28th Bled eConference June 7-10, 2015 Bled, Slovenia #eWellBeing Conference Proceedings Editors: Roger Bons, Johan Versendaal, Andreja Pucihar, Mirjana Kljajić Borštnar bledconference.org CIP - Kataložni zapis o publikaciji Narodna in univerzitetna knjižnica, Ljubljana 304.3(082)(0.034.2) 004:316(082)(0.034.2) BLED eConference (28 ; 2015) #eWellBeing [Elektronski vir] : conference proceedings / 28th Bled eConference, June 7- 10, 2015, Bled, Slovenia ; editors Roger Bons ... [et al.] ; [organized by University of Maribor, Faculty of Organizational Sciences, eCenter]. - El. knjiga. - Kranj : Moderna organizacija, 2015 ISBN 978-961-232-282-3 (pdf) 1. Gl. stv. nasl. 2. Bons, Roger 3. Fakulteta za organzacijske vede (Kranj) 279771136 The Proceedings Research Volume includes original research papers which have been selected from the submissions after a formal double-blind reviewing process and have been revised based on the recommendations of the referees. Copyright © Bled eConference Copyright for all papers resides with the author(s). By submitting the final paper to the Bled eConference, the author(s) agree to allow Bled eConference to have non-exclusive use of the paper for publication in the Bled eConference Proceedings, both in CD-ROM and web, at the time of the conference and in possible subsequent paper and electronic publications. If the paper is republished by the author(s) in another publication, acknowledgements of its original publication in the Bled eConference Proceedings should be noted. Individual authors retain copyright on papers. Please contact the author directly for reprint permission. Proceedings of the Bled eConference Permission to print and distribute content from this product must be through the approval of the Bled eConference. Duplication or replication of this Bled eConference proceedings is absolutely prohibited without written permission from the Bled eConference. University of Maribor Faculty of Organizational Sciences Kidričeva cesta 55a, 4000 Kranj, Slovenia Phone: + 386-4-237-4218 e-mail: Andreja.Pucihar@fov.uni-mb.si Research Volume Artefacts and Examples on Applications What are the factors that influence the success of the BiSL framework for business information management? Frank Van Outvorst, Jelle Van Dam, Maurits Methorst, Sjoerd Spée, Erwin Van Steijn and Benny M.E. De Waal ………………………………………………………………………………………………….1 Supporting Financial Market Surveillance: An IT Artifact Evaluation Irina Alic ……………………………………………………………………………………..16 Linking Threat Avoidance and Security Adoption: A Theoretical Model For SMEs Sean Browne, Michael Lang and Willie Golden …………………………………………. 32 Business Model Innovation Business Modelling Agility: Turning ideas into business Jukka Heikkilä, Harry Bouwman and Marikka Heikkilä …………………………………. 44 Digital Business Engineering: Methodological Foundations and First Experiences from the Field Boris Otto, Rieke Bärenfänger and Sebastian Steinbuß ………………………………….. 58 Two birds with one stone. An economically viable solution for linked open data platforms Riccardo Bonazzi and Zhan Liu ……….………………………………………………….. 77 More than a gut feeling: Ensuring your inter-organizational business model works Cristina Chuva Costa and Paulo Rupino Da Cunha ……………………………………….. 86 The Bitcoin Phenomenon Analysis Boris Tomaš and Ivan Švogor ………………………………………………………..….. 100 Towards an Artefact-Oriented Requirements Engineering Model for Developing Successful Products, Services, and Systems: Identification of Model Requirements Christian Ruf ……………………………………………………………………………… 113 Decision Support & Rules Management Financial Market Surveillance Decision Support: Demands on Explanatory Design Theory Irina Alic ………………………………………………………………………………… 127 Cloud Oriented Business Process Outsourcing using Business Rule Management Jeroen Van Grondelle, Rudolf Liefers and Johan Versendaal ……..…………………… 140 A Classification of Change Categories for Business Rules Martijn Zoet and Koen Smit …………………………………………………………… 155 eHealth The Effect of e-Mental Health Services on Saudi General Mental Health Bader Binhadyan, Konrad Peszynski, Nilmini Wickramasinghe ………………………… 169 The role of IT governance in generating business value from IT investments in healthcare: Lessons from an Australian experience Peter Haddad, Nilmini Wickramasinghe ………………………………………………… 180 Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health Luuk P.A. Simons, David van Bodegom, Adrie Dumaij, Catholijn Jonker ……………… 191 eHealth & IoT Standardisation of risk screening processes in healthcare through business rules management Joris Mens, Sander Luiten, Yannick Driel, Kobus Smit and Pascal Ravesteyn ………….. 205 Application of Lifetime Electronic Health Records: Are we ready yet? Kai Gand, Peggy Richter and Werner Esswein ………………………………………….. 217 Buying-off privacy concerns for mobility services in the Internet-of-things era: A discrete choice experiment on the case of mobile insurance Sebastian Derikx, Mark De Reuver, Maarten Kroesen and Harry Bouwman ………….. 228 eWellness Digital Wellness for Young Elderly: Research Methodology and Technology Adaptation Christer Carlsson and Pirkko Walden …………………………………………………. 239 Aggregating Community Resources of Care and Assistance Services for Elderly Population Maria Manuela Cruz Cunha, Ricardo Simoes and Isabel Miranda …………………… 251 Fuzzy optimization to improve mobile wellness applications for young-elderly Jozsef Mezei and Shahrokh Nikou ……………………………………………………… 263 Impact on people and organizational performance Human oriented performance management: Is there a gap between executives and non- executives? Benny M.E. De Waal, Pieter T. Hofste, Jeffrey Benthem, Jean-Luc J.N.T. Bonnier, Rob Ter Hedde……………………………………………………………………………….…… 276 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs Michael H. Quade and Uwe Leimstoll …………………………………………………. 290 Women and ICT: exploring obstacles and enablers of a possible career Ruxanda Berghi and Paola Biell………………………………………………………… 306 Innovations in online and mobile channels Understanding Online Channel Expansion in a SME Context: A Business Model Perspective John Jeansson, Shahrokh Nikou, Rune Gustavsson, Siw Lundqvist, Leif Marcusson, Anna Sell and Pirkko Walden …………………………………………………………………. 322 Integration of machine learning insights into organizational learning - a case of B2B sales forecasting Marko Bohanec, Mirjana Kljajić Borštnar and Marko Robnik-Šikonja ………………… 338 Ontology based Multi Agent System for the Handicraft Domain E-Bartering Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled and Khaled Ghedira ……………. 353 Social media and big data Social CRM performance dimensions: A resource-based view and dynamic capabilities perspective Nicolas S. Wittkuhn, Tobias Lehmkuhl, Torben Küpper and Reinhard Jung …..………. 368 Crowdsourcing in Software Development: A State-of-the-Art Analysis Niklas Leicht, David Durward, Ivo Blohm, Jan Marco Leimeister ……………………. 389 Social CRM Performance Model Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke and Reinhard Jung……………………………………………………………………………………… 418 The Effects of Brand Engagement in Social Media on Share of Wallet Heikki Karjaluoto, Juha Munnukka and Severi Tiensuu ...……………………………… 436 Quasi-Empirical Scenario Analysis and Its Application to Big Data Quality Roger Clarke …………………………………………………………………………….. 449 A Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance Heikki Karjaluoto and Hanna Mäkinen ………………………………………………… 469 Online Dating Sites: A tool for romance scams or a lucrative e-business model? Mohini Singh and Margaret Jackson …………………………………………………… 482 Assessment Schema for Social CRM Tools: An Empirical Investigation Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl and Reinhard Jung……………………………………………………………………………………….489 Understanding the e-Customer Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers Lauri Frank, Markus Salo and Anssi Toivakka …………………………………………. 503 Designing Tablet Banking Apps for High-Net-Worth Individuals: Specifying Customer Requirements with Prototyping Christian Ruf, Andrea Back and Henk Weidenfeld .…………………………………… 520 Business Volume Panels How to publish your conference paper in a journal? Hans-Dieter Zimmermann, Co-Editor in Chief Electronic Markets - The International Journal on Networked Business, Narciso Cerpa, Editor in Chief Journal of Theoretical and Applied Electronic Commerce Research Nilmini Wickramasinghe, Editor in Chief International Journal of Biomedical Engineering and Technology Ronald S. Batenburg, Editorial Board Member International Journal of Organisation Design and Engineering Jože Zupančič, Editor in Chief Organizacija – Journal of Management, Informatics and Human Resources Digital Wellness Services for Young Elderly: New Frontiers for Mobile Technology Harry Bowman, Professor TU Delft, The Netherlands & IAMSR/Abo Akademi University, Finland Doug Vogel, Professor Harbin Institute of Technology, China Pirkko Walden, Professor IAMSR/Abo Akademi University, Finland e-Solutions to Cluster Analysis and Knowledge Sharing Marc Pattinson, Manager Inno Group, Sophia Antipolis, France Dr John Hobbs, Senior Lecturer, Department of Management & Enterprise Cork Institute of Technology, Ireland Tamara Högler, Head of Innovations and International Affairs CyberForum e.V., Germany Darja Radić, Project Manager Automotive Cluster of Slovenia eHealth – Towards independent living Luuk P.A. Simons, Professsor Delft University of Technology, Netherlands Vladislav Rajković, Professor Emeritus University of Maribor, Slovenia Mateja Sajovic, Head of IT infrastructure, University Medical Center Ljubljana Vesna Prijatelj, Business director of Independent hospitals of University Medical Center Ljubljana Igor Košir, Strategic programs director, Smartcom, Slovenia eHealth – Security, privacy and reliability Urban Schrott, IT Security and Cybercrime Analyst, Communications manager at Reflex, ESET Ireland, Safetica UK & Ireland Vladislav Rajković, Professor Emeritus University of Maribor, Slovenia Igor Košir, Strategic progams director, Smartcom, Slovenia Business model innovation Harry Bowman, Professor TU Delft, The Netherlands & IAMSR/Abo Akademi University, Finland Jukka Heikkila, Professor, University of Turku, Finland Timber Haaker, Innovalore, The Netherlands Representative from Evolaris, Autria Cloud Computing: Towards Business e-Model Enlightenment Thomas Acton, Head of School of Business & Economics National University of Ireland Galway, Ireland Lorraine Morgan, Senior Researcher Lero - The Irish Software Research Centre, National University of Ireland Galway, Ireland Barry Reddan, Senior Manager, Hewlett Packard Ultan Sharkey, CEO Sharkey Consulting, Ireland Social Media for business: Current state and future trends Gregor Zupan, Information Society, Statistical Office of the Republic of Slovenia Marko Perme, Director, Agilcon, Slovenia Urban Schrott, Ireland - IT Security and Cybercrime Analyst, Communications Manager at Reflex / ESET Ireland / Safetica, United Kingdom and Ireland Workshops Future University Johan Versendaal, Professor of Extended Enterprise Studies, Research Centre Technology & Innovation, HU University of Applied Sciences Utrecht & Professor of E-Business, Faculty of Management, Science and Technology, Open University of the Netherlands Roger Bons, Director, Bons Academic Services, The Netherlands & 2015 Research Track Chair Tomi Ilijaš, Director, Arctur, Slovenia Amira Mujanović, Blaž Sašek, Students, University of Maribor, Slovenia Rok Kepa, Blaž Vidmar, Students, University of Ljubljana, Slovenia Information Literacy of Students: Preliminary Results of a Survey on a Slovenian Sample Alenka Baggia, Mirjana Kljajić Borštnar, Andreja Pucihar, Danica Dolničar, Tomaž Bartol, Andrej Šorgo, Bojana Boh in cooperation with Saša Aleksej Glažar, Vesna Ferk Savec, Mojca Juriševič, Blaž Rodič, Irena Sajovic, Margareta Vrtačnik Professor Rene W. Wagenaar ePrototype Bazaar WoShi: Mobile worker scheduling Simon Cvetek, Darja Štagar, Faculty of Organizational Sciences, University of Maribor, Slovenia Happly: Connecting elderly and volunteers Lior van Dam, Michelle de Groot, Maaike van Heek, Arne van Tilburg, Vincent de Wit, HU University of Applied Sciences Utrecht, The Netherlands Kornuit: Connecting autistic people Cindy Jong, Maurits van de Lagemaat, HU University of Applied Sciences Utrecht, The Netherlands GoCoach: Supporting sport coaches and trainees Mitja Rozman, Andrej Zupan, Matevž Oman, Faculty of Organizational Sciences, University of Maribor, Slovenia GlobalCulture: Culture at your fingertips Layal Storms, Vera Koopmans, Merel Wintjens, Nahom Nahom Isaac Michael, HU University of Applied Sciences Utrecht, The Netherlands GraduateStudents Consortium Design and Application of a Methodology for Unstructured Data Automated Analysis in Word-of-Mouth Lucie Šperková, University of Economics in Prague, the Czech Republic Analytics on Feedback Creation: The Application of Learning Analytics on Formative Assessments Justian Knobbout, Utrecht University of Applied Sciences, the Netherlands A conceptual framework for service system interaction Buddhi Pathak, Henley Business School, University of Reading, United Kingdom Social CRM adoption and its influence on organizational performance - SMEs perspective Marjeta Marolt, Faculty of Organizational Sciences, University of Maribor, Slovenia BACK 28th Bled eConference June 7 - 10, 2015; Bled, Slovenia What are the factors that influence the success of the BiSL framework for business information management? Frank van Outvorst The Lifecycle Company / ASL/BiSL Foundation, the Netherlands frankvanoutvorst@hotmail.com Jelle van Dam HU University of Applied Sciences Utrecht, the Netherlands jelle.vandam@student.hu.nl Maurits Methorst HU University of Applied Sciences Utrecht, the Netherlands maurits.methorst@student.hu.nl Sjoerd Spée HU University of Applied Sciences Utrecht, the Netherlands sjoerd.spee@student.hu.nl Erwin van Steijn HU University of Applied Sciences Utrecht, the Netherlands erwin.vansteijn@student.hu.nl Benny M.E. de Waal HU University of Applied Sciences Utrecht, the Netherlands benny.dewaal@hu.nl Abstract Business Information Services Library (BiSL) is a framework from Dutch origin that helps organizations shape Business Information Management. BiSL is not used by every organization in the Netherlands. The question is what moves organizations to start using BiSL or what motives do they have to reject the use of BiSL. The research question in this study is: What motivates the adoption (or non-adoption) of the BiSL Framework? To answer this question 18 interviews were conducted. The interviews have been held with organizations that do use BiSL and with organizations that do not use BiSL. Among the interviews were three interviews with experts in the field of BiSL. The conclusion of our research is that organizational readiness is the deciding factor to 1 Van Outvorst et al. use BiSL. To apply BiSL successfully there is a need of support, knowledge and a certain level of organizational maturity in business information management. Keywords: Business Information Management, BiSL framework, qualitative research, critical success factors, the Netherlands. 1 Introduction In the 1990s work on the development of a framework for business information management was initiated. Several reasons for development of such a framework existed (Outvorst and Scholten, 2013):  Organizations felt the need for structural governance and management of information and information technology (IT) due to the increasing importance and the rising cost of IT;  Organizations that outsourced their IT felt the need to professionalize their remaining IT organization;  Organizations were confronted with expensive investments in IT that in the end did not meet their expectations. On the basis of experience and lessons from practice a framework for business information management, better known as the Business information Services Library (BiSL) was developed (Pols, Donatz and Outvorst, 2012). In 2005 the BiSL book was published and the BiSL framework was adopted by the ASL BiSL Foundation in which several large profit and not-for-profit organizations participate. Since the publication of the book, BiSL has become the Dutch industry standard for business information management. Although this framework offers a lot of practical guidance to establish business information management in organizations, it seems that not everyone is adopting it and that there is some uncertainty about the contribution or value of the framework and the factors that may play a role in its successful use. Possible factors may be:  BiSL is only successful when it fills certain gaps in the alignment between business and IT;  Organizations are used to working with models, frameworks or standards, like Cobit5 (Isaca, 2012), ITIL (Bon, 2011) or TOGAF (The Open Group, 2009);  Organizations recognize and understand the importance of IT;  Organizations are only interested in international frameworks;  Thinking about information and IT is mainly influenced by technology;  Organizations are not familiar with the concept of demand and supply of information and information technology. The main research question is therefore: What motivates the adoption (or non-adoption) of the BiSL Framework? In order to answer this main question the sub-questions are: 1. Why do organizations adopt BiSL? 2. Why do organizations not adopt BiSL? 2 What are the factors that influence the success of the BiSL framework? 3. When is BiSL regarded as being successful? 4. What are the critical success factors for application of BiSL? The answers to the questions mentioned are to be used in further development of the BiSL framework. After this introductory section the theoretical background of business information management is discussed. In section 3 the research method is described. The research findings are presented in sections 4, and the paper is finalized with the conclusions in section 5. 2 Theoretical background 2.1 Organization of business information management The importance of information systems and information technology (IS/IT) for organisations is unquestionable. IS/IT is increasingly penetrating into the core of organisational performance and IT/IS usage is still growing (Azadeh, Keramati and Songhori, 2009). In general, the management of IS/IT is considered pivotal in ensuring successful use of information assets (Evans and Price, 2012). The scope of IS/IT management deals with a wide range of activities, starting with system initiation, through to design, realisation, system implementation and finally post implementation or system assimilation. Important aspects of the way business information management is organized are the relationship between (senior) management and the user community on one hand and the management of change of IT on the other hand, with attention given to matters such as organisational alignment, the contribution of an information system to the performance of the organization, and other issues which have an impact on the working practices of individual employees (Booth and Philips, 2005; Liang et al., 2007; Orlikowski, 1992; Silvius, De Waal and Smit, 2009). The way business information management is organized can be explained by the following two indicators (Pols, 2009): a. Business or IT: Is business information management part of the business (demand side) or part of the IT department (supply side)? According to the model of Looijen and Delen (Looijen, 1998) there are three different domains: Infrastructure management, Application management, and Business information management. These domains are illustrated in Figure 1. In this model Business Information Management is positioned within the business and not within the IT department. b. Centralized or decentralized: Is business information management in an organisation centralized or decentralized? Centralized means that business information management is executed at one place and by only one single department. In this situation business information management takes care of multiple business units (Pols, 2009). Decentralized means that business information management is executed by multiple units within one organization. This usually means that there are several business information management departments throughout the organization which are responsible for the different divisions or staff departments (Pols, 2009). The choice to centralize or decentralize their business information management department is usually the 3 Van Outvorst et al. result of developments that happened in the past and not necessarily a choice that was based upon a clear vision (Goense-Van den Bosch and Donatz, 2008). Application Management Business Information Management Infrastructure Management Business (demand) IT (supply) Figure 1: Model of Looijen and Delen In the next section we shortly introduce the main concepts and processes of the BiSL framework. 2.2 BiSL Framework The BiSL framework (Figure 2) divides business information management into three layers: The operational layer, the strategic layer and the managing layer (Pols, Donatz and Outvorst, 2012).  Operational layer is concerned with the use and definition of the demand of the information or information systems;  Managing layer is concerned with the profits, costs, contracts, and planning.  Strategic layer is concerned with the long term plans for the information systems. The BiSL framework is divided into seven clusters of processes. These clusters are shown in Figure 2. Process cluster 1: Use Management This cluster is build up with processes intended to create an optimal and continuous support for the user of the information system. It is about supporting the user in carrying out their tasks and managing the IT supplier (Pols, Donatz and Outvorst, 2012). Process cluster 2: Functionality Management The process cluster functionality management describes the processes where changes are specified and carried out. The goal of this process cluster is to implement a change within the constraints defined while meeting the needs, goals and demands (Pols, Donatz and Outvorst, 2012). 4 What are the factors that influence the success of the BiSL framework? Figure 2: BiSL Framework (ASL BiSL Foundation, 2014) Process cluster 3: Connecting Processes Operational Level In the third cluster the goal is to decide which changes will be realized. The other goal is to actually implement the change in the end user support (Pols, Donatz and Outvorst, 2012). Process cluster 4: Management Processes The management processes initiate all BIM processes. The management processes monitor the activities in terms of costs, benefits, needs, contracts, service levels and planning (Pols, Donatz and Outvorst, 2012). Process cluster 5: Information Strategy This cluster is to make sure the information technology is still relevant in the future when new demands are set. This cluster also solves the structural flaws in the current situation. Process cluster 5 is about formulating a long term information systems strategy. (Pols, Donatz and Outvorst, 2012). Process cluster 6: I-Organization Strategy I-Organization strategy is about defining roles, responsibilities and forms of cooperation. Relevant parties need to agree about working structures, management and processes. (Pols, Donatz and Outvorst, 2012). Process cluster 7: Connecting Processes Strategic Level Process cluster 7 connects the cluster of long term plans and policies for content and organization form. This connection requires a process by which all information plans by all different actors are aligned. (Pols, Donatz and Outvorst, 2012). 5 Van Outvorst et al. Just like other frameworks like ASL (Pols and Backer, 2006), ITIL (Bon, 2011) or CMM (Clerc and Niessink, 2004) the BiSL framework is a simplified reproduction of reality. BiSL can be used as a checklist or a tool for standardization. Apart from that best practices with experiences of different types of organizations are available (Bakker, 2014). BiSL requires a situational approach. That does not mean that BiSL can just be implemented by any organization when they feel like it. A vision about which parts from BiSL will be useful needs to be developed by the organization (Pols, Donatz and Outvorst, 2012). 2.3 Relationships with other frameworks and models This section addresses the relationships between the BiSL framework and other relevant frameworks or models. Business information management is heavily influenced by the strategic alignment model of Henderson and Venkatraman (1993). Figure 3: Strategic alignment model of Henderson and Venkatraman (from Van Bon, 2010) Henderson and Venkatraman (1993) recognize the dynamics and challenges in realizing a strategic fit between strategy and infrastructure and processes as well as a functional integration between business and IT. In practice the distance between business and IT is fairly large, due to different responsibilities, different universes in which they operate and different languages they speak. In order to create connection between business and IT, business information management comes in place. A convenient way of looking at this connection is the Amsterdam Information management Model (AIM) or the Nine Square framework (Maes, 1999). This model is an extension of the model of Henderson and Venkatraman and basically offers connecting layers between internal and external focus and between business and IT. By doing this AIM can be used to identify a business information management column that is concerned with aligning use of IT and demand for IT on one hand with supply and implementation of IT services on the other hand. Also a tactical level can be identified which is concerned with aligning long term goals with short term actions and the present situation. 6 What are the factors that influence the success of the BiSL framework? Figure 4: The Amsterdam Information management Model (from Van Bon, 2010) Besides giving explicit attention to the various alignment areas within the overall business IT alignment, the model of Maes proves to be very convenient in comparing and positioning the different relevant models and frameworks that are used in the total area which stretches from corporate governance of IT to dealing with interruptions in IT services (see Figure 5). Figure 5: The positions of frameworks in AIM (from Van Bon, 2006) 7 Van Outvorst et al. 3 Research methods 3.1 Interviews To answer the research question interviews were conducted in order to discover the experiences of different organizations with BiSL. In total 32 organizations were asked to participate in the interviews; 30 of these gave a positive reaction and with 17 an interview could be organized. In selecting the organizations for the interviews a large variation in the organizations was sought:  Organizations that use BiSL, as well as organizations that do not, although on forehand it was not always clear whether organizations do or do not use BiSL;  Organizations in as many industries as possible;  Organizations that are highly dependent on information and IT and organizations that are less dependent (based on assumptions about the organizations on forehand);  Large and smaller organizations. In total 15 information managers of 14 organizations and 3 BiSL experts were interviewed in 18 interview sessions. The organizations were active in the field of semi government (3), public services (4), education (2), healthcare (2), financial service (2) business service (3) and engineering (1). The interviews were focused on getting answers to topics related to the motives to use the BiSL framework. Therefore, mostly open questions were asked to investigate the relationship between the experience with BiSL and success factors. In general, the outline of the interview was followed, but depending on the answers given by the participants, deviation occurred. The interviews took place in November and December 2014. The interviews were tape-recorded and a report was made from each (Patton, 2002). This report was sent to the interviewee for approval. Comments and corrections were incorporated into the interview report. 3.2 Analysis procedure The interviews were analysed using a cumulative editing approach (Runeson and Höst, 2009). Each interview report was read carefully by the researchers in order to determine the meaningful fragments of text. These fragments were coded using open coding. Fragments of text from within one interview and between interviews were compared in order to determine whether or not they had the same code. If necessary, it was decided to merge codes or to change a fragment to another code following an axial coding procedure. The last step was to structure the codes at the level of main- and sub- variables/dimensions using selective coding. Thereafter the interviews were compared which resulted in a structured identification of fragments relating to the different elements that are presented in the next section (Miles and Huberman, 1994; Neuman, 2002; Boeije, 2002). 4 Findings 4.1 General setting Of the 14 organizations that were included in this study, 9 (64%) indicated that they had positioned business information management at the business side, while 5 (36%) 8 What are the factors that influence the success of the BiSL framework? organizations had positioned business information management at the IT side. Eight (57%) of the organizations have their business information management department decentralized. All other organizations have their business information management department centralized. Six (43%) of the organizations used BiSL as a reference model, three (22%) have partially implemented the framework (e.g. some processes are implemented) and another three (22%) have fully implemented the BiSL framework (e.g. all processes are implemented). Two (13%) of the organizations didn’t use BiSL at all. 4.2 Motives to adopt BiSL This section describes the motives to apply the BiSL framework. The two main topics in this section are the incentives to use the framework and the benefits that are achieved by doing so. When analysing the results, it was taken into account the way business information management was organized and the extent to which BiSL was adopted. 4.2.1 Motives to use BiSL Two main motives for using BiSL were mentioned: 1. The first motive is professionalization. Organizations want to improve and monitor their processes of business information management. This can be achieved by creating more structure within business information management. BiSL offers a framework to support this. The respondents defined structure as twofold:  Uniform processes where everyone knows his or her role, responsibilities and working methods;  People having the same view on what business information management is about. 2. Another motive that emerged in the interviews was that BiSL supports the Business-IT Alignment. Business IT Alignment means being able to translate the business needs to operational IT services. BiSL offers a structure of processes that specifically aim at this alignment. When it comes to actually selecting a framework, in all situations BiSL was an obvious choice because it is regarded as the industry standard. Every organization that took part in the study was familiar with the BiSL framework. Even the organizations that don't use BiSL agree with the statement that BiSL is the standard. In fact, BiSL is the only available model in this field. Only one respondent positioned an approach as an alternative to BiSL: Functional Service Management (FSM) (Bon and Hoving, 2013). FSM is based upon BiSL and uses a lot of elements from the BiSL framework. FSM prescribes how to use these elements like a recipe, which is contradictory to the situational approach of BiSL. FSM is not used by any of the organizations. 4.2.2 General benefits from using BiSL In general BiSL satisfies the initial needs that were mentioned as motivation for using the framework. Almost every organization mentions the support to "structure" the working process of business information management as a benefit of BiSL. Setting up a proper way of working within the business information management department is seen 9 Van Outvorst et al. as the biggest benefit from applying the BiSL framework. The majority of organizations focuses on the operational level of the framework. Another benefit that comes directly from using BiSL is the expected improvement of cooperation and relationships between the organization and their IT services provider. A lot of IT services providers also apply, or have knowledge of, the BiSL framework. This enables better communication and understanding. A third benefit that comes directly from the previously discussed benefits is the probability that using a framework like BiSL eliminates certain risks. Standardizing the way of working within business information management eliminates the risk of mistakes that could be made when employees are forced to improvise. The other way of eliminating risk is that of making clear good agreements with suppliers while both applying the BiSL framework prevents problems in later stages of the collaboration because both parties have a clear picture of their roles and responsibilities. A few other benefits that were pointed out in the interviews:  Using BiSL and being a member of the ASL/BiSL foundation also gives you the opportunity to hear about the way other organizations and branches use the framework to their advantage. The foundation also publishes a lot of templates, white papers and best practices about the framework to share knowledge.  The business information management department starts to be more critical of the work they do and the way they do it.  Organizations start to think about their long term plans for their information and information systems. 4.3 Motives not to adopt BiSL 4.3.1 The two organizations that do not use BiSL at all In our research only two organizations indicated that they do not use BiSL at all. The specific reasons mentioned for these two rejections are:  No need for a specific business information management framework because business information management is integrated in an overall IT department that uses a process model of its own;  No need for any process framework at all, because these frameworks do not support the variety of the business information management processes in the organization. All in all there are only two organizations that did not adopt BiSL. The way business information management was organized for these organizations did not significantly differ from the way it was organized in the other organizations that were studied. Because of this it is not possible to come to overall findings explaining why organizations do not adopt BiSL. However, in the interviews a number of disadvantages and possible improvements came forward. 10 What are the factors that influence the success of the BiSL framework? 4.3.2 Disadvantages and possible improvements BiSL Weak spots of the framework that were mentioned regularly in the interviews:  BiSL needs more references to other frameworks. The question that needs to be answered here is: What is the relationship between BiSL and other frameworks? ITIL was the most frequently mentioned framework here. Other frameworks that came forward in the interviews are: ASL, Prince 2 and Agile.  Some process clusters of the framework are not clear. The most confusing cluster is the cluster information coordination according to the respondents. The goal of this cluster is clear but the activities within the cluster are not. The most confusing layer is the strategic layer.  BiSL is written in a scientific way and the BiSL book does not contain prescriptions how to apply the framework in practice. Because of this, it is difficult to translate the framework into daily business practice. Other disadvantages/possible improvements that were mentioned only occasionally are:  The operational layer of BiSL offers too little control over the IT-supplier.  BiSL isn’t well known worldwide, this makes it difficult to interact for organizations who do business worldwide.  When using BiSL, you start to use a different terminology. Communicating with people who are not familiar with this terminology can become increasingly difficult.  Implementing a framework like BiSL costs a lot of money and time, and creates bureaucracy.  It is sometimes difficult to convince management to implement a model. By making the framework visually more attractive it might be easier to convince management. 4.4 Success of BiSL and critical factors It is difficult to determine whether the application of BiSL is a success. Organizations that were interviewed usually say there is no intention to make BiSL a success, “it's just a tool to reach a goal”. That goal being to make business information management a success. Apparently there is no need within organizations to assess or monitor the use of the tool BiSL. Although the success of business information management is not the scope of this research we diverge to this issues at this point in order to contextualise our findings. It illustrates that BiSL is considered as a de facto tool for which quality and success are not disputed. In 2010 a whitepaper was published on which indicators can be identified to measure the use and success of BiSL. This whitepaper suggests which criteria and measurable points of interest can be used (Faassen et al., 2010). Surprisingly none of the respondents in the interviews or the survey pointed out that they used this whitepaper or that they know about it existence. When the focus is shifted from the success of BiSL to the success of business information management the following criteria are the ones that are usually measured, according to the respondents: 11 Van Outvorst et al.  Customer satisfaction  Lead time of calls  Employee happiness  Cost reduction Customer satisfaction is the number one criterion for business information management to be successful. This is also the one that almost every organization actually measures. The lead time on calls was pointed out to a lesser degree. Employee satisfaction and cost reduction were only mentioned a few times. One way to test how processes and activities of business information management are designed is by using the BiSL self-evaluation. This evaluation shows both what the processes look like now, as well as the flaws of the organization with the possible consequences (Donatz, 2014). Not all organizations that were interviewed measure the performance of the business information management department. The reasons for this is that organizations don’t really "think about it" or the department is still in development and not ready to be measured. Initial assessments are not executed. Criteria will still have to be determined for those organizations. In the interviews the respondents were asked which critical factors influence whether or not BiSL can be made into a success for the organization. The following statements were given:  The organization needs to be conscious about working with BiSL and the value it adds.  There needs to be support from management. To make sure the departments in question can work with BiSL in a successful way, not just those departments but also management need to be convinced about the value that the framework will add. If management starts sceptical and remains sceptical chances of successful implementation are small.  It is important to understand that BiSL is a tool, not a goal in itself.  Education about the BiSL framework is important.  Create and maintain contact with the user. Customer satisfaction may be an indicator for this. Staying on top of technological developments.  The business information manager needs to have the right skill set: communicative skills, sense for organizing, and empathy. 5 Discussion and conclusion In this paper we look for an answer to the question what factors influence the success of the BiSL framework for business information management. In order to find this answer four sub-questions are defined: 1. Why do organizations adopt BiSL? 2. Why do organizations not adopt BiSL? 3. When is BiSL regarded as being successful? 4. What are the critical success factors for application of BiSL? 12 What are the factors that influence the success of the BiSL framework? All organizations included in this research know of BiSL and agree in BiSL being the (Dutch) industry standard for business information management. There is no real alternative for BiSL. Motives to make use of BiSL are a drive to professionalize business information management and a need to improve the business IT alignment. Organizations adopting BiSL are not disappointed by the results. Using BiSL results in a better structured information management and a better performance in business IT alignment. Other benefits are a better collaboration between business and IT department or an external IT services provider and an effective way to manage risks. Having pointed out the benefits of the use of BiSL, the research also shows not every organization has adopted BiSL. But there are only very few organizations that do not use BiSL and the reasons for not adopting the BiSL framework cannot be made valid in general. The reasons mentioned seem to be dependent on the need for the organization for an integrated overall IT process model on one hand and the aversion to uniform processes within the organization on the other. It is very striking that organizations have certain intentions when adopting BiSL and indicate they achieve their goals, but are not interested in explicitly evaluating the use of BiSL as a tool. However, several points that can be improved in the BiSL framework were brought forward in the interviews. These point will undoubtedly be of value for further development of the BiSL framework and are related to relationships with other frameworks, more comprehensible and accessible descriptions and use of the framework in the daily practice. Finally the findings reveal that the influencing factors for success of BiSL are very much alike influencing factors in any other organizational change:  The value BiSL adds needs to be clear to everyone and management needs to support the adoption of the BiSL framework;  It is important to keep in mind that BiSL is just a means, not a goal in itself;  People need to be educated about BiSL;  Contact with the user is important to validate if business information management is still on the right track;  The business information management people need to have the right skill set: communicative skills and organizational sensitivity and organizing abilities. Acknowledgement The authors wish to acknowledge the different organizations for making it possible to investigate the adoption of BiSL in practice. Without their corporation it would not have been possible to collect the data for this research. In that respect, many thanks to the respondents who were willing to be interviewed. References ASL BiSL Foundation. (2014). Model. Retrieved at 1-16, 2015, from ASL BiSL Foundation: http://aslbislfoundation.nl/ 13 Van Outvorst et al. Azadeh, A., Keramati, A. and Songhori, M.J. (2009). An integrated Delphi/VAHP/DEA framework for evaluation of information technology/information systems (IT/IS) investments. International Journal of Advanced Manufacturing Technology, Vol. 45(11/12), 1233-1251. Boeije, H. (2002) A purposeful approach to the constant comparative method in the analysis of qualitative interviews. Quality & Quantity, Vol. 36(4), 391-409. Bon, J. van (2006). Frameworks for IT management. Zaltbommel: Van Haren Publishing. Bon, J. van (2010). Strategic Alignment modellen: Het multifunctionele negenvlaksmodel. Best Practice – Quartely review, Vol. 1(1), 13-18. Bon, J. van. (2011). ITIL 2011 Edition - A pocket guide. Zaltbommel: Van Haren Publishing. Bon, J. van and Hoving, W. (2013). De FSM-methode. Inform-IT. Booth, M.E. and Philips, G. (2005). Information systems management: Role of planning, alignment and leadership. Behaviour & Information Technology, Vol. 24(5), 391–404. Clerc, V. and Niessink, F. (2004). IT service CMM. Zaltbommel: Van Haren Publishing. Donatz, R. (2014). BiSL self-assessment. Zaltbommel: Van Haren Publishing. Evans, N. and Price, J. (2014). Why organizations cannot justify the effective management of their information assets. In: V. Grozdanic (Ed.) Proceedings of the 10th European Conference on Management Leadership and Governance ECMLG 2014, Zagreb, Republic of Croatia, 13-14 November 2014 (111-117). Reading: Academic Publishing International Limited. Faassen, F., Smulders, A., Boers, R., Vries, J. de, Hattum-Buitenkamp, A. van and Hokke, H. (2010). BiSL meetbaar gemaakt. Utrecht: ASL BiSL Foundation. White paper, November 25, 2014. Goense-Van den Bosch, T. and Donatz, R. (2008). Centraal of decentraal. IT Service Magazine, Juni 2008. Henderson, J.C. and Venkatraman, N. (1993). Strategic alignment: Levaring information technology for transforming organizations. IBM Systems Journal, Vol. 32(1), 4-16. ISACA. (2012). Cobit 5. Zaltbommel: Van Haren Publishing. Liang, H., Saraf, N. Hu, Q. and Xue, Y. (2007). Assimilation of enterprise systems: The effect of institutional pressures and the mediating role of top management. MIS Quarterly, Vol. 31(1), 59-87. Looijen, M. (1998). Information systems: Management, control and maintenance. Deventer: Kluwer. Maes, R. (1999). A generic framework for information management. Primavera workingpaper series 1999-03. 14 What are the factors that influence the success of the BiSL framework? Miles, M.B. and Huberman, A. (1994). Qualitative data analysis: An expanded sourcebook. Thousand Oaks: Sage Publications. Neuman, W.L. (2002). Social research methods: Qualitative and quantitative approaches. Boston: Pearson Education. Outvorst, F. van and Scholten, L. (2013). Industrial experience report: BiSL as driver for innovating business information management in the Dutch police organization(s). In: 13th International Conference, SPICE 2013, Bremen, Germany, June 4-6, 2013. Proceedings. Heidelberg: Springer Verlag Patton, M.Q. (2002). Qualitative research and evaluation methods. Thousand Oaks: Sage Publications. Pols, R. van (2009). Business informatiemanagement en BiSL in de praktijk. Zaltbommel: Van Haren Publishing. Pols, R. van and Backer, Y. (2006). ASL, Application Service Library - A management guide. Zaltbommel: Van Haren Publishing. Pols, R. van, Donatz, R. and Outvorst, F. van (2012). BiSL® - A framework for business information management. Zaltbommel: Van Haren Publishing. Orlikowski, W.J. (1992). The duality of technology: rethinking the concept of technology in organizations. Organization Science, Vol. 34(3), 398-427. Runeson, P. and Höst, M. (2009). Guidelines for conducting and reporting case study research in software engineering. Empirical Software Engineering, Vol. 14(2), 131-164. Silvius, A.J.G., De Waal, B.M.E. and Smit, J. (2009). Business and IT alignment: Answers and remaining questions. In: Proceedings of the 13th Pacific Asia Conference on Information Systems (PACIS), 10-12 July, Hyderabad, India. The Open Group. (2009). The Open Group Architecture Framework TOGAF Version 9. Zaltbommel: Van Haren Publishing. 15 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Supporting Financial Market Surveillance: An IT Artifact Evaluation Irina Alić University of Göttingen, Germany irina.alic@wiwi.uni-goet ingen.de Abstract In this paper, an IT artifact instantiation (i.e. software prototype) to support decision making in the field of financial market surveillance, is presented and evaluated. This artifact utilizes a qualitative multi-attribute model to identify situations in which prices of single stocks are affected by fraudsters who aggressively advertise the stock. A quantitative evaluation of the instantiated IT artifact, based on voluminous and heterogeneous data including data from social media, is provided. The empirical results indicate that the developed IT artifact instantiation can provide support for identifying such malicious situations. Given this evidence, it can be shown that the developed solution is able to utilize massive and heterogeneous data, including user-generated content from financial blogs and news platforms, to provide practical decision support in the field of market surveillance. Keywords: Big Data, User-Generated Content, Decision Support Systems 1 Introduction Financial market manipulation has received much attention from regulatory authorities, resulting in trading suspensions of those companies that may have been hijacked by fraudsters. Recent suspensions included two large groups of penny stock companies suspended by the U.S. regulatory authority on a single day (SEC, 2012) before they could harm investors. All companies have been traded on over-the-counter (OTC) markets with low regulatory standards. For the assessment of potentially affected companies, the supervisory authority used information technology to recognize thinly 16 Irina Alić traded, low-capitalization stocks that could be affected by fraudsters. The system is calibrated to detect companies with low-priced penny stocks that are traded in low volumes or not at all; such companies must also be delinquent in their public disclosures (SEC, 2012). Unfortunately, the system reveals its weak point: namely, insufficient real- time surveillance on voluminous and heterogeneous data. The pump-and-dump manipulation scheme is the most common form of information- based market manipulation in financial markets (Dunham, 2007; Zaki, 2013). It first appears as the spreading of false positive information to market participants by fraudsters; this information, typically user-generated content (UGC) on financial blogs and news platforms, is enthusiastically spread by fraudsters attempting to pump up a stock price to an artificial level. Usually, the affected stocks are penny stocks that trade below $5. Before starting the manipulation, fraudsters buy a significant quantity of the stocks over a longer period of time. Motivated by spread positive information, uninformed market participants relying on the bogus story buy the stocks, effectively forcing abnormal price increases (pump) (Bouraoui, 2009; Huang & Cheng, 2013). Finally, fraudsters make a profit by selling their stocks at the increased price level, which accordingly causes the stock price to drop (dump). This paper addresses the question of whether the assessment of user-generated content has the potential to help regulatory authorities and financial institutions detect such situations. Therefore, the focus of this study is to present a quantitative evaluation of the final IT artifact that has been developed within a major EU project in order to demonstrate the feasibility of the approach. Accordingly, an evaluation based on a voluminous dataset is performed; for this purpose, OTC trading data and corresponding UGC related to 1700 companies were collected during 2012 and 2013. The collected data is used by the developed and instantiated IT artifact to identify situations of abusive behavior, and to assess whether those identified situations do actually demand more attention compared to those that were not identified by be suspicious. At this point, it is important to note, that the EU project involved several activities in order to provide services including technologies development as well as the development of data mining algorithms suitable to handle big data complexity and provide metadata for further usage for the different use cases. The corresponding development processes and associated findings resulted in more than 20 research articles and project deliverables (“Project FIRST,” 2013). Thus, in this present paper, the voluminous data utilized for this evaluation is provided by the mentioned services. In the next section, related background research is introduced. Next, I provide an overview of the developed IT artifact instantiation. Section 4 describes the quantitative evaluation based on real data, and the final section concludes this paper. 17 Supporting Financial Market Surveillance: An IT Artifact Evaluation 2 Background Relevant areas of study relating to this research in terms of stock fraud detection include literature on (1) stock touting impacts based on spam mails; (2) the impact of user- generated news on stock activity; and (3) financial markets services architecture. Each area will be briefly discussed in the following paragraphs. Several studies examine of the impact of stock touting using spam emails. To generate profit, scammers spread misleading information. Authors (Frieder & Zittrain, 2008) investigate market activity prior to and following a stock touting email campaign, making use of a dataset containing about 75,000 spam emails. The research reveals high market activity beginning one day before email spamming commences and continuing until the day with the most considerable number of touting mails. The authors find that volume and return respond positively to touting, whereas the returns dip significantly following the conclusion of the campaign. Hence, the study suggests two main indicators for market manipulation: abnormal price changes and trading volumes, which will utilized by our IT artifact. Research conducted by (Bouraoui, 2009) demonstrates similar findings. The author provides an evolution model of volatility to assess the impact of stock spam emails based on a sample of 110 penny stock companies. The research shows abnormal returns three days after the commencement of spam emails. In both studies, the outcomes have been explained as the behavioral effect of market participants who have responded positively to the touting. Thus, these studies consider statistical approaches to explicate the influence of email stock promotion. Other studies introduce data mining techniques that help identify stock touting spam emails. Research by (Zaki, Theodoulidis, & Solis, 2011) observes spam massages to detect highly fraudulent stock activities by utilizing data mining techniques to identify stock touting spam emails. The accuracy detection of these experiments ranges from 58% to 71%. A considerable amount of research on the predictive power of UGC (such as tweets, financial forums, and blogs) on stock prices has been documented in the literature. In one instance, the research (Delort, Arunasalam, Leung, & Milosavljevic, 2011) introduce evidence of manipulation and examine the effect of such misuse in online financial forums. The authors show that manipulative user-generated content regarding companies with lower prices and market capitalization positively correlates to stock returns, volatility, and volume. One recent study (Smailović, Grčar, Lavrač, & Žnidaršič, 2013) presents a support vector, machine-based, sentiment classifier; here, a set of about 150,000 tweets thematizing eight companies (e.g., Apple, Google, and Microsoft) served as the data basis. The authors find that positive sentiment predicts positive movement in the closing price. A study proposed by (Zhang & Skiena, 2010) examines the ways in which blog and news data is reflected in trading volumes and returns. The authors demonstrate a significant positive correlation between media content and trading volume as well as stock returns. Therefore, in this study the UGC data will be incorporated. 18 Irina Alić Several commercial stock market systems for monitoring and detecting abuse in structured and unstructured data exist; some, such as the Securities Observation, News Analysis and Regulation Systems (SONAR), are presented in scientific research (Goldberg, Kirkland, Lee, Shyr, & Thakker, 2003). SONAR, which aims to monitor the stock market, applies data mining, text mining, statistical regression, and rule-based detection to recognize both abuse patterns in structured data and unusual trading following publication of the news. A study by (Mangkorntong & Rabhi, 2008) compares two different surveillance systems as event-processing systems in such areas as memory usage, scalability, and flexibility. The authors reveal the strengths and weaknesses of the two systems and suggest a generic approach that uses numerous different event-processing systems to support the detection process (Mangkorntong & Rabhi, 2007). The next section introduces the Financial Market Surveillance Decision Support System (FMS-DSS), which was developed within an EU project between 2010 and 2013 (“Project FIRST,” 2013). As an IT artifact instantiation it demonstrates how regulatory authority can be supported in detecting malicious pump-and-dump market manipulations utilizing voluminous and heterogeneous data streams. 3 Instantiated IT Artifact During the research project, the opportunity to develop an instantiated IT artifact (FMS- DSS) was provided. Within the research consortium consisting of research institutions, industry partners, and financial regulatory authority, software components have been developed in close researcher-practitioner collaboration. The problem owner (i.e., domain experts and regulatory authority) intervened according to the project needs and aligned the design principles with their surveillance issues (Alić, Siering & Bohanec 2013). Generally, decision support system configurations are built on the basis of three basic technology components related to: (1) data, (2) models, and (3) user interface (UI) (Turban, Sharda, & Delen, 2010). The following subsections will present these components. The data management component preprocesses and stores the needed data. The model component assesses whether or not given stock are suspicious of being currently affected by pump-and-dump manipulation schemes. The user interface component allows meaningful representation of suspicious situations. During operation, FMS-DSS continuously searches for the appearance of user-generated content related to monitored companies. For this purpose, unstructured input data is continuously retrieved, preprocessed, and stored in a database. Based on rules and models that were developed in close collaboration with the domain experts that were involved in the project, the UI shows alerts, indicating assessments of 19 Supporting Financial Market Surveillance: An IT Artifact Evaluation how likely a stock is affected by manipulation (ranging from very high, high, medium, low, to very low). 3.1 Data Selection of potential input data categories based on pump-and-dump manipulation scheme evidence: In meetings with experts (i.e., compliance officers and the regulatory body), the specific decision problem in revealing typical factors for pump-and-dump market manipulation was explained. The three main types of information incorporated into the decision model are: the manipulation of information concerning the company, the manipulation of the financial instrument, and news as user-generated content about the company and its financial instrument. Consequentially, the three main input data categories of market abuse suspiciousness are Company, Financial Instrument, and News:  Company: According to the experts, there are two determiners for company suspiciousness. First, in those cases where the company is already involved in financial market manipulation, the financial authority issues litigation releases and puts the company on a blacklist, which is later refined into company, country, and industry blacklists. The second determiner of a company’s reliability is its history; the manipulator often targets newer companies and companies that have gone bankrupt and have recovered again. The History attribute is thus refined into the attributes Age and Bankruptcy.  Financial Instrument: This category is refined into the attributes Market and Trading. If a company’s financial instrument is listed in a market segment with low regulatory requirements, and the company itself has low market capitalization, then this instrument is seen as an additional indicator of suspiciousness. A change in trading volume or trading behavior is also seen as suspicious.  News: The user-generated content spread in social communities is closely analyzed by the model. The attribute News is refined into attributes Content and Sentiment; the former analyzes whether the web publication includes suspicious phrases (e.g., increase in revenue, new product development), and the latter captures the sentiment expressed within the news. This input data is provided by the developed services and described in (Grčar, 2012; Smailović, Žnidaršič, & Grčar, 2012). 3.2 Model Description Based on further interviews, the attributes structure was transformed into a hierarchical tree (Alić et al., 2013), with the root node ‘Pump and Dump’ in the following P&D, based on a qualitative- multi-attribute-method as proposed in (Bohanec & Zupan, 2004), 20 Irina Alić differentiating between the pure user-generated-content data (News) and the heterogeneous data regarding the company and related financial instruments (Comp_FinInst). Hence, the tree consists of the two sub-trees: one for ‘Company’ and its related financial instrument ‘Comp_FinInst’ and the second one for ‘News’. The proposed model aggregates the attributes into assessment of pump-and-dump market manipulation: CountryBlackList IndustryBlackList BlackLists CompanyBlackList Company Age History Bankrupt Comp_FinInst MarketSegment Market MarketCapitalization Pump and Dump FinancialInstrument (P&D) TradingVolume Trading Sentiment News Nb. Of Trades Content Figure 1: The hierarchical tree of attributes (Alić et al., 2013) Attribute scales: For each attribute, the qualitative values are scaled in the range from highly suspicious (red colored) to not suspicious (green colored), where v-low is an abbreviation for very low, and v-high an abbreviation for very high (Figure 2). Figure 2: The attribute scales (Alić et al., 2013) 21 Supporting Financial Market Surveillance: An IT Artifact Evaluation The scales for each attribute value are defined by the regulatory authority members and can be reconfigured when in use by the end user. In our case, for example, the default setting for the attribute ‘Market Capitalization’ is ‘low’ (highly suspicious) if the company’s value is under 5 million (given currency), ‘medium’ if the value is between 5 and 30 million, and ‘high’ (not suspicious) if the value is greater than 30 million. In the next example the attribute ‘Market Segment’ has two values ‘no’ and ‘yes’. In this case the value ‘no’ means that the stock is not traded at the market segment with low regulations. Accordingly, the value ‘yes’ means, that the stock is traded at the market with low regulations (e.g. OTC market). Manipulation scheme indicators: Within the predefined timespan the proposed calculation intends to identify abnormal changes which can be seen as indicators (Eren & Ozsoylev, 2006; Goldstein & Guembel, 2008; Zaki, Diaz, & Theodoulidis, 2012) for pump-and-dump abuse. Hence, the suspiciousness is assessed as follows: Firstly, to assess recent changes in trading, long-and short-term average trading volumes are computed by taking the monthly and three-day averages of the trading volume: where TVi is the trading volume of the i-th day; n = 3 days; m = 30 days. Secondly, to assess recent changes in number of trades, long-and short-term average trading volumes are computed by taking the monthly and three-day averages of the trading volume: where NTi is the number of trades of the i-th day; n = 3 days; m = 30 days. 22 Irina Alić Additional indicators for pump-and-dump market manipulation are calculated accordingly as presented in Table 1. Name Description / Definition Sentiment Long- Sentiment of news based on assessment of long-term sentiment. Term Period Based on the overall picture of the mood of the news = ( =1 )/ Sentiment Sentiment of news based on short-term sentiment (daily). Based Short-Term on the overall picture of the mood of the news of one to three days Period ℎ = (=1 )/ User-Generated Content of news based on assessment of specified terms. Based on Content Long- the overall picture of the mood of the news Term Period = (=1 )/ User-Generated Content of news based on short-term specified terms (daily). Content Short- Based on the overall picture of the mood of the news of one to Term Period three days ℎ = ( =1 )/ Table 1: Calculation of average values of the input variables Thirdly, in order to calculate jumps in e.g. price (Frieder & Zittrain, 2008), the deviation of the short-term as related to the long-term average is calculated by dividing the short- term average by the long-term average and multiplying by 100, as presented in Figure 3. Three cases are assessed: when the short-term value is smaller than, greater than, or equal to the long-term value. Suspiciousness is assessed using aggregated numerical input values, which are then mapped according to qualitative scales, as defined by the problem owner as high, med and low or as v-high, high, med, low, and v-low. Accordingly, structured data (such as e.g. trading volume) and unstructured data (such as e.g. user-generated content) are thereby taken into account in order to identify abnormal changes that may be indicators of pump-and-dump market manipulations. The recalibration of the indicator values or even the deployment of predefined default values can be adjusted by the end user. 23 Supporting Financial Market Surveillance: An IT Artifact Evaluation Indicators: Low= less than 50% Med=between 50% and 150% High= more than 150% Short term (Short / Long) *100 V-low = between 0% Method for formalizing the business idea -> Identifying the extent of change, or gap -> Feeding forward BMI objectives -> changing and feeding back the strategic position with operational and economic performance. 3 Beta-testing the analysis of BMI agility We chose four European SMEs providing KIS to ‘beta-test’ our idea and discuss the reasoning. The selection of cases (Table 1) was motivated by both theoretical and pragmatic considerations so that they represent differing industries and have differing approaches to BMI, and the researchers had access to the case data. We applied design case approach (Sein et al., 2011; Van Aken and Romme, 2009) focusing on developing new business models. Case 1: Case 2: Case 3: Case 4: Big data analytics PA prescription & Mobile operator Wind turbine platform service measurement service service diagnosis service Industry Information Health & Wellbeing Telecom Energy Technology Company size SME SME SME SME Maturity Growth Start-up Mature Mature Location Netherlands Finland Finland Denmark Market National National Local International B-to-B B-to-B, B-to-C B-to-B, B-to-C B-to-B Table 1: Case organisations Case 1 is developing service platform that would make big data analytics more accessible for Small and Medium sized Enterprises (SMEs). It is a knowledge intensive service aiming at new market disruption, opening a new business branch for the company. The case focused on utilising technological advancements in cloud computing, mainly virtualization and reduced cost of storage. After initial development rounds, SMEs were found to experience different kind of needs: varying access and collaboration mechanisms that required major design revisions and change of goals for the development. Only after piloting could the developers renew the technical architecture to achieve specified goals and requirements, which called for using STOF BM-ontology (Bouwman et al, 2008). Two iterations were made starting from value proposition for the SMEs, while the other customer segment was analysed in less detail. A marketing strategy was defined in order to achieve critical mass for the platform, but in practice lead customers were targeted to generate cash flow for further development. Because the case was technology driven the major development was focused on fast download speed, open source infrastructure technologies, development of APIs for data providers, as well as device agnostic applications, etc. with extra attention to security and management of users' profiles. Furthermore, the 51 Jukka Heikkilä, Marikka Heikkilä, Harry Bouwman business ecosystem was described in detail including value-, and information flows and actual operational processes in the latter stages of development, where after costs were estimated, risks assessed, a pricing strategy defined, and the revenue model, a mixed model combining a commission, a one time dataset download fee and a licence fee for longitudinal data, was developed. The BM was then evaluated by project team and external experts and customers using criteria such as completeness, consistency, viability, scalability and sustainability. The conclusion was that the business model was not yet complete and viable, and the project was discontinued. We can depict the development in Figure 5 as example 1, where technology dominates the design and BMI starts late. Case 2 of Physical Activity Prescription and measurement service, is a reincarnation of a well-known, proved idea with new technology and business model, and has characteristics of both Low-End Disruption and New-Market Disruption. Ideation was started years earlier by an entrepreneur willing to expand from pharmacy consulting towards novel platform based data fusion and analysis services for b-to-b customers in occupational health. The idea was developed into initial business model CANVAS in a project with a network of business partners, some of whom were incumbents having good contacts to potential customers. With the help of an IT solution provider the entrepreneur defined the information systems requirements for information exchange between the parties. The project group piloted the service with real customers. Based on a set of business model metrics (Heikkilä et al., 2014) the pilot indicated that one business model would not fit all customer segments, and introducing varying customer segments complicated the technical requirements. Furthermore, one of the incumbents questioned the benefits from committing to risky and complex network of partners. After three years of the business modelling design kick-off, the service innovation development was suspended for the time being by the incumbent. In Fig 5. the case can be illustrated with the market opportunity in the beginning that after tentative BM is aiming at joint technological solutions, but the complications of the technology and unsolved BM problems halt the BMI. Case 3, a company of 300 employees providing mobile and Internet access and services was willing to redefine the incumbent dominated, oligopolistic market by introducing mass-customized services for their existing and potential clients. It is a low- end disruption strategy providing better, more agile and affordable service with the leap of new technology. The work started with a consultant’s recommendation of building a specific information technology solution as add-on to existing operating model. The management realised that there were inherent discrepancies of the suggested technical solution with various client segments’ needs and desired objectives of winning new markets. Business modelling discussion started to clarify the design into a level of service technology design and building on automated common infrastructure and service components to all customer segments, distinct from customer segment specific processes. CSOFT was used for innovating the BM. The case changed quickly in the beginning from technology dominant design to parallel technology and business model development, and finally into iterative, alternating design of technology and business model. It also appeared that the metrics of guiding the process and design outcome were different for the business model and technology development in order to maintain the integrity of strategic objectives. The business model has been implemented for three years now, and the company has been expanding according to its strategic objectives in a profitable manner, taking advantage of the separate realms, and performance measures of technology development and business model implementation with mergers and acquisitions (Fig 5, example 3). 52 Business Modelling Agility: Turning ideas into business Figure 5 Proposed BMI processes Case 4 concerns a SME company in the wind energy industry. The case is an example of servitization. The company intended to exploit patented prototype add-on technology to increase wind turbine efficiency by reducing turbine misalignment with dominant wind flow based on Low-End Market -disruption. One-off sales BM was to replaced, or at least tried with an alternative mass-customized service-oriented BM, starting with competing technologies and stakeholders analysis. The requirements analysis followed in two-layers: requirements for the BM design as well for the service offering. Interviews with lead-users from different customer groups within the value chain were executed to get insights in the service offering, as well as interviews & design workshop within the company were conducted to get an understanding of the requirements for the BM design, and the availability of companies’ resources and capabilities. The results were continuously confronted for attestation on usability and on the BM. As a consequence present one-off sales got replaced with a two-step approach: First offering a cheap software based wind turbine diagnosis service, based on which the company will be able to estimate potential efficiency loss. At second stage an add-on hardware device can correct the misalignment and its cost to the customer is charged based on performance improvement. This two stage BMI appeared universally attractive to all newer customer segments, where the diagnostic can be done with software. The technological know-how and production capabilities reside in-house to provide both hardware as well as complementary software in a bundle, but the iterative critical reviews and analyses of the product in parallel with BM development did really lead to an alternative BMI without keen competition. Case 53 Jukka Heikkilä, Marikka Heikkilä, Harry Bouwman 4 followed the last process, example 4, depicted in Fig 5, that is the business modelling, technology development and customer validation were carried out in cyclical manner. 4 Lessons Learnt and Further Research Our study is a tentative probe into BM and BMI, illustrated with four select design cases. We intend to use the results for developing more lightweight, BM method for Knowledge Intensive Services entrants’ fast implementation of BMI for different strategic situations and grounding the development of the method on sound research. The better understanding of the strategic intent is crucial to concentrate resources in improving existing products for existing customer base with Sustaining Strategy, or to introduce disruptive technological solutions with BMI on Low-End Disruption and New- Market Disruption. Against this backdrop, we claim that innovation calls for the utilization of new technology and novel business models simultaneously. Simultaneously means in this context that they are considered in together, preferably in the market context, because the business model can then act as a boundary object between the stakeholders, customers and developers (Heikkilä, 2010; Heikkilä & Heikkilä, 2013). It also can act as an intermediator of the external market forces (reception and competition) similarly as agile product development probes the alternative solutions and user experiences. Balanced view on both is necessary in quick BMI realization - emphasising too much of the other seem to have negative, and different, problematic consequences on the speed of the implementation of the BMI. The cases are divers, but nevertheless they indicate the need for rapid iteration. Iteration can be implemented by dividing innovation in parallel product development and Business Modelling streams and by maintaining integrity between the two by rapid review cycles, or by balancing the performance measures between product development and BM according to the strategic intent. This helps to maintain customer focus and meet the market requirements early enough. The postulated need for iteration is well in line with some recent findings in KIS: “a SaaS client’s agility to leverage external resources, reconfigure its internal resources for strategic move and service provision, and mitigate the effect of environmental turbulence plays a critical role in competitive performance.” (Chou et al., 2014). In other words, in response to the turbulent market and technological environment an organization must be constantly and quickly scanning and responding to the changes by reconfiguration of internal and external resources and activities. In this paper we illustrated our ideas with four illustrative design cases. Further research on agility in BMI is clearly needed. Our intention is to confirm the findings with a larger data set collected from various entrants in the forthcoming years. Acknowledgement The authors wish to thank Lyubomir Nedyalkov and Stefan Marges for data collection and analyses in two of the cases. The work leading to these results has received funding from Data to Intelligence (D2I) Programme and the European Community’s Horizon 2020 Programme (2014–2020) under grant agreement 645791. The content herein reflects only the authors’ view. The European Commission is not responsible for any use that may be made of the information it contains. References Beck, K. & Andres, C. (2005). Extreme Programming Explained: Embrace Change. Boston: Addison-Wesley. Bouwman, H., M. De Reuver, S. Solaimani, D. Daas, T. Haaker, W. Janssen, Iske, P. & Walenkamp, B. (2012). Business Models, Tooling and Research Agenda. In R. Clark, A. Pucihar & J. Gricar (Eds), The first 25 years of the Bled Conference. Kraj: Moderna organizacija. 54 Business Modelling Agility: Turning ideas into business Bouwman, H., de Vos, H. & Haaker, T. (2008). Mobile Service Innovation and Business Models. Berlin: Springer-Verlag. Bouwman, H., de Reuver, M., Hampe, F., Carlsson, C. & Walden, P. (2014). Mobile R&D prototypes; what is hampering market implementation? International Journal of Innovation and Technology Management, 11(1). doi: 10.1142/S0219877014400033 Brandenburger A., M., & Nalebuff, B. J. (1997). Co-opetition, Currency Doubleday, 1st ed., 304 p. Bullinger, H-J., Fähnrich, K-P. & Meiren, T. (2003). Service Engineering-methodical development of new service products. Int. Journal of Production Economics, 85(3), 275-287. Casadesus-Masanell, R., & Ricart, J. E. (2010). From strategy to business models and onto tactics. Long range planning, 43(2), 195-215. Chan, L.-K. & M.-L. Wu (2002). Quality function deployment: A literature review . European Journal of Operational Research, 143, 463-497. Chiaroni, D., Chiesa, V., & Frattini, F. (2011). Organisational modes for Open Innovation in the bio-pharmaceutical industry: An exploratory analysis. Technovation, 31, 34- 43. Christensen, C.M., (1997). The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail. HarperBusiness. Christensen, C.M., (2006). The Ongoing Process of Building a Theory of Disruption. The Journal of Product Innovation Management, 23, 39-55. Christensen, C.M., & Raynor, M.E., (2003). The Innovator’s Solution. Harvard Business School Publishing, MA. Christensen, C.M., Grossman, J.H., & Hwang, J., (2009). The Innovator’s Prescription: A Disruptive Solution for Health Care. New York: McGraw-Hill. Chou, S-W., Hsu, C.S., Min, H-T., Chiang, C.-H., & Chang, Y-C. (2014). Understanding Competitive Performance of Software-as-a-Service (SAAS) — The Competitive Dynamics Perspective. PACIS 2014 Proceedings. Paper 362. http://aisel.aisnet.org/cgi/viewcontent.cgi?article=1025&context=pacis2014 Coad, P., De Luca, J., & Lefebre, E. (1999). Java Modeling in Color. Englewood Cliffs, NJ: Prentice Hall. Cohen, L., (1995). Quality Function Deployment: how to make QFD work for you. Reading, Mass: Addison-Wesley. Cooper, R.G. (2000). Stage gate systems: a new tool for managing new products. Business Horizons, may-June, 44- 54. Cortimiglia, M. N., Ghezzi, A., & Frank, A. G. (2015). Business model innovation and strategy making nexus: evidence from a cross‐industry mixed‐methods study. R&D Management. Dan, Y. & Chieh Hang Chang (2008). A Reflective Review of Disruptive Innovation Theory. PICMET 2008 Proceedings 27-31 July, Cape Town, South Africa, pp. 402-414. Dedehayir, O., Ortt J.R., & Seppänen M. (2014). Reconfiguring the Innovation Ecosystem: An Explorative Study of Disruptive Change , International Conference on Engineering, Technology and Innovation 2014, 23-25 June 2014, Bergamo, Italy. De Reuver, M., Bouwman, H. & Haaker, T. (2013). Business model roadmapping: A practical approach to come from an existing to a desired business model. International Journal of Innovation Management, 17, 1-18. Doz, Y.L., & Kosonen, M. (2008). Fast Strategy: How Strategic Agility Will Help You Stay Ahead of the Game. Pearson Education. El Sawy, O. A., & Pereira, F. (2012). Business Modelling in the Dynamic Digital Space: An Ecosystem Approach. Springer Science & Business Media. Galbraith J. R. (2014). Designing Organizations: Strategy, Structure, and Process at the Business Unit and Enterprise Levels, 3rd ed. Jossey-Bass, 352 pages. Gnyawali, D.,R., & Park, B.-J., (2011). Co-opetition between giants: Collaboration with competitors for technological innovation, Reseach Policy, 40, pp. 650-663. Goldstein, S., Johnston, R., Duffy, J. & Rao, J. (2002). The service concept: the missing link in service design research?. Journal of Operations Management, 20(2), 121-134. 55 Jukka Heikkilä, Marikka Heikkilä, Harry Bouwman Govindarajan, V. & Kopalle, P. K. (2006). The Usefulness of Measuring Disruptiveness of Innovations, Ex Post in Making Ex Ante Predictions. Journal of Product Innovation Management, 23 (1), 12-18. Heikkilä, M., & Heikkilä, J. (2013). Collaborative Business Model Innovation Process for Networked Services. In J. Järveläinen, H. Li, A.-M. Tuikka, & T. Kuusela (Eds.), Co- created Effective, Agile, and Trusted eServices (in Series: Lecture Notes in Business Information Processing Vol. 155, pp. 133–147). DOI: 10.1007/978-3-642-39808- 7_12 Heikkilä, J., Tyrväinen, P. & Heikkilä, M. (2010). Designing for performance - a technique for business model estimation. Seppä, M., Helander, N. & Ilvonen, I. (eds.) Proceedings of EBRF, Research Forum to Understand Business in Knowledge Society. Heikkilä, M. (2010). Coordination of complex operations over organisational boundaries, Phd. thesis, University of Jyväskylä, Jyväskylä Studies in Computing 111. Heikkilä M., Solaimani, S., Soudunsaari, A., Hakanen, M., Kuivaniemi, L. & Suoranta, M. (2014). Performance Estimation of Networked Business Models: Case Study on a Finnish eHealth Service Project. Journal of Business Models, 2 (1), 71- 88. Heikkilä, M., Saarni, J., Kaartemo, V. & Koponen, A. (2015). Viability radar – a practical tool for assessing viability of transformative service innovations in the health care context. Technology Innovation Management Review, 5(3). Hung S.-V. & Chang C.-C. (2012). A co-opetition perspective of technology alliance governance modes, Technology Analysis & Strategic Management, 24(7), 679-696, DOI: 10.1080/09537325.2012.705120 Karni, R. & Kaner, M. (2007). An engineering tool for the conceptual design of service systems. In Spath, D. & Fähnrich, K-P. (eds.) Advances in Services Innovations. Springer. pp. 65-85. Ktata, O., & Lévesque, G. (2009). Agile development: Issues and avenues requiring a substantial enhancement of the business perspective in large projects. In proceedings of the 2nd Canadian conference on computer science and software engineering (pp. 59-66). ACM. Kim, W.C. & Mauborgne, R. (2005). Blue Ocean Strategy: How to Create Uncontested Market Space and Make the Competition Irrelevant, Harvard Business School Press, 256 pages. Kunz, V.,D., & Warren L. (2011). From innovation to market entry: a strategic management model for new technologies, Technology Analysis & Strategic Management, 23(4), 345-366, DOI: 10.1080/09537325.2011.558388 Lee, G. & Xia W. (2010). Toward Agile: An Integrated Analysis Of Quantitative And Qualitative Field Data On Software Development Agility. MIS Quarterly 34(1), 87- 114. Osterwalder, A. & Pigneur, Y. (2010). Business Model Generation: A Handbook for Visionaries, Game Changers, and Challengers. Wiley, 288 p. Markides, C. (2006). Disruptive Innovation: In Need of Better Theory. The Journal of Product Innovation Management, 23, 19–25. Menor, L., Tikonda, M. & Sampson, S. (2002). New Service Development: Areas for exploitation and exploration. Journal of Operational Management, 20(2), 135-157. Norman, D. (1998). The Invisible Computer: Why good products can fail, the personal computer is so complex and information appliances are the solution. MIT Press, London, England. Ritala, P. & Hurmelinna-Laukkanen, P. (2009). What’s in it for me? Creating and appropriating value in innovation-related coopetition. Technovation 29(12), 819– 828. Ritala P. & Sainio, L-M. (2014). Coopetition for radical innovation: technology, market and business-model perspectives. Technology Analysis & Strategic Management, 26(2), 155-169. Schwaber, K. & Beedle, M. (2002). Agile Software Development with Scrum, Upper Saddle River, NJ: Prentice-Hall. 56 Business Modelling Agility: Turning ideas into business Sein, M.K., Henfridsson, O., Purao, S., Rossi, M. & Lindgren, R. (2011). Action Design research. MIS Quarterly, 35, 37-57. Shostack, L.G. (1984). Designing services that deliver. Harvard Business Review. 62(1), 133-139. Sinha, R.K. & Noble, C.H. (2005). A model of market entry in an emerging technology market. IEEE Transactions on Engineering Management 52(2), 186–98. Stapleton, J. (1997). DSDM: Dynamic Systems Development Method. Harlow, England: Addison Wesley. Teece, D. J. (2010). Business Models, Business Strategy and Innovation, Long Range Planning, 2–3, 172–94. Van Aken J. & Romme, G. (2009). Reinventing the future: adding design science to the repertoire of organization and management studies. Organization Management Journal, 6, 5-12. Wessel, M. & Christensen, C.M. (2012). Surviving Disruption. Harvard Business Review, Dec. 2012. Wohlfeil, F. & Terzidis, O. (2014). Critical Success Factors for the Strategic Management of Radical Technological Innovation, International Conference on Engineering, Technology and Innovation 2014, 23-25 June 2014, Bergamo, Italy. Zott, C. & Amit, R. (2010). Business model design: an activity system perspective. Long range planning 43, 216-226. 57 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Digital Business Engineering: Methodological Foundations and First Experiences from the Field Boris Otto TU Dortmund University, Germany Boris.Otto@tu-dortmund.de Rieke Bärenfänger University of St. Gallen, Switzerland Rieke.Baerenfaenger@unisg.ch Sebastian Steinbuß Fraunhofer ISST, Germany Sebastian.Steinbuss@isst.fraunhofer.de Abstract Digitization is affecting almost all areas of business and society. It brings about opportunities for enterprises to design a digital business model. While a significant amount of research exist examining the conceptual foundation of business models in general, no comprehensive approach is available that helps enterprises in designing a digital business model. This paper addresses this gap and proposes Digital Business Engineering as a method for digital business model design. The activities are structured into six phases, namely End-to-End Customer Design, Business Ecosystem Design, Digital Product/Service Design, Digital Capability Design, Data Mapping, and Digital Technology Architecture Design. The method development follows principles of design- oriented research. Five case studies are used to analyse method requirements and evaluate it within is natural context. Keywords: Digitization, Business Model; Digital Business Engineering, Method Engineering, Case Study 1 Introduction 1.1 Research Motivation Digital Business is a term that has recently created much attention both in the scientific and in the practitioners’ community. Driven by native digital companies such as Facebook and Google as well as by start-up businesses such as MyTaxi, MyDryClean, 58 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß and Uber and many enterprises see themselves confronted with questions for future revenue streams, customer segments, new market entrants and innovative operational models. Along with the proliferation of digital businesses, both researchers and practitioners are frequently discussing the concept of business models. Consensus exists with regard to the question as to what a business model is (Alt and Zimmermann 2001; Hedman and Kalling 2003; Zott et al. 2010). Apart from that, there is a significant amount of contributions on electronic business models (Osterwalder and Pigneur 2002; Timmers 1999; Gordijn et al. 2000). Furthermore, some contributions address the task of business modelling. A prominent technique supporting this task is the so-called Business Model Canvas (Osterwalder and Pigneur 2010). Another approach that gained significant attention both among practitioners and researchers is Business Engineering (Österle 1996; Österle and Winter 2003) providing methodological support when it comes to business transformation induced by information technology (IT). However, the design of a digital business model is still relatively unexplored. The research community proposes first approaches for particular questions in the design of a digital business model. Otto and Aier (2013), for example, examine business models in the data economy. Krishnan et al. (2007) study business models of peer-to-peer networks. A comprehensive approach, though, is not available yet. Furthermore, a study conducted by consulting company McKinsey finds that barriers to digital business models include management, organizational, and technical aspects ranging from lacks of expertise to poor data quality, for example (Brown and Sikes 2012). It is this gap in the understanding of how to design digital business models which motivates the research presented in this paper. 1.2 Research Goal This research aims at methodological support for designing digital business models. The paper proposes Digital Business Engineering as a comprehensive methodological framework. The paper takes a design-oriented approach to the object of investigation. It wants to understand the underlying means-end principles of digital business model design, rather than corresponding cause-effect relationships (Winter 2008). Five case studies are used to analyse method requirements and evaluate it within is natural context. From an epistemological perspective, the paper is positioned in the design-theoretical (cf. Gregor 2006) realm, i.e. it aims at contributing to the scientific knowledge base while at the same time being useful for practical application. The contribution to the scientific knowledge base stems mainly from the methodological foundations of the Digital Business Engineering approach. Practitioners may benefit from the results as the method fragments help structuring and accelerating digital business modelling activities in enterprises. In particular the six phases in which the activities of the method are structured give guidance to digital business modelling efforts in practice. The remainder of the paper starts with a presentation of related work, which also introduces the basic conceptualisation of the research. The third section lays out the research approach, before section 4 presents the method itself. Selected method components are presented as they are used in the corresponding case studies. Section 5 59 Digital Business Engineering summarizes and interprets the findings that result from the case studies. The paper closes with a conclusion section. 2 Related Work 2.1 Business Model Research Business model research is rooted in the resource-based view (RBV) of an enterprise. Scholars from management science introduced RBV to the academic discourse, for example Barney (1991). RBV proposes that competitive advantage originates from strategic enterprise resources. Strategic resources meet the so-called VRIN criteria, i.e. they are valuable, rare, inimitable and non-substitutable (Dierickx and Cool 1989). RBV as an exploratory model for competitive advantage gained significant attention in the late 1980s and early 1990s not only on the scientific community, but also among practitioners, mainly driven by contributions such as the book on competitive advantage by Porter (1985). However, resources themselves are not able to create value (Bowman and Ambrosini 2000). The creation of value is brought about by competencies, which are portrayed as the ability to deploy combinations of firm-specific resources to accomplish a given task (Teece et al. 1997). In this context, an organizational capability is the “ability of an organization to perform, across individuals or groups, a coordinated set of tasks, utilizing organizational resources, i.e. tangible or intangible assets and inputs for production, for the purpose of achieving a particular end result” (Helfat and Peteraf 2003). In addition to explicit elements such as methods and functions, capabilities also comprise tacit elements, such as knowledge of individuals or leadership. Of course, RBV is only one conceptualization of business models. Competing views see business models as activity systems (e.g. Zott and Amit 2010) or even ingrained strategic orientations (e.g. Aspara et al. 2010). However, the paper follows the RBV perspective on business models, mainly because it views the activity of business modelling as an organizational capability. 2.2 Digital Business Models The term “digital business” is experiencing a renaissance at present. While it was initially coined in the 1990s, it is today used in broader context. The traditional understanding of digital business was very much influenced by the debate around treating information as an enterprise asset (Horne 1995; Oppenheim et al. 2001). This perspective on digital information acknowledged the important role that information plays for enterprise success. However, it mainly materialised in the digitization, often “electronification”, of business processes. Today’s notion of digital business, though, takes a business model view looking at the enterprise as a whole and asking what opportunities digitization brings about to transform and advance current business models. Research groups embracing this perception of the term “digital business” have formed across the globe. Examples are the MIT Center for Digital Business (The MIT Center for Digital Business 2015) and the research programme “Digital Business Transformation” at the University of St. Gallen (Leimeister et al. 2014b). Furthermore, 60 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß the European Commission published recommendations for the transition to the data- driven economy (European Commission 2014). The practitioners’ community is discussing the fundamental principles of digital business models, too, and came up with first recommendations. In Germany, for example, the Smart Service Welt Working Group (2014), which consists of delegates from industrial partners, research as well as policy makers, investigates business models around so-called smart services. One fundamental design principle of such services is consumer-centricity (Leimeister et al. 2014a) and, closely related to that, multi-channel integration. The term consumer-centricity looks at the individualization trend from a business perspective. Furthermore, companies reach out to their customers via many channels, not just the direct sales channel, and keep track of one unique customer identity across those channels. Another indication for the increased focus on the consumer is the involvement of customers in the value creation process, for example through crowdsourcing, which changes the consumer role into a “prosumer” (Ernst & Young 2011). In the digital economy, products are evolving into “hybrid service offerings”. Traditional products become increasingly computerized and “smart” thanks to the close integration of IT into physical products. Examples are embedded software systems in modern cars (“Car IT”) and “wearables”, clothes with integrated computer chips. Companies try to gain more from additional services around the core product (Yoo et al. 2010). While some research exists on digital business models, the practitioners’ community is experiencing a number of drawbacks in leveraging the full potentials of digitisation. A McKinsey study points to various barriers among which are inappropriate organisational structures, lack of IT systems, lack of IT and business expertise, lack of leadership and poor data quality (Brown and Sikes 2012). Apparently, there is a need for methodological support when it comes to digital business. 2.3 Business Modelling Methodologies Business models are of conceptual nature. Thus, many proposals about the constituents of a business model are presented as conceptual models (cf. Hedman and Kalling 2003). The process of creating such a conceptual model is referred to as business model design and follows the general principles of model design. Early research on business model design stems from the 1990s. Business Engineering, for example, is a model-oriented and method-driven approach for managing IT-induced transformation (Österle 1996; Österle and Winter 2003). It integrates different views of an enterprise, mainly business strategy, business processes, and information systems and can be operationalised using a set of individual methods. Later on, the research community proposed further methods, which often focused on a particular aspect of a business model. MacInnes and Hwang (2003), for example, focussed on peer-to-peer business models, Timmers (1999) and also Alt and Zimmermann (2001) examined business models of electronic markets, and Wirtz et al. (2010) provided guidance for internet business models using Web 2.0 ideas. 61 Digital Business Engineering Apart from that, De Vos and Haaker (2008) discuss how to apply the STOF business modelling method in practical steps, and Heikkilä and Heikkilä (2013) suggest a practical approach to use their C-SOFT business modelling method in action design research. There are also some methods that focus on specific elements of business model design and evaluation, e.g. the business model roadmapping method by De Reuver et al. (2013). Recently, the Business Model Canvas proposed by Osterwalder and Pigneur (2010) gain much attention in the scientific community, but even more among practitioners. Similarly, Gassmann et al. (2013) propose a set of practical business model blueprints. Some initial work is available for developing digital business models (Berman 2012; Bharadwaj et al. 2013; Eichentopf et al. 2011). However, a comprehensive approach that covers all the various concepts of digital business models is not available yet. 3 Research Design 3.1 Research Process The paper aims at designing a method to guide the process of digital business modelling. In general, methods are typical design artefacts (March and Smith 1995; Hevner et al. 2004) as they embody the scientific knowledge about means-end relationships for a phenomenon under investigation. Thus, the design of the Digital Business Engineering method follows Design Science Research (DSR) principles. Consortium Research Phase Activities/Methods Expert interviews Focus group workshops Analysis Case study research Analysis of literature in the scientific and practitioners’ domain Business Engineering as a conceptual foundation Design Method Engineering as a design paradigm Participative case studies Expert interviews Evaluation Case studies Presentation at practitioners’ conferences Diffusion Present research paper Table 1: Consortium Research Approach Over the last decade, a number of guidelines emerged supporting the DSR process. A prominent example is the Design Science Research Methodology proposed by Peffers et al. (2007). The majority of approaches has in common that a DSR process is iterative in nature and combines both scientific and practitioners’ knowledge during the artefact design. As in particular the latter is of paramount importance to achieve both scientific knowledge accumulation and practical utility, the paper follows consortium research as a methodological frame. Consortium research is a multilateral form of DSR in which researchers work closely with practitioners over a significant amount of time. Practitioners contribute their knowledge and test the design artefacts regularly within their organisational environments (Österle and Otto 2010). Consortium research consists 62 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß of four phases, namely analysis, design, evaluation, and diffusion. The cycle itself is conducted repeatedly and typically, researchers perform multiple iterations within the four phases. As Table 1 shows, key to the research process, in particular for data collection, are case studies and expert interviews for analysis and evaluation purposes. Furthermore, Method Engineering forms the conceptual foundation for the method design. 3.2 Data Collection The design of the Digital Business Engineering method requires data from the field for requirements specification and artefact evaluation purposes. Data was collected via two research methods, namely case studies and expert interviews. Case study research is adequate when a relatively new phenomenon is investigated that cannot be separated from its organizational environment (Yin 2014) - as in the case of Digital Business Engineering. The case studies used various data sources as input, such as interviews with company representatives, internal and external documentation and material. Period of Data Type of Key Other Case Industry Country Collection Case Study Experts Material A FMCG DE 02/2012-10/2012 participative Head of Presentation Supply on industry 7 interviews of 4 Chain Data event hours Managemen t, PM Digital Marketing B Retail CH 12/2010-12/2013 explorative Head of Two Customer presentations 3 interviews of 2 Intelligence, at industry to 3 hours PM Web events Intelligence C Online DE 12/2012-09/2014 explorative Head of BI Presentation Fashion on industry 2 interviews of 2 Retail event hours D Insurance CH 09/2014-01/2015 participative Head of Internal Innovation, documents 10 interviews of PM at least 2 hours Marketing E Power LI 01/2013-06/2014 explorative Digital Presentation Tools Business on industry 2 interviews of 2 Project event hours each Leader Table 2: Case Study Overview Table 2 shows key information about the five cases in this paper. The companies analysed in the case studies were members of the industry network of the Competence Center Corporate Data Quality (CC CDQ). The CC CDQ is a consortium research project aiming at the advancement of quality-oriented data management in large enterprises (Otto and Österle 2010). Two of the authors of the paper are part of the team of CC CDQ which forms the context of the study presented in this research. Companies in Cases A and B were regular consortium partners of the CC CDQ, the remaining companies were well-known companies from the wider project network. 63 Digital Business Engineering Cases A and D were participative, i.e. the researchers did actively engage with the case study company and did not limit their role to a purely observing one. Baskerville (1997) points to the difficulties that occur as a consequence of research participation in action research cases. However, case study research with a strong active part on the researchers’ side is more and more seen as useful in DSR settings, as the proposition and adoption of methods such as Action Design Research (ADR) shows (Sein et al. 2011).Apart from that, expert interviews were conducted to triangulate findings. The expert interviews were design as focus groups in order to leverage consensus-finding mechanisms that come with group set-ups (Chiarini Tremblay et al. 2010; Stewart et al. 2007). Table 3 shows the focus groups that were conducted within the research endeavour presented in this paper. The participants in the focus groups were delegates of the CC CDQ partner companies, i.e. mainly line and project managers responsible for enterprise-wide data and digitization activities. The fact that focus group participants only in some cases also were included in on-site case study interviews contributed to the triangulation objective. The focus group workshops started with an impulse presentation by the research team, an invited company presentation on the focus topic (with exception of the session on October 10th, 2013), and a moderated discussion. Date Time Location Focus Topics Participants 2012-06-14 09.00-16.00 h St. Gallen (CH) Consumer-centric 16 participants information management, from 10 companies Consumer services, Product information 2013-09-24 09.00-17.00 h St. Gallen (CH) Consumer-centricity, 16 participants Consumer services, Value from 14 companies of data 2013-10-10 10.00-11.00 h St. Gallen (CH) Business in the data- 46 participants driven economy from 23 companies 2014-04-30 09.00-12.00 h Munich (DE) Capabilities for Big Data 9 participants from management 6 companies 2014-06-26 09.00-10.30 h Stockholm (SE) Towards the data-driven 41 participants organization, business from 15 companies opportunities and needs for action, organizational capabilities 2014-12-11 08.45-12.15 h Berlin (DE) Capabilities for Big Data 10 participants management from 7 companies Table 3: Focus Group Overview 3.3 Method Engineering While Business Engineering forms the conceptual framework of the method, Method Engineering is used as a concrete design technique. Method Engineering stems from the software engineering domain and services the design of methods through the definition of method components and their relationships (Heym 1993; Nuseibeh et al. 1996). Methods give guidance for design and development processes by providing recommendations on the activities and techniques needed to achieve a certain result type (Brinkkemper 1996). 64 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß Gutzwiler (1994) identifies five components which constitute a method according to Method Engineering. First, a meta-model identifies and describes the relevant concepts (and their relationships) for the application domain of the method (Digital Business Engineering, in the case of this research). Second, result types describe the various outcomes of applying the method. Third, activities describe which steps must be carried out in order to achieve the result types. Fourth, roles are defined which perform the activities. Fifth, techniques are defined which have to be deployed within the activities. 4 A Method for Digital Business Engineering 4.1 Requirements and Method Overview Two sources of knowledge led to the requirements of Digital Business Engineering, namely analysis of literature and findings from the field. Table 4 summarises the functional method requirements. Besides, there are non-functional requirements which mainly stem from good modelling practice (such as usability, technical comprehensiveness etc.). However, these requirements are addressed implicitly by following a widely adopted modelling approach such as Method Engineering. A B C D Req. Description Supporting Literature e e e e Cas Cas Cas Cas R1 Comprehensive enterprise (Bharadwaj et al. 2013), X X X perspective (Brown and Sikes 2012) R2 Consumer-centric perspective (Ernst & Young 2011), X X X X (Leitner and Grechenig 2008), (Rajagopal and Burnkrant 2009), (Ross 2009), (Schuster and Dufek 2004) R3 Digital product/service (Leimeister et al. 2014a), X X perspective (Rajagopal and Burnkrant 2009) R4 Data-centric perspective (European Commission X X X 2014), (Newman 2011), (Otto et al. 2014), (Otto and Aier 2013) R5 Organisational capability (Berman 2012), (Yoo et al. X X perspective 2010) R6 Business ecosystem (El Sawy and Pereira 2013), X X perspective (Corallo et al. 2007) Legend: X - Requirement addressed in Case. Table 4: Digital Business Engineering Requirements Figure 1 shows an overview of the Digital Business Engineering Method. It comprises strategic, business process, and system technology aspects, thus providing an integrated approach for addressing both business and IT related design tasks. The method consists of six phases. 65 Digital Business Engineering Digitization 1 E2E Customer Process Design Strategic Perspective 2 Business Ecosystem Design 3 Digital Product & Service Design Process Perspective 4 5 Digital Capabilities Design Data Mapping Systems Perspective 6 Digital Technology Architecture Design Digital Business Model Figure 1: Digital Business Engineering Overview Phase 1 analyses the end-to-end customer process, which forms the ultimate starting point for digital business modelling. It is based on findings from research and practice that a key success factor in the digital economy is a comprehensive understanding of the future customer process, instead of focusing on optimizing the traditional customer- supplier interaction points. Phase 2 aims at understanding and designing the business ecosystem that must be in place to support the end-to-end customer process in a comprehensive way. Phase 3 deals with the design of digital products and services needed in the support of the customer process. Yoo et al. (2010), for example, propose a general architecture for digital artefacts. It is evident that digital products and services rely on organizational capabilities, which are subject of Phase 4. Digital capabilities are dynamic capabilities, which allow rearranging enterprise resources in order to make use of the digitization. Furthermore, digital artefacts (as a generic term for both digital products and services) require data of various kinds and from various sources. Otto et al. (2014), for example, analyse cases of the networked economy with regard to the data variety. Data can come from internal or external sources, be under the organization’s or under third-party ownership, be in different data quality, occur in streams or in batches, follow a certain schema or be unstructured. Thus, Phase 5 deals with data mapping making sure that the business objects required in the end-to-end customer process are transparent and that corresponding data objects are identified and described (including their format, occurrence, and source, for example). Finally, Phase 6 of Digital Business Engineering designs the digital technology architecture. Table 5 shows the components of the Digital Business Engineering Method, namely the goals, involved roles and techniques for all six phases. Section 4.2 introduces selected method components, in particular techniques, as they were applied in the case studies. The entire method was not applied in full in any of the 66 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß cases. However, the design research processes aggregates the findings of the five cases into a comprehensive methodological framework. DBE Description Goal Involved Roles Techniques Phase Customer Journey E2E Understand end-to-end Business Analysis, Customer Customer 1 customer process from an Development, Sales, Process Modelling, Process outside view Marketing Multi-Channel Design Analysis Business Business Understand actors within Development, Sales, SWOT Analysis, 2 Ecosystem customer process and Marketing, Product Network Analysis Design customer interaction points Development Digital Design digital products and Business Business Model Product and services based on end-to- Development, Sales, Canvas, Digital 3 Service end understanding of Marketing, Product Artefact Design, Design customer process Development Design Thinking Business Digital Identify capabilities needed Development, IT, Capability Reference 4 Capabilities to provide digital products Product Model Design and services Development Business Data Architecture Identify data assets needed Development, IT, Management, Data 5 Data Mapping to provide digital products Data Management, Mapping, Data and services Product Value Chain Development Digital Technology Design digital technology Data Management, Digital Architecture 6 Architecture architecture IT Management Design Legend: DBE - Digital Business Engineering; E2E - End-to-End; SWOT - Strengths, Weaknesses, Opportunities, Threats; IT - Information Technology. Table 5: Digital Business Engineering Method Components 4.2 Method Components 4.2.1 Customer Journey Analysis Figure 2 shows the first version of an end-to-end customer design technique used in Case D for the scenario “life insurance”. Internally coined as “customer journeys” the technique takes an outside-in perspective to the company. The process starts with the customers’ need for information and ends with an electronic invoice. Throughout the entire process, various digital technologies (e.g. social media, chats, digital signature) are deployed across multiple channels (e.g. Internet, e-mail, telephone, chats, communities). 67 Digital Business Engineering Figure 2: End-to-End Customer Design in Case D 4.2.2 Multi-Channel Analysis Figure 3 shows the use of Multi-Channel Analysis in Case B. The company used the technique to analyse the relationship between the customer process and the various interaction channels the company offers to the customer. The analysis shows that the general customer process includes nine steps, starting with information about a certain product and ending with service activities on the purchased product. Of course, not all steps are relevant in the traditional food division, but occur very often in the electronics division, which offers consumer electronics such as television sets, computers etc. In addition, nine channels exist through which interaction with the customer takes place. E2E Customer Process Track Information Availability Advice Sale Payment Shipment Return Service and Trace Print TV/Radio Store lsenn Internet Cha Mailing ion E-Mail ractteIn Phone Fax Text message Channel used in consumer process Exemplary E2E customer process. Figure 3: Multi-Channel Analysis in Case B A typical “path” across the various channels is as follows: A customer receives information about a new TV product via printed advertisements. If interested, he/she checks availability of the particular product via the company’s internet shop. The customer then goes into a store in order to have a look at the product (as a TV is a relatively expensive item). He seeks advice from a store employee, but purchases and 68 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß pays the product over the internet. When the product is available, he receives a text message on his mobile phone. The customer picks up the item in the store and might, in case something is wrong with it, also return it there. Service claims are then handled over the phone. In Case B, the multi-channel analysis was used for achieving understanding of the customer process, but mainly for making sure that information about the products (e.g. product features, availability, price etc.) is provided consistently across the different interaction channels. 4.2.3 Network Analysis Figure 4 shows the result of a business ecosystem analysis in Case A. The technique was used in the preparation phase of establishing a digital business responsibility in the company. The focus is on how product information is created, used and distributed through the company’s ecosystem. Product information comprises standard data such as product name, content, manufacturer information, and GTIN (Global Trade Identification Number), “value-added” information on allergen sensibility, ingredients, “how-to-apply” and “where to buy” information etc. The technique used network analysis to illustrate the “betweenness centrality” of the various actors in the ecosystem when it comes to controlling the flow of product information. Betweenness centrality equals the number of shortest paths from all notes on the ecosystem to all others that pass through that node. A node with high betweenness centrality has a large influence on the transfer of product information, thus, is considered most powerful (cf. Wasserman and Faust 1994). 2007 2012 GDSN Consumer Information Consumer Provider Technology Forums & Provider Blogs Consumer Information Provider Brand Owner Media Agency Consumer Agency Brand Owner Retailer Social Web shop Network Retailer Consumer Online Retailer Legend: GDSN - Global Data Synchronization Network Figure 4: Business Ecosystem Analysis in Case A The analysis was conducted with experts from the marketing, supply chain, sales and data management departments in Case A. Participants were asked to describe the ecosystem and the flow of product information through the ecosystem. They were asked to reflect on the current situation (2012) and five years ago. All paths are considered equally important, i.e. no weighing was applied. Results of the analysis were twofold. First, the number of actors in the ecosystem has increased from six to ten, i.e. the ecosystem became more complex. Second, the 69 Digital Business Engineering betweenness centrality changed. While the “brand owner” (the company in Case A) remained on the outer circles with a low value of betweenness centrality, the consumer gained much more power and moved from the periphery to the centre of the ecosystem. 4.2.4 Data Mapping Figure 5 shows the data map for digital services in Case E. The company operates in a direct sales and services model, thus, the data map is centred on data about products and services. Case E Relevant Data NB: Abbreviations are information systems acronyms used in Case E. Figure 5: Data Mapping in Case E In addition to these data, digital services use various internal data sources, both structured and unstructured. Structured data comes mainly from large enterprise information systems such as ERP and CRM whereas unstructured data comes from call centre activities, for example. A third data domain is external data, which comes from various sources such as data traders and social networks. Web Shop Catalogue/Print Relation Center E-Mail Offsite Decision Engine Further Channels Real-Time Layer Architecture Further Channels Model Management and Advanced Analytics Near-Real-Time and Batch Layer Architecture Reporting and Business Analytics Legend: Data Transfer; Model Deployment. Figure 6: Digital Technology Architecture in Case C 70 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß 4.2.5 Digital Technology Architecture Design Figure 6 shows the digital technology architecture in Case C. The online fashion retailing business requires decision making in almost real-time. For example, if a customer is likely to stop a purchasing activity, the digital technology architecture helps with customer churn prevention in real-time. Three components form the digital technology architecture, namely the decision-engine that is fed by mathematical and statistical models about customer behaviour. The real- time layer architecture is capable of analysing online shopping data. The near-real-time and batch layer architecture processes information about wish lists, customer master data etc. The architecture design follows the requirements of the company’s digital business model including multi-channel management and digital service design. 5 Findings from Method Application in the Field Using the method in the field led to a number of findings with regard to model design and its usefulness with regard to current barriers to digital business model development. First, the method facilitates the business modelling process as it provides a common language between multiple stakeholders from various departments in a company. For example, in Cases A and D the method components guided the activities in which employees from marketing, IT, business development, supply chain management etc. were involved. Second, the method helps to stay focussed on the customer perspective. Often, in the course of the digital business design processes, employees tend to concentrate on the existing customer interaction. In particular, the customer journeys in Case C helped to keep the outside-in perspective. A third finding relates to the issue of appropriate organisational structures. As pointed out by Brown and Sikes (2012), applying the method requires a clear mandate for action. In Case A, however, it was unclear whether marketing or sales were leading the digital business transformation initiative. The Digital Business Engineering method helped to structure the activities needed. However, the full potential could not have been reached. Fourth, the method needs refinement as in its present form it focuses on digital business. However, enterprises are unable of simply switching from traditional to digital business. A big challenge in all cases stemmed from the fact that the traditional business model must be kept while at the same time digital business transformation must be driven forward. That is why the company in Case D was not only developing customer journeys, but also “employee journeys” to make explicit the need for organizational change during digital transformation. Sixth, the method needs further development in order to be able to cope with different levels of digital business maturity. For example, in Case A it turned out that the sales organisation in the United Kingdom was much more advanced in terms of digital business compared to the markets in Germany and Switzerland. Finally, Digital Business Engineering takes an enterprise perspective, but leads to individual digital products and services. It does not answer the question, though, how 71 Digital Business Engineering experiences from one area can be transferred (our scaled) across the rest of the enterprise. 6 Conclusions 6.1 Result and Contribution The research presented in this paper addresses a gap in literature as well as in practice. Both communities observe a lack in methodological support in designing digital business models. Digital Business Engineering is proposed as a method for digital business model design. The paper contributes to the scientific body of knowledge as it is among the first research results that identifies guidelines needed for digital business transformation. The method embodies both scientific knowledge and knowledge from the practitioners’ community. It is an artefact which instantiates the general principles for digital business design methods. The method is beneficial for practitioners. In particular, the participative case studies showed that Digital Business Engineering is useful in “real-life” situations. 6.2 Limitations and Outlook to Future Research Qualitative research in general is limited with regard to validity and generalizability. The research uses five cases and focus group interviews for analysing the requirements and evaluating the method. The paper illustrates the method through describing and analysing the application of method components in the five cases. Further research is needed to validate the structure of the method, the method components and their relationships, e.g. through a more rigorous case analysis (within case and cross-case). Apart from that, the method was not applied in full in any of the five cases. All components were used at least once, but not in one comprehensive case. Thus, the usefulness and applicability of the method as a whole has not been evaluated yet, as well as the efficacy of the method. Apart from that, the method components are to some extent immature and need refinement. One example is the lack of conceptual depth of a meta-model. The method in its current form is a result of a first design iteration. Future research activities should include incorporating the results of the focus groups into the next design iteration. Further research opportunities lie also in the identification of general design principles for digital business methods, which could then form the foundation of a design theory for this particular domain. References Alt R & Zimmermann H-D. (2001). Business Models. Electronic Markets 10 (1), 3-9. Aspara J, Hietanena J & Tikkanen H. (2010). Business model innovation vs replication: financial performance implications of strategic emphases. Journal of Strategic Marketing 18 (1), 39-56. DOI 10.1080/09652540903511290. Barney J. (1991). Firm Resources and Sustained Competitive Advantage. Journal of Management 17 (1), 99-120. DOI 10.1177/014920639101700108. Baskerville RL. (1997). Distinguishing action research from participative case studies. Journal of Systems and Information Technology 1 (1), 24-43. 72 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß Berman SJ. (2012). Digital transformation: opportunities to create new business models. Strategy & Leadership 40 (2), 16-24. DOI 10.1108/10878571211209314. Bharadwaj A, El Sawy OA, Pavlou PA & Venkatraman N. (2013). Digital Business Strategy: Toward a Next Generation of Insights. MIS Quartely 37 (2), 471-482. Bowman C & Ambrosini V. (2000). Value Creation Versus Value Capture: Towards a Coherent Definition of Value in Strategy. British Journal of Management 11 (1), 1-15. DOI 10.1111/1467-8551.00147. Brinkkemper S. (1996). Method Engineering: Engineering of Information Systems Development Methods and Tools. Information and Software Technology 38 (4), 275-280. Brown B & Sikes J. (2012). Minding your digital business. McKinsey on Business Technology. McKinsey & Company, New York, NY. Chiarini Tremblay M, Hevner AR & Berndt DJ. (2010). The Use of Focus Groups in Design Science Research. In: Hevner A, Chatterjee S (eds) Design Research in Information Systems, vol 22. Integrated Series in Information Systems. Springer, Berlin, pp 121-143. DOI 10.1007/978-1-4419-5653-8_10. Corallo A, Passiante G & Prencipe A (eds) (2007). The Digital Business Ecosystem. Edward Elgar Publishing, Glos, UK De Reuver M, Bouwman H & Haaker T. (2013). Business Model Roadmapping: A Practical Approach to Come from an Existing to a Desired Business Model. International Journal of Innovation Management 17 (1), DOI 10.1142/S1363919613400069. De Vos H & Haaker T. (2008). The STOF Method. In: Bouwman H, De Vos H, Haaker T (eds) Mobile Service Innovation and Business Models. Springer, Berlin. DOI 10.1007/978-3-540-79238-3_5. Dierickx I & Cool K. (1989). Asset Stock Accumulation and Sustainability of Competitive Advantage. Management Science 35 (12), 1504-1511. DOI 10.1287/mnsc.35.12.1504. Eichentopf T, Kleinaltenkamp M & van Stiphout J. (2011). Modelling customer process activities in interactive value creation. Journal of Service Management 22 (5), 650-663. DOI 10.1108/09564231111174997. El Sawy OA & Pereira F. (2013). Business Modelling in the Dynamic Digital Space: An Ecosystem Approach. Springer, Heidelberg. Ernst & Young. (2011). The digitisation of everything: How organisations must adapt to changing consumer behaviour. Ernst & Young LLP, London. European Commission. (2014). Towards a thriving data-driven economy. European Commission, Brussels. Gassmann O, Frankenberger K & Csik M. (2013). Geschäftsmodelle entwickeln: 55 innovative Konzepte mit dem St. Galler Business Model Navigator. Hanser, Munich. Gordijn J, Akkermans H & van Vliet H. (2000). What's in an Electronic Business Model? In: 12th International Conf. on Knowledge Engineering and Knowledge Management (EKAW 2000), October 2nd, 2000 (257-273). Juan-les-Prins. Gregor S. (2006). The Nature of Theory in Information Systems. Management Information Systems Quarterly 30 (3), 611-642. Gutzwiller TA. (1994). Das CC RIM-Referenzmodell für den Entwurf von betrieblichen, transaktionsorientierten Informationssystemen (Germ.: The CC 73 Digital Business Engineering RIM Reference Model for the Design of Transaction-oriented Enterprise Systems). Physica, Heidelberg, Germany. Hedman J & Kalling T. (2003). The business model concept: theoretical underpinnings and empirical illustrations. European Journal of Information Systems 12 (1), 49- 59. DOI 10.1057/palgrave.ejis.3000446. Heikkilä M & Heikkilä J. (2013). Collaborative Business Model Innovation Process for Networked Services. In: Järveläinen J, Li H, Tuikka A-M, Kuusela T (eds) Co- created Effective, Agile, and Trusted eServices. Lecture Notes in Business Information Processing. Springer, Berlin, pp 133-147. DOI 10.1007/978-3-642- 39808-7_12. Helfat CE & Peteraf MA. (2003). The Dynamic Resource-Based View: Capability Lifecycles. Strategic Management Journal 24 (10), 997-1010. DOI 10.1002/smj.332. Hevner AR, March ST, Park J & Ram S. (2004). Design Science in Information Systems Research. MIS Quarterly 28 (1), 75-105. Heym M. (1993). Methoden-Engineering: Spezifikation und Integration von Entwicklungsmethoden für Informationssysteme (Germ.: Method engineering: Specification and integration of design methods for information systems). Rosch-Buch, Hallstadt, Germany. Horne NW. (1995). Information as an Asset: The Board Agenda. Computer Audit Update (9), 5-11. Krishnan R, Smith MD, Tang Z & Telang R. (2007). Digital Business Models for Peer- to-Peer Networks: Analysis and Economic Issue. Review of Network Economics 6 (2), 194-213. Leimeister JM, Österle H & Alter S. (2014a). Digital services for consumers. Electronic Markets 24 (4), 255-258. DOI 10.1007/s12525-014-0174-6. Leimeister JM, Winter R, Brenner W & Jung R. (2014b). Research Program "Digital Business & Transformation IWI-HSG". University of St. Gallen, Institute of Information Management, St. Gallen. Leitner P & Grechenig T. (2008). Consumer Centric Communities: Integrating Community Based Features into Online Shops. In: IADIS International Conference on Web Based Communities 2008, July 24-26, 2008 Amsterdam. MacInnes I & Hwang J. (2003). Business Models for Peer to Peer Initiatives. In: 16th Bled eCommerce Conference, June 9-11, 2003 (44-58). Bled, Slovenia. March ST & Smith GF. (1995). Design and natural science research on information technology. Decision Support Systems 15 (4), 251-266. Newman D. (2011). How to Plan, Participate and Prosper in the Data Economy. Gartner, Stamford, CT. Nuseibeh BA, Finkelstein, Anthony & Kramer J. (1996). Method Engineering for Multi-Perspective Software Development. Information and Software Technology 38 (4), 267-274. Oppenheim C, Stenson J & Wilson RMS. (2001). The attributes of information as an asset. New Library World 102 (11/12), 458-463. DOI 10.1108/03074800110696979. Österle H. (1996). Business Engineering: Transition to the Networked Enterprise. Electronic Markets 6 (2), 14-16. 74 Boris Otto, Rieke Bärenfänger, Sebastian Steinbuß Österle H & Otto B. (2010). Consortium Research: A Method for Researcher- Practitioner Collaboration in Design-Oriented IS Research. Business & Information Systems Engineering 2 (5), 283-293. Österle H & Winter R. (2003). Business Engineering. In: Österle H, Winter R (eds) Business Engineering, vol 2. Springer, Berlin, pp 3-18 Osterwalder A & Pigneur Y. An e-Business Model Ontology for Modeling e-Business. In, Bled, 18.06.2002 2002. 15th Bled Electronic Commerce Conference. Osterwalder A & Pigneur Y. (2010). Business Model Generation: A Handbook for Visionaries, Game Changers, and Challengers. Wiley, Hoboken, NJ. Otto B, Abraham R & Schlosser S. (2014). Toward a Taxonomy of the Data Resource in the Networked Industry. In: 7th International Scientific Symposium on Logistics (ISSL 2014), (382-420). Cologne. Otto B & Aier S. (2013). Business Models in the Data Economy: A Case Study from the Business Partner Data Domain. In: 11th International Conference on Wirtschaftsinformatik (WI2013), February 27th, 2013 (475-489). Leipzig. Otto B & Österle H. (2010). Relevance Through Consortium Research? A Case Study. In: 18th European Conference on Information Systems (ECIS 2010), 2010-06- 06 Pretoria. Peffers K, Tuunanen T, Rothenberger MA & Chatterjee S. (2007). A Design Science Research Methodology for Information Systems Research. Journal of Management Information Systems 24 (3), 45-77. Porter ME. (1985). Competitive Advantage: Creating and Sustaining Superior performance Free Press, New York, NY. Rajagopal P & Burnkrant RE. (2009). Consumer Evaluations of Hybrid Products. Journal of Consumer Research 36 (2), 232-241. Ross B. (2009). Ten tips to winning at consumer centricity: for retailers and manufacturers. Journal of Consumer Marketing 26 (6), 450-454. DOI 10.1108/07363760910988265. Schuster CP & Dufek DD. (2004). The Consumer...or Else! Consumer-Centric Business Paradigms. International Business Press, New York, NY. Sein MK, Henfridsson O, Purao S, Rossi M & Lindgren R. (2011). Action Design Research. MIS Quarterly 35 (1), 37-56. Smart Service Welt Working Group. (2014). Smart Service Welt: Recommendations for the Strategic Initiative Web-based Services for Businesses. Berlin. Stewart DW, Shamdasani PN & Rook DW. (2007). Focus Groups: Theory and Practice. 2. edn. Sage, Thousand Oaks, CA. Teece DJ, Pisano G & Shuen A. (1997). Dynamic Capabilities and Strategic Management. Strategic Management Journal 18 (7), 509-533. DOI 10.1002/(SICI)1097-0266(199708)18:7<509::AID-SMJ882>3.0.CO;2-Z. The MIT Center for Digital Business. (2015). Research. Massachusetts Institute of Technology, Sloan School of Management. http://ebusiness.mit.edu/research/index.html. Accessed February 1st, 2015. Timmers P. (1999). Business models for Electronic Markets. Electronic Markets 8 (2), 3-8. Wasserman S & Faust K. (1994). Social Network Analysis: Methods and Applications. Cambridge University Press, Cambridge. Winter R. (2008). Design Science Research in Europe. European Journal of Information Systems 17 (5), 470-475. 75 Digital Business Engineering Wirtz BW, Schilke O & Ullrich S. (2010). Strategic Development of Business Models - Implications of the Web 2.0 for Creating Value on the Internet. Long Range Planning 43, 272-290. Yin RK. (2014). Case study research: design and methods. 5th edn. Sage Publications, Thousand Oaks, CA. Yoo Y, Henfridsson O & Lyytinen K. (2010). Research Commentary—The New Organizing Logic of Digital Innovation: An Agenda for Information Systems Research. Information System Research 21 (4), 724-735 Zott C & Amit R. (2010). BusinessModel Design: An Activity System Perspective. Long Range Planning 43 (2-3), 216–226. DOI 10.1016/j.lrp.2009.07.004. Zott C, Amit R & Massa L. (2010). The Business Model: Theoretical Roots, Recent Developments, and Future Research. IESE Business School, University of Navarra, Barcelona. 76 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Two birds with one stone. An economically viable solution for linked open data platforms Riccardo Bonazzi University of Applied Sciences Western Switzerland (HES-SO), Switzerland Riccardo.bonazzi@hevs.ch Zhan Liu University of Applied Sciences Western Switzerland (HES-SO), Switzerland Zhan.liu@hevs.ch Abstract Linked open data has been described by scholars as the logic evolution and the main benefit of open data. Nonetheless, the cost of data integration and platform management cannot be simply covered by selling the data, which is freely available by definition. Moreover, existing classifications of business models for linked open data platforms are rather descriptive instead of being prescriptive, and they do not take into account the notion of economic sustainability. Hence, this paper extends the existing literature in order to understand how to define a value proposition and a revenue model to increase the adoption of linked open data. We have developed a simple typology and we have identified a new revenue model for a linked open data platform, which is currently being tested. Keywords: open data, linked open data, business model, revenue model, value proposition, open innovation 1 Introduction This paper describes the first phase of an on-going project, and it is addressed to managers and scholars seeking for new ways to assure economic resources to develop and maintain a linked open dataset. The notion of linked open data (hereinafter referred to as LOD) comes from two concepts: (1) open data and (2) linked data. Open data is data that can be freely used, re-used and redistributed by anyone - subject only, at most, to the requirement to attribute and sharealike (Open Knowledge Foundation, 2012). Linked data describes a method of publishing structured data, upon standard Web 77 Bonazzi and Liu (2015) technologies, in a way that can be read automatically by computers so that it can be interlinked and become more useful (Bizer, Heath, & Berners-Lee, 2009). Figure 1: Stakeholders of a linked open data platform Driven by the success of Linked Data (LD), LD related business models have been discussed in the literature. Hence, we refer to a set of different provider’s roles proposed by (Latif, Saeed, Hoefler, Stocker, & Wagner, 2009) to support the conceptualization of successful business cases: raw data provider, linked data provider and linked data application provider. Based on these roles, (Tammisto & Lindman, 2011) claim that the main benefits of open data related activities is LOD transformation, consulting, and the application development by using these data. Figure 1 represents four stakeholders of a LOD platform: (a) the user of the application, who sometimes is willing to pay for contextual information obtained by aggregated data; (b) the application developer, who looks for a large amount of consistent data to exploit, in order to increase the usefulness of the application; (c) the owner of open data, who hopes to increase the usage of the dataset; (d) the manager of the LOD platform, who has to offer a service that is outperforming traditional data services while finding new ways to cover the cost of data integration and platform management. Accordingly, we refer to the notion of business model, as defined by (Osterwalder & Pigneur, 2010), and we focus on two key elements: (1) the revenue model, which is the description of how a business monetizes its services and (2) the value proposition, which is a promise of value to be delivered to the customer. Therefore, our research question is: how to define a revenue model to increase the adoption of linked open data? The rest of the paper proceeds as it follows. Section 2 briefly illustrates the existing literature, which addresses our research question. Section 3 illustrates the methodology used to address the gap in the literature. Section 4 illustrates the theoretical model obtained. Section 5 briefly illustrates the evaluation procedure, which is currently on- going. Section 6 summarizes the key elements of the paper and illustrates further directions of investigation. 78 Two birds with one stone. An economically viable solution for linked open data platforms 2 Literature review In order to obtain a descriptive review, we followed the argumentative strategy suggested by (Rowe, 2014). Accordingly, we used the keywords "linked open data" "value proposition" "revenue model" on Google scholar and we selected academic articles that were available online on January 2015, that offered insights about revenue models for linked open data platforms, possibility under the shape of classification or typologies. We initially obtained nine papers: three were dismissed since out of topic, one was dismissed since it was not containing any sort of classification; one was dismissed since it was not an academic paper and one was dismissed since it was not available. Of the remaining three articles, one focuses on linked data, whereas the other two on open data. On the one hand, (Vafopoulos, 2011) proposes eleven distinct business model categories for linked data. On the other hand, (Lindman, Kinnari, & Rossi, 2014) induce from a set of case studies the open data value network structure and propose five business model for the data network profiles, whereas, (Zeleti, Ojo, & Curry, 2014) merge emerging value disciplines for open data businesses into five major categories: (1) freemium; (2) premium; (3) cost saving;(4) support primary business;(5) razor and blade. Table 1 compares the elements used by (Zeleti et al., 2014) with the elements of the other two classifications. In the following sections we intend to extend these models by introducing the notion of economic performance of the LOD platform. Table 1: Mapping among elements of the three classifications 79 Bonazzi and Liu (2015) 3 Methodology In this section, we describe how we use design science to obtain a theory under the shape of a typology. According to (Hevner, March, Park, & Ram, 2004), design science addresses wicked problems and seeks out usefulness, rather than truth. According to (Doty & Glick, 1994), typologies are conceptually derived, interrelated sets of ideal types that meet three criteria: (1) they contain explicitly de-fined constructs that can be quantified, (2) relationships among the constructs are articulated, and (3) predictions associated with the typology are testable and subject to disconfirmation. Constructed in this way, a typology can account for multiple causal relationships in a given setting, and it can reduce complexity to manageable levels both conceptually and methodologically. 4 Our typology In this section we present (1) the constructs of our typology, (2) the relationships among the constructs, and (3) the predictions associated with the typology. Figure 2: Money flows among stakeholders of the linked open data platform 4.1 Our first order constructs Figure 2 represents the money flows among the four stakeholders already introduced in figure 1. In this study we focus on recurring transactions. The profit of the LOD platform manager depends on three flows: (a) the money paid by the application developer, who uses the data, which can be nothing or any amount above zero. Therefore, Price = [Low; High]; (b) the money that the platform has to pay to the data owner, which can be nothing or any amount above zero. Therefore, Cost = [Low; High]; (c) the money paid by a third-party, who takes indirect advantage from the platform, which can be nothing or any amount above zero. Therefore, Support = [Low; High]. 80 Two birds with one stone. An economically viable solution for linked open data platforms In this article we do not take into consideration: (d) the money flows between the application user and the application provider and (e) the money flow, which goes between the data owner, if it is a public institution, and the application user, intended as citizen. 4.2 Relationships among our first order constructs Following what stated in the previous paragraph, we obtain the following relationship. Equation 1: Profit Platform Manager = Price Application developer – Cost Data Owner + Support Third-Party Moreover, we assume that the amount of money obtained by third-party is less than the amount of money obtained by application developers. The resulting typology and its associated predictions Table 2 illustrates the resulting set of ideal types, which we named by using the five categories of (Zeleti et al., 2014). Nonetheless, since we obtained eight ideal types, we had to split some categories into two sub-components. The first ideal type offers community services (Vafopoulos, 2011), which are not meant to be profitable. Dbpedia.org is an example of service offering Wikipedia as LOD. The second ideal type refers to public services (Vafopoulos, 2011), which are supported by public institutions. ItoWorld.com offers public LOD. The third ideal type is used to increase traffic towards other services. Google public data explorer is meant to increases the overall traffic. The fourth ideal type is used to promote data owners. Musicbrainz.org offers linked music data to promote artists and it is supported by Google, which uses the dataset to improve its search results. The fifth ideal type refers to a LOD platform offering paid services beside its free datasets. According to their website, Mapbox.com offers custom online maps for major websites such as Foursquare, Pinterest, Evernote and the Financial Times. The sixth ideal type does not appear as such in the existing literature, and it will be described in the next section. The seventh ideal type refers to a LOD platform offering high quality data at a price, which is given back to the data owner (minus a transaction fee). The Azure data market allows Microsoft to obtain some money while increasing usage of the Azure platform. The eighth ideal type offers paid dataset in exchange for money from users for complementary services, while receiving supporting money from the provider of the complementary services. The recent acquisition of Datamarket by Olik (Park, 2014) can be seen as an example of this type of platform. 81 Bonazzi and Liu (2015) Table 2: Our typology 82 Two birds with one stone. An economically viable solution for linked open data platforms 5 Evaluation According to (Snow & Ketchen, 2014) most typologies fail to be assessed by using the five guidelines offered by (Doty & Glick, 1994). Therefore, we explain in details how we have addressed each guideline. 5.1 Typological theorists should make explicit their grand theoretical assertion(s). We refer to (Karahanna, Straub, & Chervany, 1999), who use the theory of diffusion of innovation to show that the perception of usefulness increases the chances of acquisition and retention of new users. Indeed, (Zeleti et al., 2014) have claimed that the freemium and the premium value propositions, which increase usefulness. Therefore, we claim that: Proposition 1: Over time, the diffusion of the ideal types premium and freemium will be greater than the other ideal types. 5.2 Typologies must define completely the set of ideal types. We have defined the full set of ideal types, and we have discussed the soundness of each result obtained. For one ideal type (the sixth) we did not find a correspondence in the existing literature. Indeed, it describes a service that relies on paid services based on freely available data, while obtaining sponsorship from third-party. Therefore, we named it 2b1s (“two birds with one stone”), and we speculate that it could refer to a LOD platform that offers high quality data at application owners, while selling to third- party the usage statistics of its datasets. Indeed, one could expect that, data owners belonging to public institutions would be willing to know how to fine tune their datasets to increase usage. 5.3 Typologies must provide complete descriptions of each ideal type using the same set of dimensions. We have presented our ideal types and we have done two actions: (a) we gave an example for those that are currently implemented and (b) we suggested a business case for those that are theoretically sound. 5.4 Typological theories should explicitly state the assumptions about the theoretical importance of each construct used to describe the ideal types. We have derived three first order constructs by extending the roles of (Latif et al., 2009). For sake of simplicity we have simplified our set of first order constructs in order to obtain the lowest set of ideal types that answers our research question. 83 Bonazzi and Liu (2015) 5.5 Typological theories must be tested with conceptual and analytical models that are consistent with the theory. The testing of our typology is currently on-going. We are collecting experts’ opinions to validate our theoretical model and we have been collecting second-hand data about LD, OD and LOD platforms to falsify our testable proposition. In parallel, we have partnered with a public institution and we have developed a LOD platform, which will follow the guidelines of the sixth ideal types (2b1s), to verify if it is feasible in practice. 6 Conclusions The purpose of this paper was to extend the existing literature to understand how to define a revenue model to increase the adoption of linked open data. We have developed a simple typology and we have identified a new revenue model for linked open data, which is currently being tested. We recognize that the major limitation of our paper is the lack of first-hand data. Nonetheless, we believe that this study already offers a major contribution in the field of business model for LOD by combining existing classifications for open data and linked data into a new prescriptive model that introduces the notion of business performance in the equation. Acknowledgements : This project was supported by the HES-SO Valais-Wallis under grant number 40160 (OverLOD Surfer). 84 Two birds with one stone. An economically viable solution for linked open data platforms 7 References Bizer, C., Heath, T., & Berners-Lee, T. (2009). Linked data-the story so far. Doty, D. H., & Glick, W. H. (1994). Typologies as a unique form of theory building: Toward improved understanding and modeling. Academy of Management Review, 19(2), 230–251. Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design Science in Information Systems Research. MIS Quarterly, 28(1), 75–105. Karahanna, E., Straub, D. W., & Chervany, N. L. (1999). Information Technology Adoption Across Time: A Cross-Sectional Comparison of Pre-Adoption and Post- Adoption Beliefs. MIS Quarterly, 23(2), 183. doi:10.2307/249751 Latif, A., Saeed, A. U., Hoefler, P., Stocker, A., & Wagner, C. (2009). The Linked Data Value Chain: A Lightweight Model for Business Engineers. In I-SEMANTICS (pp. 568– 575). Citeseer. Lindman, J., Kinnari, T., & Rossi, M. (2014). Industrial open data: Case studies of early open data entrepreneurs. In System Sciences (HICSS), 2014 47th Hawaii International Conference on (pp. 739–748). IEEE. Open Knowledge Foundation. (2012). Open Data Handbook Documentation. Release 1.0.0. Retrieved from http://opendatahandbook.org/ Osterwalder, A., & Pigneur, Y. (2010). Business Model Generation: A Handbook For Visionaries, Game Changers, And Challengers Author: Alexander Osterwalder, Yves." . Wiley. Park, H. (2014, November 4). Qlik Acquires DataMarket | Blue Hill Research. Retrieved from http://bluehillresearch.com/qlik-acquires-datamarket/ Rowe, F. (2014). What literature review is not: diversity, boundaries and recommendations. European Journal of Information Systems, 23(3), 241–255. Snow, C. C., & Ketchen, D. J. (2014). Typology-driven theorizing: A response to Delbridge and Fiss. Academy of Management Review, 39(2), 231–233. Tammisto, Y., & Lindman, J. (2011). Open data business models. In The 34th Information Systems Seminar in Scandinavia, Turku, Finland. Vafopoulos, M. (2011). A framework for linked data business models. In Informatics (PCI), 2011 15th Panhellenic Conference on (pp. 95–99). IEEE. Zeleti, F. A., Ojo, A., & Curry, E. (2014). Emerging business models for the open data industry: characterization and analysis. In Proceedings of the 15th Annual International Conference on Digital Government Research (pp. 215–226). ACM. 85 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia More than a gut feeling: Ensuring your inter-organizational business model works Cristina Chuva Costa Institute Polytechnic of Coimbra, ISEC CISUC, Department of Informatics Engineering, University of Coimbra, Portugal chuva@isec.pt Paulo Rupino da Cunha CISUC, Department of Informatics Engineering, University of Coimbra, Portugal rupino@dei.uc.pt Abstract We present an approach, called BIZ2BIS (from Business Models to the Blueprint of the Information System), to help design, discuss, and evaluate inter-organizational business models without a central point of authority, and also derive high-level requirements for their underlying IS. Its iterative and incremental nature enables the identification of attractive value propositions for all participants, thus ensuring a resilient value network. We have used three case studies to craft the first draft of our approach, accounting for principles, ideas, and concepts from the business model field. We then used action research to refine it, while simultaneously assisting a consortia tasked with setting up an inter-organizational business model and supporting IS for a wine producing region. The varied viewpoints provided by BIZ2BIS and its systematic nature enable the analysts to cope with the complexity of modern networked business models and their IS implications in an integrated manner. Keywords: Business models, information systems, high-level requirements. 1 Introduction Information and communication technologies have been gradually changing the playing field for organizations. The unprecedented ubiquitous connectivity achieved at negligible costs reduced coordination and transaction costs among firms (Heck & Vervest 2007). The balance between external and internal transaction costs in firms (Coase 1937; Coase 1960) changed dramatically, leading to the emergence of new organizational structures. It became cheaper and more convenient to procure, contract, and coordinate the services of globally distributed partners than to integrate all the needed functions in-house. In this context, in which the Internet is used as a business 86 Cristina Chuva Costa, Paulo Rupino da Cunha platform, firms are more properly viewed as participants in multiple networks (Gulati, Nohria & Zaheer 2000). Organizations were given the chance to open their boundaries and define innovative processes with different business rules and original value propositions. Existing approaches to business model design and evaluation usually neglect this complexity of partner networks. They do not manage the contributions and returns of the participants to ensure that all end up with attractive value propositions, that, in turn, ensure the collective satisfaction of the network (Iansiti & Levin 2004). Vague business ideas can hide inconsistencies and lead to false assumptions. This lack of information and imprecision can compromise the elicitation of business constraints that ultimately should be met by the IS supporting the business model (Gordjin 2002). Even though it is recognized that there are clear advantages in tracking these mutual influences (Chan & Reich 2007), the business model implications for the design of its supporting IS is underrepresented in the literature (Bouwman et al. 2012). To answer this, we developed BIZ2BIS, an approach to help design, discuss, and evaluate inter-organizational business models without a central point of authority and also derive high-level requirements for their underlying IS. In the remainder of the paper we detail BIZ2BIS as follows: in section 2, we introduce the work related with our proposal. Then, in section 3, we describe the four phases of BIZ2BIS, illustrate its steps and supporting artifacts. Finally, in section 4, we present the conclusions and discuss future research. 2 Related work To develop BIZ2BIS, we started by reviewing the literature on business models. To make sense of the field, we used the framework proposed by Pateli and Giaglis (2004) with its seven-subdomains (definitions, components, taxonomies, representation models, evaluation, adoption factors, and change methodologies). By reviewing definitions and components, we became aware that concepts like value propositions, partnerships, business architecture, financial issues, performed activities, or available resources stood out. Most of these concepts were also present in the proposals capable of visually representing business models. In spite of the noted limitations (the proposals graphical notations were not much elaborated), we detected an additional effort in detailing the activities performed by the actors and the resulting business flows (e.g., financial, information, goods and services, and intangibles). In turn, the subdomain taxonomies showed us a tendency in perceiving different categories of business models (e.g., freemium, razor/blades, or reverse auction) as building blocks that can be combined in multiple ways, as jump-starts of a creative business model process of discussion (Osterwalder & Pigneur 2010; Johnson 2010). When researchers like Osterwalder (2004), Shafer et al. (2005) and Morris et al. (2005) started to synthesize the research mentioned above, they created a base of knowledge that promoted the development of conceptual tools. The Business Model Canvas (Osterwalder & Pigneur 2010), the e3-value ontology (Gordijn 2002), and the STOF framework (Bouwman et al., 2005a) are unavoidable references in the field. They showed us the importance in defining an outlined plan for the application of BIZ2BIS to 87 More than a gut feeling: Ensuring your inter-organizational business model works … offer guarantees that its users do not overlook critical issues. The former two inspired us to use our approach as an effective communication tool to discuss business models and promote collaborations with all the participants. The latter two underlined the importance of examining the network of relationships in which firms are implanted and the role of the IS. The e3-value ontology establishes a link between value propositions and their underlying business processes, while the STOF framework details how the service offering can be carried out from a technical perspective. However, despite the previous efforts, there is still a gap between business model design and the specification of the IS to enable and support it. Furthermore, that gap needs to be bridged in a manner understandable by all business and technical stakeholders. When exploring the remaining sub-domains of the Pateli and Giaglis framework (2004) (evaluation, adoption factors, and change methodologies), we became aware of additional lines of research. There were no indications on how to address social factors in the development, adoption, or modification of real-world business models, which can also compromise the identification of the features that the supporting IS should satisfy. Furthermore, the existing proposals to evaluate business models were mainly focused on financial flows (Linder & Cantrell 2000; Gordijn 2002). However, we share Allee’s (2008) conviction that the benefits obtained through intangible forms of value can be vital in disclosing motivations for partners to engage with a network. Gathering this extra knowledge on intangible flows enhances network comprehension and provides valuable hints to design, discuss, and evaluate the network. We also acknowledged the difficulty in evaluating all kinds of value propositions using an economic unit of measure. For instance, the financial value assigned to a product or service is very volatile; what was established when conceiving the business model may not be valid after a month. Furthermore, something that can be extremely valuable for an actor, may not appeal to another. Table 1 summarizes the various contributions from the literature that had a role in shaping our approach. Number Author Influence in the development of BIZ2BIS Guideline 1 Timmers (1998), Al- Address dimensions of the business model concept such as value Debei and Avison proposition, value architecture, value network, and value finance (2010) Guideline 2 Osterwalder (2004), Take into account business model components like value proposition, Shafer et al. (2005) technology, revenue model, customers, distribution channel, and partners Guideline 3 Gordijn (2002), Define an outlined plan for using the approach in the field, in order to ensure Osterwalder (2004), that critical issues are not overlooked Bouwman et al. (2008) Guideline 4 Gordijn (2002), Use the approach as a communication tool to reflect on, discuss, innovate, Osterwalder (2004) and articulate a business model Guideline 5 Shafer et al. (2005), Address the potential offered by the network concept in the business model Gordijn et al. (2009) domain Guideline 6 Allee (2008) Detail the kind of ties established among the network participants to elicit clues on how these could strengthen the business model or obstruct undesirable movements 88 Cristina Chuva Costa, Paulo Rupino da Cunha Number Author Influence in the development of BIZ2BIS Guideline 7 Osterwalder and Develop easy-to-use field tools that promote collaboration among all the Pigneur (2010) stakeholders Guideline 8 Gulati et al. (2000) Identify vital dependencies in a the networked business model (e.g., important resources, indispensable actors, and critical value propositions) Guideline 9 Normann and Ramírez Develop negotiation mechanisms to promote eventual adjustments to new (1993), Iansiti and circumstances and balance the network pursuit for joint value creation Levin (2004) Guideline 10 Gordijn (2002), Acknowledge the need to change, to reconsider adopted options, revisit past Bouwman et al. (2012) assumptions, and rebuild taking into account new contexts Guideline 11 Gordijn et al. (2009), Make use of alternative business model scenarios to encourage discussion Bouwman et al. (2012) and explore new opportunities Guideline 12 Pateli and Giaglis Address social factors in the discussion, design, adoption, and change of (2004), business models Guideline 13 Tapscott et al. (2000), Consider other influences beyond financial flows in the business model Allee (2008) evaluation (e.g., prestige and brand recognition) Guideline 14 Gordijn (2002), Explore connections points between business models and their technological Bouwman et al. (2012) support Guideline 15 Gordijn (2002) Translate business models into high-level requirements for the specification of its underlying IS Table 1: Guidelines for the development of BIZ2BIS Supported by the literature review, we used the first draft of BIZ2BIS and two of its updated versions to analyze our three case studies. They enabled us to test and improve BIZ2BIS (e.g., concepts, phases, steps, and artifacts), as well as to detect and weed out any glaring omissions or misfits, before moving on to our last case, a complex action research project (2 Million Euros). We used BIZ2BIS to describe the scenario under study, diagnose problems, support negotiations, conceive interventions, readjust the business model, evaluate value propositions, and reflect on the obtained findings (for researchers and practitioners). Next, we present the resulting version of our approach. 3 BIZ2BIS: Business model and IS design BIZ2BIS consists of four phases. In Phase I – “Business model characterization”, we characterize the network, by identifying its actors and detailing their relationships. Then, in Phase II – “Business model refinement”, we analyze the network and suggest eventual adjustments to better align the interests of the actors. In Phase III – “Stability assessment”, we assess the business model stability by systematically verifying if the value propositions in the business model bring benefits to all the actors. In Phase IV – “Information system specification”, we use the gathered information about the network and its actors, as well as the arrangements established to align their interests, to detail the high-level requirements of the IS underlying the business model in a service- oriented fashion. The approach is flexible enough to interrupt, at any moment, the sequential order of its phases and return to previous ones in order to answer to unexpected network events or 89 More than a gut feeling: Ensuring your inter-organizational business model works … to indications ascertained when applying its steps. For instance, if a new actor is identified, independently of the phase in use, it is mandatory to return to Phase I. This flexibility enables BIZ2BIS to account for the dynamic nature of the networks, as suggested in the literature, Guideline 9 (Table 1). The importance of defining an outlined plan for using the approach was inspired by the Guideline 3 and 9 (Table 1). 3.1 Phase I - Business model characterization Phase I analyzes the business model by looking at its network. It comprises the identification and characterization of the participating actors, as well as their relationships. It consists in three steps with complementary perspectives:  Step I.a – “Exploration of the business model”: allows analysts to specify the aims of the networked business model, who contributes to its success, and how, as well as contextual influences that guide the performed activities. It is supported by the “Networked business model chart” (Table 2) and was inspired by the literature, namely guidelines 1, 2, 4, 5, 8, and 12 (detailed in Table 1).  Step I.b – “Description of the participating actors”: identifies actors and describes their roles, relationships, as well as expectations through the “Actor description chart” (Table 3), which should be filled for each actor. Guidelines 6 and 8 (Table 1) inspired this step.  Step I.c – “Representation of the business model”: represents the business model using two different tools: the “Flow Diagram” and the “Flow Matrix”. The former depicts the business model using a graph notation, in which the nodes represent the actors and the arrows the direction of the business model flows. These are categorized in four types: material or service, finance, information, and intangible connection (e.g., reputation, influence, and cooperation). To avoid the need to follow intricate configurations of arrows, the latter tool shows the same data in a matrix. The “Flow Matrix” should be read as indicated by the red arrow, starting with the “actor-source” (lines) and moving upward to the “actor-target” (columns). Step I.c was inspired by guidelines 6 and 8 (Table 1). Business model scenario The name assigned to the business model Network goals Gathers all the data obtained, analyzes it, and presents a first draft of the network’s goals Network opportunities Describes advantageous circumstances that can arise if the network is created Network threats Identifies possible threats to the network creation or maintenance Mutual obligations and expectations Describes established commitments and provides indications about the degree of cooperation in the network Shared interpretations and representations Identifies common codes, languages, and narratives that guide actors behavior Existing rules Describe policies that the actors must adhere to Available resources/actors Identifies the existing resources and the actors who provide them 90 Cristina Chuva Costa, Paulo Rupino da Cunha Business model scenario The name assigned to the business model Institutional sanctions Describes actions that must be carried out if the actors do not follow an acceptable behavior Version: 0.3 Date: Author: First name Surname DD/MM/YYYYY Table 2: “Networked business model chart” Description of the actor Identification of the actor Network interactions Depicts the interactions of the actor in the network Relationships and flows Details the business flows, e.g., information, associated with each interaction of the actor Roles Describes the activities carried out by the actor Goals Identifies the individual interests of the actor Business model: Name Version and Date: 0.3, Author: First and Surname DD/MM/YYYY Table 3: “Actor description chart” Table 2 provides a succinct view of the main guidelines established for the business model by its proponents. Table 3 provides clues on the business model participants. In turn, Figure 1 exemplifies a “Flow diagram” of the conceived business model on the left side and its corresponding “Flow matrix” on the right side (based on the data obtained in Table 2 and Table 3). Actor 2 Actor 1 web application Actor 3 Actor 1.1 Actor 1.2 mobile application Legend: Network actor Group of actors with financial flow information flow common features intangible flow material or service flow Name Name Business model scenario: Name Artefact: Flow diagram Version/Date: 0.2, DD/MM/YYYY Author: First name and surname Figure 1: “Flow Diagram” and “Flow Matrix" 91 More than a gut feeling: Ensuring your inter-organizational business model works … 3.2 Phase II – Business model refinement Having detailed what was planned for the networked business model, as well as the expectations of its participants in Phase I, Phase II addresses the need to perform refinements by providing a negotiation mechanism that looks for alignments among actors. This phase takes an optimistic view of the negotiation process, searching for win-win value propositions based on the assumption that the actors are engaged in a positive-sum activity in which they jointly create value. Five steps support Phase II:  Step II.a – “Detection of dependencies among goals”: - highlights how the goals of each actor contribute to reaching the aims of the overarching value network. It also exposes the dependencies among those goals and discloses how individual expectations interlock in a network of interactions that directly influences the ultimate business model objective. Step II.a uses the “Common goal diagram” (illustrated at the top of Figure 2) to support its analysis. In its upper part shows the network goal(s). Below, it depicts the goals of the actors in several lanes, one per actor, that support the overarching network goal. The bonds among actors’ goals are represented through arrows. For instance, “Goal 1 of the Actor 3” depends on “Goal 1 of the Actor 2” and on “Goal 1 of the Actor 1”. It is also possible to show that “Goal 3 of the Actor 3” is extremely dependent from the user’s goals. These insights enable the exploration of appealing synergies (e.g., possible cooperation), or risky situations (e.g., implications of actor abandonments) that can support or jeopardize the accomplishment of the business model goals. Step II.a was inspired by guidelines 4, 6, and 8 (Table 1).  Step II.b – “Identification of actor affinities”: supports the identification of goals common to various actors and promotes collaborations to minimize individual effort. It uses the “Actors/Goals affinity chart” (the middle chart in Figure 2) that maps the actors (first column) to their identified goals (first row) based on the data collected about the actors in Phase I. If a certain actor intends to accomplish a given goal, an “X” is placed at their intersection. This step points out common goals, which provides clues in order to strengthen the collaborations or minimize conflicts/problems identified in Step II.a. Guideline 6 (Table 1) inspired this step.  Step II.c – “Negotiation of actor contributions”: balances gains and efforts of the actors involved in the goal to clarify their interests. Step II.c uses the “Negotiation diagram” (the middle figure in Figure 2) to illustrate the analysis of the “Goal 1 of the Actor 3”, identified in Step II.a as critical. The diagram places it and the actor(s) that own(s) it at the center of the diagram. Below are the actors that carry out the supporting activities that sustain the central goal achievement (based on the data collected in Phase I). We rated the effort spent, as well as the gain obtained on these activities in a scale from . The gain obtained has two beneficiaries: the actor(s) that own(s) the goal under study (at the center of the diagram) and the set of actors that directly benefit from its achievement (the ones at the top). 92 Cristina Chuva Costa, Paulo Rupino da Cunha Step II.a – Common Goal Diagram Step II.b – Actors/Goals Affinity chart List of goals expressed by the actors in Phase I Gain Gain Goal 1 that benefits from the Goal 2 that benefits from the 5 5 goal under analysis goal under analysis Act or 3 Act or 3 3 Act or 4 Step II.c – Negotiation diagram Actor 1 (responsible by the goal under analysis) Goal under analysis Goal identified in Step II.a as critical by the actors Effort Effort Activity 1 that supports the Activity 2 that supports the 5 5 goal under analysis Actor 3 goal under analysis Act or 1 3 Actor 4 Legend: Gain Effort Result Actor 1 6 5 +1 Ef fort Gain Actor with the goal Effort spent by an actor Gain that a goal brings to under analysis Name Activity [1..5] to perform an activity Goal [1..5] Actor 3 10 5 +2 an actor Goal Actor Actor Actor 4 3 3 0 Identify goals influenced positively Gai Gain that the actor with the goal n by the goal under analysis under analysis obtains Business model scenario: Name Artefact: Negotiation diagram Version/Date: 0.3, DD/MM/YYYY Author: First name and surname Value proposition 1 Step II.d – Description of critical dependencies Value proposition 2 Actor 1 The cause that can compromise the flow Actor 2 Domino effect caused by the Legend: absence of a critical flow Network actor Group of actors with Dependency Existing depencies common features X Dependency origin Name Name business flow Direction of the dependency Value Dependent value proposition proposition Business model scenario: Name Artefact: Dependence flow diagram Version/Date: 0.2, DD/MM/YYYY Author: First name and surname Figure 2: Artifacts of Step II.a, Step II.b, Step II.c, and Step II.d 93 More than a gut feeling: Ensuring your inter-organizational business model works … The results achieved for each goal in Step II.c cannot be analyzed from a narrow perspective. For instance, a goal may not be appealing for a particular actor, but the business model may offer other advantages that can make it worthwhile. When a positive balance is not reached, analysts should initiate a negotiation process and consider adjustments to the conceived business model in order to stimulate and encourage actors’ participation. Analysts should base their attempts on the data gathered in the previous steps of BIZ2BIS. Step II.c was inspired by Guidelines 9, 10, and 11 (Table 1).  Step II.d – “Description of critical dependencies”: discloses domino effects caused by the extinction of a particular business flow. For instance, if an actor leaves the network, his/her activities will not be performed and the resulting flows will be compromised, which will consequently jeopardize value propositions that depend upon those flows, as well as the aspirations of participating actors interested on those value propositions. The “Dependency flow diagram” (on the bottom part of Figure 2) depicts these dependencies, which can help analysts mitigate possible threats. When indications of events that may jeopardize the business model no longer exist, analysts should advance to Step II.e. Step II.d was inspired by Guideline 8 (Table 1).  Step II.e – “Stabilization of value propositions”: uses the data gathered in the previous steps of BIZ2BIS to list the existing business flows. Then, based on the contribution of the flows to the activities performed by the actors, analysts should refine and stabilize the list of value propositions provided by the business model. The analysis of the existing relationships is supported by the “Business flows/Value propositions chart” (Table 4), which maps all the flows (first column) into all the derived value propositions (first row). If a certain flow gives rise to, contributes to, or influences, a given value proposition, that situation is marked with an “X” at the intersection. ws ws ws ws lo lo lo lo f f f f ss ss ss ss ine ine ine ine us us us us b b b b pshi he he he he t t t t by by by by ions lat Business flow/Value proposition chart rted rted rted rted re o o o o pp pp pp pp r of su su su su be n n n n tio tio tio tio Num si si si si po po po po ro ro ro ro p p p p ue ue ue ue Val Val Val Val V1 V2 V3 V5 Business flow (material or service, finance, information, or intangible connection) obtained from Phase I, Step I.c F1 X X X X 4 Business flow (material or service, finance, information, or intangible connection) obtained from Phase I, Step I.c F2 X X X X 4 Business flow (material or service, finance, information, or intangible connection) obtained from Phase I, Step I.c F3 X 1 Business flow (material or service, finance, information, or intangible connection) obtained from Phase I, Step I.c F4 X 1 Number of relationships 3 2 3 2 Version and Date: 0.3, Author: First Business model: Name DD/MM/YYYY name and Table 4: “Business flows/Value propositions chart” 94 Cristina Chuva Costa, Paulo Rupino da Cunha 3.3 Phase III – Stability assessment Phase III, Step III.a – “Evaluation of actors perspective” assesses the idealized business model based on the value propositions obtained in Phase II. The performed evaluation integrates two perspectives (inspired by Guideline 13, Table 1). One shows the actors’ perception of the effort spent to support the value propositions, as well as the gain obtained. The other discloses how the actors perceive influences among value propositions and may expose dependencies not yet detected. The “Interview chart” (at the top of Figure 3) supports the evaluation performed by the actors. It maps each one (second column) with the identified value propositions (first column). We represent the relationship among the two by pairs of integer numbers (g, e), where “g” represents the gain obtained with a value proposition in the range {1,...,5}, and “e” denotes the effort involved in the range {-1,...,-5}. The influences that a specific value proposition has on others is available in its own row, after the pair (g, e), and separated by a “/”, such as (g, e)/ , where “+|-” further informs whether that same value proposition has a positive (“+”) or negative (“-”) impact towards the value proposition “Vi”. The superscript “+” denotes iteration, since a value proposition may influence none, one, or more value propositions. Figure 3 exemplifies the “Interview chart”. For example, it shows that “Actor 1” assigns an effort of “5” to support V3 and assigns an importance of “1” to the benefits obtained from it. Taking into account all the value propositions, the balance between gain and effort shows if the actor is pleased or, on the contrary, disappointed. In this case, extra benefits must be considered to maintain the actor in the network. To relate the different concepts used in BIZ2BIS, we developed the “Value proposition traceability diagram” (at the bottom of Figure 3). Its compact representation traces the business flows (the dashed rectangle), their supporting activities (the rectangles), the actors that perform them, as well as dependencies among value propositions based on the data filled in by the actors in the “Interview chart”. In this example, it shows the influence of V3 on V1 and V5 (the relationship between the rounded rectangles): the plus sign means that one value proposition influences other(s) positively, while the minus implies a negative influence. By exposing the influences among the different value propositions, we have the chance to identify critical actors and value propositions, anticipating potential problems. 95 More than a gut feeling: Ensuring your inter-organizational business model works … Actors Value propositions Actor 1 Value proposition 1 V1 (+5,-1) Value proposition 2 V2 Value proposition 3 V3 (+1,-5) / +V1,+V5 Value proposition 4 V4 (-2) / +V1,+V5 Value proposition 5 V5 (+3) / +V1, +V6 Date: DD/MM/YYYY Sum (+9,-8) Actor 3 Actor 2 Activities performed by Activities performed by Actor 3 Actor 2  Business model flow 1  Business model flow 2 V1 V3 Actor 2 Actor 1 V5 Actor 3 Figure 3: “Interview Chart” and the “Value proposition traceability diagram” 3.4 Phase IV- Information system specification When an agreement is achieved, analysts should advance to Phase IV, Step IV.a – “Consolidation and description of requirements” (it was inspired by Guidelines 14 and 15, Table 1). Step IV.a establishes a bridge between business models and their supporting IS by using the data obtained in the first three phases of BIZ2BIS to identify and detail the features to be provided. To enable this translation of knowledge, we used the concept of service (Marks & Bell 2006), which establishes a point of contact between what organizations provide to their customers or partners, and the functionalities delivered via the interface of an IS. As a result, we developed the “Service specification chart” to detail the services that must be provided to make the value propositions acknowledged by the available actors (in Phase II, Step II.e, “Business flows/ Value propositions chart”). Table 5 exemplifies this artifact. 96 Cristina Chuva Costa, Paulo Rupino da Cunha Business model Service Specification Name/Identifier Presents the service name and its identification number Id: 1 Version Identifies version, data, and author Goal Presents the aim of the service (data obtained from Phase II, Step II.e) Description Describes the activities performed when using it (based on Phase II, Step II.e) Actor that provides the Identifies the actor(s) that provide(s) it (data obtained from Phase I, Step I.b service and Step I.c) Actor that uses the service Identifies the actor(s) that use(s) it (based on Phase I, Step I.b and Step I.c) Input data and their source Depicts input information flows and their source (data obtained from Phase I, Step I.b and Step I.c) Output data and its target Describes output information flows and their target (data obtained from Phase I, Step I.b and Step I.c) Service dependencies Identifies supporting services (data obtained from Phase III, Step I.a) Access control Details permissions and access rights (data obtained from Phase I, Step I.a) mechanisms Business flows leading to Identifies the business flow(s) that contributed to the service detection (data the service obtained from Phase II, Step II.e and Phase I, Step I.c) Reasons for its existence Explains the motives behind the service creation (data obtained from Phase I, Step I.a and Step I.b) Service restrictions Presents the rules employed by the service in its activities (data obtained from Phase I, Step I.a and Step I.b) Information system Describes how the IS supports the service (based on Step I.a and Step I.b) support Remarks Additional data Table 5: “Service specification chart” Analysts and IT teams can easily perceive the actors that interact with the service, how they do it, the reasons for the service existence, the involved business flows, the activities related to the service, rules that govern its operation, and how the IS should made it available. At the end of Step IV.a, a service-oriented high-level specification of the supporting IS is available as the full set of “Service specification charts” - a blueprint of how a network of organizations creates and delivers value. 4 Conclusion BIZ2BIS guides the search towards stable networked business models. It gathers data on the network, its context, and its actors to clarify and expose their different opinions, preferences, and instincts. By providing a common language between analysts and practitioners, the approach encourages the collaboration of the latter and promotes communication and discussion among all the involved. Its insights support the actors with the power to make decisions to carry out adjustments in inter-organizational business models. Our proposal was inspired by the review of the field literature, which allowed us to identify key topics to take into account. However, we complemented the existing common ground with additional perspectives of analysis. First, we introduced a negotiation mechanism to promote the alignment of the actors’ interests. Second, we moved beyond the usual accounts of an organization business model and focused our attention on inter-organizational business models. Third, we involved the participating actors in the evaluation of the conceived business models and introduced more than 97 More than a gut feeling: Ensuring your inter-organizational business model works … economic units of measure (e.g., data and intangible flows). Fourth, taking into account the wealth of information collected by BIZ2BIS, we gathered promising conditions to overcome the gap between business models and the development of their IS. The concept of service helped us to establish a point of contact between the value propositions made available and the internal business processes supported by the IS, which allows BIZ2BIS to derive the high-level requirements of the underlying IS in a business model driven way. The increasing importance assigned to the co-creation of value propositions and service innovation, led us to our future research aim: use BIZ2BIS’ ability to characterize the network in order to design service-oriented architectures in a systematic way. References Al-Debei, M. M. & Avison, D. (2010), Developing a unified framework of the business model concept, European Journal of Information Systems, 19 (3), p. 359–376. Allee, V. (2008), Value network analysis and value conversion of tangible and intangible assets, Journal of Intellectual Capital, 9 (1), p. 5-24. Bouwman, H., Faber, E., Haaker, T., Kijl, B. & Reuver, M. D. (2008), Conceptualizing the STOF model, In Mobile Service Innovation and Business Models, ed. HdVTH H. Bouwman, Heidelberg: Springer, p. 31-70. Bouwman, H., Reuver, M. D., Solaimani, S., Daas, D., Haaker, T., Janssen, W., Iske, P. & Walenkamp, B. (2012), Business models tooling and a research agenda, In 25th Bled eConference, Bled, Slovenia. Chan, Y. & Reich, H. (2007), IT alignment: what have we learned?, Journal of information technology, 22 (4), p. 297 – 315. Coase, R. (1937), The Nature of the Firm, Economica, 4 (16), p. 386–405. Coase, R. (1960), The problem of social cost, Journal of Law and Economics3), p. 1-44. Gordijn, J. (2002), Value-based requirements engineering: exploring innovative e- commerce ideas, PhD Thesis, Free University Amsterdam. Gordijn, J., Kinderen, S. d., Pijpers, V. & Akkermans, H. (2009), e-services in a networked world: from semantics to pragmatics, In Future Internet – FIS 2008, eds J Domingue, D Fensel & P Traverso, Heidelberg: Springer Berlin, p. 44-57. Gulati, R., Nohria, N. & Zaheer, A. (2000), Strategic networks, Strategic Management Journal, 21 (3), p. 203-215. Heck, E. v. & Vervest, P. H. (2007), Smart business networks: how the network wins, Communications of the ACM, 50 (6), p. 28-37. Iansiti, M. & Levin, R. (2004), Strategy as ecology, Harvard Business Review, 82 (3), p. 68-78. Johnson, M. W. (2010), Seizing the white space: business model innovation for growth and renewal, Boston: Harvard Business Press. Linder, J. & Cantrell, S. (2000), Changing business models: surveying the landscape, Accenture Institute for Strategic Change. Marks, E. & Bell, M. (2006), Service-oriented architecture: a planning and implementation guide for business and technology, New Jersey: John Wiley & Sons. Morris, M., Schindehutteb, M. & Allenc, J. (2005), The entrepreneur's business model: toward a unified perspective Journal of Business Research, 58 (6), p. 726-735. 98 Cristina Chuva Costa, Paulo Rupino da Cunha Normann, R. & Ramírez, R. (1993), From value chain to value constellation: designing interactive strategy, Harvard Business Review, 71 (4), p. 65-77. Osterwalder, A. (2004), The business model ontology - a proposition in a design science approach, Universite de Lausanne. Osterwalder, A. & Pigneur, Y. (2010), Business model generation: a handbook for visionaries, game changers, and challengers, New York: John Wiley & Sons. Pateli, A. & Giaglis, G. (2004), A research framework for analyzing eBusiness models, European Journal of Information Sciences, 13 (4), p. 302-314. Shafer, S., Smith, H. & Linder, J. (2005), The power of business models, Business Horizons, 48 (3), p. 199-207. Tapscott, D., Ticoll, D. & Lowy, A. (2000), Digital capital: harnessing the power of business webs Boston: Harvard Business School Press. Timmers, P. (1998), Business models for electronic markets, Electronic Markets, 8 (2), p. 3-8. 99 BACK 28th Bled eConference #eWellBeing June 7 - 105, 2015; Bled, Slovenia The Bitcoin Phenomenon Analysis Boris Tomaš Faculty of Organization and Informatics, Croatia boris.tomas©foi.hr Ivan Švogor Faculty of Organization and Informatics, Croatia ivan.svogor@foi.hr Abstract Bitcoin, the latest Internet phenomenon branded as a game-changer. It is a revolution in financial system and money creation mechanics which currently holds the market of almost 13 billion USD. At the same time, its inner working is presented as hard to understand but perfectly trustworthy and safe. This trust is what gives Bitcoin value, however due to its mystical creator and recent headlines with negative connotation, what to really make of it? - In this paper we try to explain the Bitcoin and present the latest related research to provide a discussion ab out its potential impact on economy, financial world and society. Keywords: Bitcoin, economy, cryptocurrency, digital money, society 1 Introduction Very recently a new online phenomenon came to a focus in the form of digital currency. Among others, the best known one is Bitcoin. Since its appearance in 2009 it has been a source of controversy and myths, but recently, there has been a campaign to popularize it and brand it as a state of the art, mathematics-security game changer in a world of finance. In this paper we will explain the mysterious Bitcoin phenomenon circumventing oversimplified explanations of it, address the state of the art and infer some social implications of its applications. 2 Bitcoin - a crash course Although one can find a lot of papers about Bitcoin, in this section we will briefly present the most important facts about how its inner workings. 100 Ivan Švogor, Boris Tomaš 2.1 What is really a Bitcoin? Bitcoin refers to two things. First, Bitcoin is a currency like a dollar, euro, yen, etc. with a smaller called satoshi which represents 1 billionth of a single bitcoin value (10− 9). The second meaning, more accurate one, referees to Bitcoin as a name for the pseudo- anonymous, peer-to-peer currency protocol (Hobson, 2013). In reality, Bitcoin is a number created out of nothing, and its value is given by the Bitcoin protocol, or even more so by the trust of its users. It has been designed by unknown individual who calls himself Satoshi Nakamoto. In his paper (Nakamoto, 2008) Nakamoto described Bitcoin as a peer-to-peer electronic cash system. 2.2 Typical transaction and Bitcoin creation Figure 1: Typical Bitcoin transaction A typical Bitcoin transaction involves 3 parties: a) sender (in Figure 1 Alice), b) receiver (in Figure 1 Bob) and c) Bitcoin miners. Additionally there are two more; Bitcoin exchange and Bitcoin developers, however for this example they are irrelevant. In the Bitcoin network, both Alice and Bob have their addresses which represent their account numbers. Each address, as a bank account does, has a balance (in Bitcoins). Alice and Bob can have multiple addresses and instead of remembering them, they use a software called a Bitcoin wallet. It uses a public key cryptography, and for every address it generates an appropriate private key pair (Nechvatal, 1991). Let us consider a transaction depicted in Figure 1, where Alice sends 1 Bitcoin to Bob (1). To do so, Alice uses a Bitcoin client software to create a transaction. Using the private key from Alice’s Bitcoin wallet (2), Bitcoin client signs the transaction request (3). Using Alice’s public key, anyone on the Bitcoin network can verify its authenticity. Alice’s transaction, along with other transactions on the Bitcoin network are bundled together in something known as Bitcoin block (4). The transaction is now verified by Bitcoin miners (5). In order for a block to be recorded in the Bitcoin network, miners need to provide the proof of work (6). Once proof of work is confirmed, block is added to a 101 The Bitcoin Phenomenon Analysis block chain which is a list of all confirmed blocks in a Bitcoin network. Each block contains the address of a previous block, thus forming single chain. A process of adding a block (also known as: completing a block) to a block chain is called Bitcoin mining. In Bitcoin protocol proof of work is called nonce, i.e. a 32-bit number. When the block is hashed a nonce with a specific lead of zeros must be found and be confirmed by more than 50% of the network. Miner, who has found the nonce is rewarded with newly minted Bitcoins (currently 25BTC) (7). Finding such nonce is tremendously difficult for individual find so Bitcoin users join to mining groups called a mining pools. 3 Related work – state of the art In this section we present related work in the academic community regarding the Bitcoin. For this purpose we browsed the following bibliographic databases: Springer, Web of Science, Scopus, Science direct, IEEE Xplore and ACM Digital Library. The best papers we found for this overview were categorized in seven topics: • Introductory papers - reporting general remarks about Bitcoin • Analytical papers - investigating some aspect of Bitcoin in detail and presenting results • Anonymity papers - reporting on the issues, benefits and improvements of anonymity • Economic papers - investigating and reporting economic implications of Bitcoin and its market influence • Improvement suggestion papers - suggesting an improvement of some aspect of Bitcoin • Mining papers - reporting on technical challenges of mining • Bitcoin issues papers - presenting weaknesses and drawbacks of BitcoinAuthors’ Information 3.1 Introductory papers Most of the introductory papers about Bitcoins are popular articles in magazines or featured articles in scholarly journals. However, most of these papers present only a partial and simplified image of a Bitcoin. Also, depending the background of the author, focus is on different aspects of Bitcoin. In his paper (Hobson, 2013), Hobson explains the nature and proof of concept of Bitcoin virtual currency. While author pinpoints several weaknesses of Bitcoin, he presents the 51% problem as the most significant one. 51% problem refers to hash power owned by single entity. This amount of power cannot be matched with any today’s supercomputer, however it can be matched by mining pools. Actually, there are some pools like ghash.io which are very close in reaching that number (CEX.IO Ltd., 2014). Peck presented a Bitcoin as a “Crypto-anarchist’s answer to hash” (M. E. Peck, 2012). In non-bias way he presented Bitcoin as a flow of balances between peer-to-peer entities without a bank, credit card company and any other central authority. Peck also says that this is by no means a novel idea, and arguments that Timothy May and 102 Ivan Švogor, Boris Tomaš Cyberpunks had a similar idea in 1992. Author points out that restoring of online payment privacy is a good thing but wans that it can introduce new problems such as collecting donations for political decedents or money laundering. In order to get more insight to Bitcoin we suggest some credible sources such as: (Grinberg, Primer, Ecosystem, Sustainable, & Issues, 2012; Hobson, 2013; Nakamoto, 2008; Ron & Shamir, 2013). 3.2 Analytical papers In Nature Scientific Reports, Kristoufek published a great analysis of Bitcoin (Kristoufek, 2013). He presented absurd profits of people who invested in Bitcoin in beginning of 2013 when its value was 13 USD. By the end of 2013, price of a single Bitcoin went to 395 USD bringing the owners 2900% increase of investment. At the time of writing of this paper, the value is 450 USD, however, the largest was recorded in end of November of 2013 when it was 1126 USD (Coinbase, 2014). Author analyzed the phenomenon of Bitcoin and argues that such behavior cannot be explained by standard economic and financial theories, e.g. future cash-flows model, purchasing power parity, and uncovered interest rate party. Furthermore, he analyzed market entities and reports that the market is dominated by short-term investors, trend chasers, noise traders and speculators. Fundamentalist segment of the market is missing due to the fact that there is no basis for setting a fair price. It is solely driven by investors’ faith in the growth. Kristoufek used user search queries on Wikipedia and Google Trends to analyze the dynamic relationship between Bitcoin price and interest in currency measured by search queries. There is a causal bidirectional relationship between prices and search terms. If prices are high, the increase of interest pushes them further to top when they are low, growth of interest pushes them deeper. This follows the pattern of bubble behavior which has been observed with Bitcoin (Kristoufek, 2013). Kondor et. al. analyzed the Bitcoin by measuring network characteristics in function of time (Kondor, Pósfai, Csabai, & Vattay, 2014). They identified two distinct phases in the lifetime of the system: 1) when it was new and 2) when it received public attention. In first phase this was characterized by large fluctuations in its network characteristics, heterogeneous in-degree and homogeneous out-degree distribution, and in the second phase one can see stable network measures. Ron and Shamir published a paper with quantitative analysis of Bitcoin transactions (Ron & Shamir, 2013). They used transactions carried out since the beginning of Bitcoin, up to May 2012. In the Bitcoin transaction flow graph authors discovered that most Bitcoins remain dormant, this means that they are not used, but simply stored on a wallet for safe keeping. Most Bitcoins (55% of total volume) are not circulating in the system. 3.3 Anonymity papers Due to its anonymity properties Reid and Harrigan see a Bitcoin as an alternative to cash (Reid & Harrigan, 2011). According to them cash still has competitive advantage in anonymity of payment, however, Bitcoin seems like an alternative. Authors actually found that using the appropriate network representation, it is possible to associate public-keys with identifying information. Also, large centralized services such as exchange markets and online wallets are capable of identifying and tracking user activity. Bitcoin community says that anonymity is not a prominent design goal, however users should be aware when sending Bitcoins to organizations they would 103 The Bitcoin Phenomenon Analysis prefer not to be publicly associated with. Peck has been able to identify several key entities in the Bitcoin economy: markets, mining pools, stores, gaming sites, and many more (M. Peck, 2013). Some papers in this group are not directly related to Bitcoin, however they present an anonymity aspect. One of them analyses drug dealing site called Silk Road inside Deep Web and paying options with anonymous currency – Bitcoin (Van Hout & Bingham, 2014). Authors have shown that cash replacement for Bitcoin is actually being valid, valuable and reliable digital currency in the case of Silk Road. Since Bitcoin is actually pseudo-anonymous, Miers et.al. proposed a solution to this issue, with the Zerocoin, i.e. a Bitcoin extension (Miers, Garman, Green, & Rubin, 2013). However, accepting changes in existing environment could cause some issues, but since it has been done before without larger consequences, it is up to the community to decide. 3.4 Economic papers Due to its anonymity features Bitcoin is prone for money laundering. Bryans analyzed the Bitcoin network form an economic viewpoint focusing on legal issues (Bryans, 2014). He identified that a typical transaction involves 5 parties. 1) sender (e.g. dirty money), 2) receiver (launderer which obfuscates the source), 3) Bitcoin miners which act as transaction verifiers by completing a block, 4) development team which updates codebase, 5) currency exchange. Bryans asked the question, which party should carry the legal responsibility when a breakdown occurs (money laundering, theft, fraud, etc.), and therefore whom to regulate? It is hard to regulate a sender due to pseudo-anonymity and dispersed nature of identities on the Bitcoin network. The same applies to the receiver and miners. Perusing them would be inefficient and detrimental. Miners act as payment processors with no real interest other than small (and non-obligatory) fee, and in the most cases they are unaware of the nature of transaction. Also, mining is done by a software without any intervention of people. Bitcoin developer team has hardly any actual input on the individual transactions and they act more like a standards agency rather than central authority. Finally, we reach currency exchanges. Bryans concludes that because exchanges deal with fiat currency, they will more likely fall under money exchange laws which define money as currency backed by government. Regulation of such currencies should occur at point where law enforcement can most effectively punish civil and criminal violators with the least overhead. Since Bitcoin is decentralized it makes little sense to regulate others than Bitcoin currency exchanges. In fact, some exchanges showed interest by registering as MSBs under AML schemes. Author suggests that instead of predicting regulations for next generation of disruptive technology like Bitcoin, we need to understand current ones better, and police the points of public contact with existing legal schemes. Although the idea of digital currency exists for a while, they never cough on. Barber et.al. analyses the success of Bitcoin and reported that despite no fancy cryptography features, it is “ingenious and sophisticated” but not perfect (Barber, Boyen, Shi, & Uzun, 2012). There are several advantages of Bitcoin over its predecessors: a) Bitcoin ecosystem ensures that users have economic incentive to participate by mining the coins, b) coin generation has an exponential rate which enables predictable money supply, c) it is distributed which is appealing due to is ease of dividing, d) although it is a e-cash system, denominations are possible, e) it is open and flexible, f) transactions are irreversible, g) fees are low and h) implementations exist. But there are also 104 Ivan Švogor, Boris Tomaš imperfections such as: a) Zombie coins1, b) deflationary spiral, c) history-revision attack2, d) countering revisionism by checking the past. They conclude the following: ”While instantiation is impaired by its poor parameters, the core design could support a robust decentralized currency if done right” (Barber et al., 2012). Yermack reports that proper currency functions as medium of exchange, unit of account and sore of value. Bitcoin has first two functions but lack function to store value. This is mainly because Bitcoin volatile nature. Traditional currencies have institutional stability and security, deposit insurance and international treaties. Bitcoin lacks such instruments. If Bitcoins are widely used then controlled inflation will not be sufficient and effects of deflation may arise. This could lead to political protests like one happened in U.S during Populist movement at the end of 19th century. (Yermack, 2013) 3.5 Improvement suggestion papers Decker and Wattenhofer analyzed how information in the Bitcoin network is disseminated in order to synchronize the ledger replicates (Decker & Wattenhofer, 2013). They report that reliance on blocks delays clearing of transactions and poses a threat. An example of this are larger blocks which are propagated slower. This causes a Blockchain fork3. Authors implemented changes in Bitcoin protocol which reduces a risk of Blockchain fork by 50%. Babaioff et. al. studied a scenario in which all the nodes that become aware of the information compete for the same prize, have incentive not to propagate the information (Babaioff, Dobzinski, Oren, & Zohar, 2011). This is related to false identities where one would keep the information about any transaction that offers a fee for itself as other nodes compete to authorize it and claim the associated fee. This would be the problem when payment to the nodes is slowly phased out and Bitcoin owners who want their transactions approved would pay a fee to authorizing nodes. Authors propose a novel rewarding scheme in Bitcoin mining and propose a novel low cost reward scheme that incentivizes information propagation and is Sybil proof. Clark and Essex presented the problem with proof of work which Bitcoin network uses to generate a block. Since it is costly and time consuming it limits the rate at which new blocks can be generated. Authors argue that employing a proof of work protocol as commitment time will later allow anyone to ”carbon date” when the commitment was made, approximately without trusting any external party. Authors present commitcoin, an instantiation of this approach which harnesses the existing computing power of Bitcoin network to mint and trade digital cash. With their approach users do not need to trust the timestamps or any node in the network, and proof of work would itself be used to carbon date the transaction. (J Clark & Essex, 2012) 3.6 Mining papers Bitcoin mining is not the focus of many scientific publications. We suspect that this is because all the algorithms are well known, and from the computational viewpoint it is just the matter of making mining faster, i.e. engineering better chips (ASIC, FPGA, 1 Bitcoins which are lost cause reduction of available supply 2 When incentive to mine will diminish, then, the computers will become stronger and revision attack will be easier 3 A time when nodes in the network don’t agree on which block, in current Blockchain, is the head 105 The Bitcoin Phenomenon Analysis CPU, GPU, etc.). In his publication Taylor presented a review of advantages and disadvantages of Bitcoin mining hardware (Taylor, 2013). It is proven that Bitcoin, using existing hardware, on a small scale is inefficient. This is the basis of the protocol of value creation mechanism. Over time mining will become completely inefficient and will reside on transaction fee which is going to grow as Bitcoin transaction volume grows. Next big thing in mining technology would the application of quantum computers, which will do computing several orders of magnitude faster and efficient. Unfortunately, use of quantum computer will probably endanger core of Bitcoin currency, so its appearance should result in changes of Bitcoin protocols. 3.7 Bitcoin issues papers Bitcoin supporters often use mathematics to argument the validity of Bitcoin idea. It is said that mathematics has trust embedded into it and since the mathematical proofs which back up Bitcoin is irrefutable. It presents trust unlike one tied to present day banking system. In his report, Bradbury presents this idea of a solid currency however issues are also mentioned; stealing identities, double spending, dust transactions etc. (Bradbury, 2013). Here are several large incidents related to Bitcoins. Sheep Marketplace has been closed following the theft of millions of dollars’ worth of Bitcoins. Article reports it as one of biggest cybercrime heists in history. Also, there was another heist in Denmark BIPS using DDoS and a similar one in Australia (1m BTC - 1295), Inputs.io (1.2m BTC - 4100). In Germany, hackers used Botnets to mine 700.000 BTC, while in US E-Sports Entertainments reported that their anti-cheating software for online gaming has been used to mine. A hacker made added some mining code which was distributed among gamers, making him 4.000,00 USD (Computer Fraud & Security, December, 2013). Analyzing Bitcoin exchange markets volume and characteristics has proven that high volume markets are more often hacked and targeted for fraud (Moore & Christin, 2013). Moore and Christin investigated this issue and found that low level markets are irrelevant for malicious actions, however, it is proven that trading on high volume markets is safer because foul action will be easier to detect. Authors claim that high volume markets have lower probability of crashing. However, crash of MtGox in 2014 obviously proves them wrong (Moore & Christin, 2013). Finally, one issue with Bitcoin is the botnet exploit, which first occurred in 2010. Bitcoin mining feature was used using multiple Bitcoins addresses and mining pools, but due to Bitcoin pseudo- anonymous nature it is hard to reveal identity of botnet creators. It is notable that existence of botnet Bitcoin miners is not the issue of Bitcoin system rather the standard issue of computer security. More on this topic can be found in a paper by Plohmann and Gerhards-Padilla (Plohmann & Gerhards-Padilla, 2012). 4 Discussion Bitcoin is a product of computer science, i.e. cryptography, and as such represents the latest achievement. However, it is expected that some changes to the Bitcoin will occur in future as they occurred several times in history (bitcoin, n.d.). Bitcoin is by no means perfect and like any trail entity it needs to go through some changes in order to evolve. Changing the Bitcoin is not easy because it is decentralized and open source. This means that everyone can contribute to the Bitcoin, but changes will not come to place until the community (users) accepts them. This is the problem, since the user chooses 106 Ivan Švogor, Boris Tomaš weather to update or not it is likely that one point there will be two Bitcoin networks. Process of updating peer-to-peer network is security critical and has to be carried out carefully. Bitcoin and inflation Bitcoin has implemented inflation and it is called Controlled supply4. It states that by the end of 2140 there will be not more than 21 million of Bitcoins. Current Bitcoin value is around 450 USD which means that all 21 million of Bitcoins would be worth 9 .45 ∗ 109 USD. Current USA money supply in circulation is around 1.05 ∗ 1013 USD, i.e 10.5 trillion USD (Josh and Clark & Whitbourne, 2013). Imagining that USA swap USD for BTC it would mean that Bitcoin should be worth 1100 times more nearing 500.000,00 USD for 1 BTC, assuming that current demand comes only from USA. So, 1 USD beverage would cost 0,000.002 BTC. This is only for USA, considering rest of the world in this thought experiment would make this fraction several magnitudes smaller which is quite impractical for everyday use. Although, Bitcoin network can handle such small quantities, impression can be made that the creator of Bitcoin did not had global coverage in mind, i.e. to have one global currency which will replace all national currencies. Moreover, the implemented inflation is not ”true inflation”. It is simply controlled money supply with finite number of units. Bitcoins will get lost or forgotten, and no new Bitcoins can be produced. So, the value of Bitcoin will rise even more, over time. Many of those issues simply prove that Bitcoin is just the first attempt for ubiquitous and global currency. It would be wise to wait see other, alternative cryptocurrencies (Luther, 2013) that have different characteristics. There are several interesting alternative cryptocurrencies such as the one with only 42 coins minted5, or Doge coin which is available in unlimited volume. Bitcoin can be observed as unifying currency like Euro. Introducing Euro to the euro-zone took several years of preparation and lots of political good will. Although practical, some countries like UK still won’t join euro-zone due to political and economic interests (Mulhearn & Vane, 2005). Current state of politics is such that it will try to prevent any loss of power over money (currency) because implicitly money is power. This is also why some countries oppose Euro. Bitcoin is decentralized and under no authority in the world. 4 en.bitcoin.it/wiki/Controlled supply 5 www.42coin.org 107 The Bitcoin Phenomenon Analysis Figure 2 Bitcoin volatility (Albrecht, n.d.) Bitcoin in the world Due to limited inflation Bitcoin price depends only on supply and demand. If more people need Bitcon value of it will rise. Consider a scenario in which there are enough Bitcoins and is used worldwide. It would be expected that Bitcoin in one country has greater value than Bitcoin in other country. Earning wages would produce Bitcoins for individual, and he/she would possess a value which is greater in developed country due to greater labor value. In undeveloped countries prices would be lower because of low labor value. Although Bitcoin is decentralized the market which defines Bitcoin value, is not. This means that Bitcoin has different values in different countries, which might lead to social, economic and political instability. If we assume that market is decentralized and everyone has the same opportunities to earn and spend money, then all people would be equal, which often defined as a desirable society by Utopian socialists (Mises, 1994). Value of Bitcon can be controlled by country using taxes and social policy. In a country with higher taxes one Bitcoin would be worth less because value added tax (VAT) would consume large part of Bitcoin value. So, there is mechanism to change or control the value of Bitcoin; this means that powerful countries will be able to influence the value of Bitcoin. There is a danger that first large country that accepts Bitcoin will tie Bitcoin to its economy. Bitcoin volatility Being a newborn currency, Bitcoin on Figure 2 one can notice it is very volatile. According to Albrecht (Albrecht, n.d.), Bitcoin volatility is declining, slowly but steadily. Albrecht also concludes that Bitcoin behaves more like an asset rather than currency. Over time, volatility will settle with more acceptance of Bitcoin as valid currency. Unfortunately this behavior is circular, meaning that volatility will decline if more people accept it, and people will accept it more if it is less volatile. Bitcoin vs. other currencies Bitcoin is not the first digital currency; currently, digital currency is a currency stored on our bank accounts and it can be used digitally using Internet banking systems. This process is safe, centralized, non-anonymous and fast. Table I shows the comparison 108 Ivan Švogor, Boris Tomaš between different types of currencies: Digital currency, Bitcoin and paper money. Currency characteristic used to compare are: • Volatile: states the currency volatility. • Anonymous: defines anonymity of the currency. Currency is anonymous if transaction source and destination cannot be linked to physical person or any other entity. • Centralized: currency is centralized if it is monitored and governed by central institution like Central Bank or government. • Secure: is currency theft and fraud secure? • Offline: can currency be used without Internet connection? Selection of characteristics used is based on a previous research papers, identified in this work, that pinpoint key features and weaknesses of cryptocurrency comparing to real currency. Characteristic Digital currency Bitcoin Paper money Volatile No Yes No Anonymous No Yes (Maybe)6 Yes Centralized Yes No Yes Secure Yes Yes (Maybe)7 Maybe8 Offline No No (Maybe)9 Yes Table 1: Summary comparison of different currency types 4.1 Conclusion This paper presented facts about Bitcoin and current research about its influence in order to clarify some of the reasons for its popularity. It is pinpointed that Bitcoin is noteworthy idea which has great economic and social potential, however there are some issues that may delay adoption of this digital currency and postpone the revolution it promised. By design, Bitcoin is decentralized and cannot be controlled - due to this feature it is not politically friendly, hence there is huge effort among large economies (China, USA, Russia, etc.) to slow down Bitcoin. Even though, every day more and more merchants accept Bitcoin as legal payment option. The future is still to show what it holds for Bitcoin. 6 In this paper it has been noted that Bitcoin is not completely anonymous. 7 There has been several security incidents. 8 Paper money has been tried to be copied constantly. 9 There are attempts to make offline version of Bitcoin: www.casascius.com 109 The Bitcoin Phenomenon Analysis References Albrecht, R. (n.d.). Bitcoin Volatility – The 4 perspectives. Retrieved from http://bitcoinmagazine.com/6543/bitcoin-volatility-analysis/ Babaioff, M., Dobzinski, S., Oren, S., & Zohar, A. (2011). On Bitcoin and red balloons. ACM SIGecom Exchanges, 10(3), 5–9. doi:10.1145/2325702.2325704 Barber, S., Boyen, X., Shi, E., & Uzun, E. (2012). Bitter to better—how to make bitcoin a better currency. Financial Cryptography and Data …, 399–414. Retrieved from http://link.springer.com/chapter/10.1007/978-3-642-32946- 3_29 bitcoin. (n.d.). Bitcoin version history. Retrieved from https://bitcoin.org/en/version-history Bradbury, D. (2013). The problem with Bitcoin. Computer Fraud & Security, 2013(11), 5–8. doi:10.1016/S1361-3723(13)70101-5 Bryans, D. (2014). Bitcoin and Money Laundering: Mining for an Effective Solution. Indiana Law Journal. Retrieved from http://www.repository.law.indiana.edu/cgi/viewcontent.cgi?article=11100&c ontext=ilj CEX.IO Ltd. (2014). Crypto and Bitcoin Mining Pool. Retrieved from https://ghash.io/ Clark, J. and, & Whitbourne, K. (2013). How much actual money is there in the world? Retrieved from http://money.howstuffworks.com/how-much-money- is-in-the-world.htm Clark, J., & Essex, A. (2012). CommitCoin: carbon dating commitments with Bitcoin. Financial Cryptography and Data Security, 390–398. Retrieved from http://link.springer.com/chapter/10.1007/978-3-642-32946-3_28 Coinbase. (2014). Bitcoin Wallet - Coinbase. Retrieved from https://coinbase.com/charts December, I. (2013). Massive Bitcoin thefts and seizures leave many users nervous and poorer. Computer Fraud & Security, 2013(12), 1–3. doi:10.1016/S1361-3723(13)70106-4 Decker, C., & Wattenhofer, R. (2013). Information propagation in the bitcoin network. Peer-to-Peer Computing (P2P), …. Retrieved from http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6688704 Grinberg, R., Primer, B., Ecosystem, B., Sustainable, I. B., & Issues, L. (2012). Bitcoin: an innovative alternative digital currency. Hastings Sci. & Tech. LJ, 50. Retrieved from http://heinonlinebackup.com/hol-cgi- 110 Ivan Švogor, Boris Tomaš bin/get_pdf.cgi?handle=hein.journals/hascietlj4§ion=6\nhttp://papers.ss rn.com/sol3/papers.cfm?abstract_id=1817857 Hobson, D. (2013). What is Bitcoin? XRDS: Crossroads, The ACM Magazine for Students, 20(1), 40. doi:10.1145/2510124 Kim, G. (2014). New Virtual Currency Bitcoin Its Past, Present, and Future. Kondor, D., Pósfai, M., Csabai, I., & Vattay, G. (2014). Do the rich get richer? An empirical analysis of the Bitcoin transaction network. PloS One, 9(2), e86197. doi:10.1371/journal.pone.0086197 Kristoufek, L. (2013). BitCoin meets Google Trends and Wikipedia: quantifying the relationship between phenomena of the Internet era. Scientific Reports, 3, 3415. doi:10.1038/srep03415 Luther, W. (2013). Cryptocurrencies, Network Effects, and Switching Costs. Mercatus Center, 37. doi:10.2139/ssrn.2295134 Martins, C. S., & Yang, Y. (n.d.). Introduction to Bitcoins : A pseudo-anonymous electronic currency system Theme : Abstract : Integrated Solution : Smarter Commerce, 349–350. Miers, I., Garman, C., Green, M., & Rubin, A. D. (2013). Zerocoin: Anonymous Distributed E-Cash from Bitcoin. 2013 IEEE Symposium on Security and Privacy, 397–411. doi:10.1109/SP.2013.34 Mises, L. von. (1994). Economic Calculation in the Socialist Commonwealth. In Classics in Austrian economics: A Sampling in the History of a Tradition (Vol. 3, pp. pp. 2–30). Retrieved from http://www.mises.org/ Moore, T., & Christin, N. (2013). Beware the middleman: Empirical analysis of Bitcoin-exchange risk. Financial Cryptography and Data Security, (June 2011), 25–33. Retrieved from http://link.springer.com/chapter/10.1007/978- 3-642-39884-1_3 Mulhearn, C., & Vane, H. R. (2005). The UK and the Euro: Debating the British Decision. The World Economy, 28, 243–258. doi:10.1111/j.1467- 9701.2005.00648.x Nakamoto, S. (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. Consulted, 1–9. doi:10.1007/s10838-008-9062-0 Nechvatal, J. (1991). Public-key cryptography. Peck, M. (2013). The Bitcoin Arms Race is on! IEEE Spectrum, 50(6), 11–13. doi:10.1109/MSPEC.2013.6521016 Peck, M. E. (2012). The cryptoanarchists’ answer to cash. Spectrum, IEEE, 49(6), 50–56. 111 The Bitcoin Phenomenon Analysis Plohmann, D., & Gerhards-Padilla, E. (2012). Case study of the Miner Botnet. Cyber Conflict (CYCON), …. Retrieved from http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=6243985 Reid, F., & Harrigan, M. (2011). An Analysis of Anonymity in the Bitcoin System. 2011 IEEE Third Int’l Conference on Privacy, Security, Risk and Trust and 2011 IEEE Third Int'l Conference on Social Computing, 1318–1326. doi:10.1109/PASSAT/SocialCom.2011.79 Ron, D., & Shamir, A. (2013). Quantitative analysis of the full bitcoin transaction graph. Financial Cryptography and Data Security, 6–24. Retrieved from http://link.springer.com/chapter/10.1007/978-3-642-39884-1_2 Taylor, M. B. (2013). Bitcoin and The Age of Bespoke Silicon How Bitcoin Works : User Perspective Bitcoin Mining : Miner ’ s Perspective. Van Hout, M. C., & Bingham, T. (2014). Responsible vendors, intelligent consumers: Silk Road, the online revolution in drug trading. The International Journal on Drug Policy, 25(2), 183–9. doi:10.1016/j.drugpo.2013.10.009 Yermack, D. (2013). Is Bitcoin a Real Currency?, 1–14. Retrieved from http://www.nber.org/papers/w19747 112 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Towards an Artifact-Oriented Requirements Engineering Model for Developing Successful Products, Services, and Systems: Identification of Model Requirements Christian Ruf University of St.Gallen Institute of Information Management, Switzerland christian.ruf@unisg.ch Abstract Despite extensive research in the domain of requirements engineering (RE), companies still struggle with this discipline. Moreover, practitioners are challenged with developing successful products, services, and systems which address the true needs of their customers. This gives rise to a new research field in the domain of RE, namely artifact orientation. According to the literature, this artifact orientation should increase the success of RE significantly. By conducting a literature review and 7 expert interviews, we identified 7 model requirements (MRs) for an artifact-oriented RE model. Furthermore, the results of this paper suggest that existing artifact- oriented RE models do not sufficiently address all identified MRs. In particular, these models lack the combination of traditional RE practices, such as goal orientation, documentation, and traceability with novel agile approaches. Furthermore, there is a need for a more holistic RE which merges the domains of product, service, and software engineering. Keywords: Requirements Engineering, Artifact Orientation, Software Engineering, Service Engineering 1 Introduction Over the past 20 years, extensive research has been conducted and the literature has shown an emphasis on the importance of requirements engineering (RE) in order to develop successful products, services, and systems (Méndez Fernández & Wagner, 2014). RE is the process of capturing, analyzing, prioritizing, negotiating, and documenting user needs or requirements (Sommerville & Kotonya, 1998). Despite the acknowledged relevance of this research domain, companies still struggle with adopting RE (Beecham et al., 2005). Only 48% 113 Christian Ruf of projects are completed on time and 55% do not meet the budget plan due to insufficient RE (Kassab, Neill, & Laplante, 2014). In order to increase the successful adoption of RE practices, which will lead to more successful products and services, researchers have introduced artifact orientation (Broy, 2006b). Artifact orientation focuses on what kind of requirements the RE team should elicit, document, analyze, and negotiate. Such artifacts might include high-level customer goals or specific software specifications. Recent findings suggest that this young discipline of artifact orientation will lead to more successful RE which facilitates developing products, services, and systems that meet the needs of the relevant stakeholders and customers (Méndez Fernández & Penzenstadler, 2014). Within this article, we focus on the artifact orientation of RE and propose the following research question: What are requirements for an artifact-oriented RE model that facilitates developing products, services, and software that meet the needs of stakeholders and customers? In order to address this research question, we introduce and define the relevant terms and the related work in Section 2; in particular, the artifact orientation. Subsequently, we describe the research approach, including a literature review and 7 expert interviews, in Section 3. Moreover, we reveal the findings regarding the identified requirements for developing an artifact-oriented RE model in Section 4. A comparison with existing artifact-oriented models suggests potential research gaps. We discuss the implications based on these results in Section 5 and provide guidance on how to design a new artifact-oriented RE model. Finally, Section 6 elaborates on future research and limitations. 2 Related Work In this section, three concepts are defined in further detail: (1) model requirements (MRs); (2) requirements engineering (RE); and (3) artifact orientation. (1) Model requirements (MRs). A requirement describes a fundamental attribute of a system along with an appropriate value statement (Grady, 2010). Hence, a requirement refers to an attribute or a characteristic of a product, service, or system. Such requirements originate from stakeholders, e.g. users, customers, or employees, and address a particular problem or need. There are three categories of requirements. Functional requirements relate to specific features that a product, service, or system possesses. Robertson and Robertson (2013) define a functional requirement as follows: “Functional requirements are things the product must do” . The second category of non-functional requirements describes the more general characteristics of a product, service, or system. Such characteristics might include the usability, security, or availability of a system. By constraints, a third category, the literature refers to specific limits on how products and services should be developed. Such constraints may relate to the design of a software artifact for a particular operating system (for example, Android or iOS). In this study, we identify model requirements (MRs) for an artifact-oriented RE model. These MRs incorporate one of the three requirement categories and describe how the artifact- oriented RE model should be developed. 114 Towards an Artefact-Oriented Requirements Engineering Model (2) Requirements engineering (RE). RE is the process of capturing, analyzing, prioritizing, and documenting requirements. There are four common RE processes which the RE community acknowledges (Pohl, 2008). (1) Elicitation focuses on capturing requirements from different stakeholder groups. Many different techniques facilitate this first step of RE. (2) Analysis and negotiation address the resolution of conflicts between elicited requirements. The result should provide the RE team with consistent and unambiguous requirements. (3) Documentation is facilitated with natural or formal language, such as Unified Modeling Language. (4) Validation incorporates prototyping and testing. By applying such techniques, we ascertain whether the elicited, specified, and documented requirements represent customer and stakeholder needs (Sommerville & Kotonya, 1998). The output of RE should lead to the successful design of a product, service, or system (Hall, Beecham, & Rainer, 2002). (3) Artifact orientation. Artifact orientation focuses on describing the content elements, identifying the kind of requirements the RE team should capture (Penzenstadler et al., 2013). As opposed to the artifact orientation, the activity orientation includes RE techniques and methods which advise how the team should go about the RE process (Jiang et al., 2007). Consequently, requirement categories such as goals, constraints, or specific software specifications refer to requirement artifacts, while focus groups, workshops, and process models refer to RE activities. In this study, we focus on artifact-oriented RE models (Loucopoulos & Kavakli, 1995; Méndez Fernández & Penzenstadler, 2014; Nuseibeh & Easterbrook, 2000). 3 Research Approach We first elaborate on details with respect to the literature review and, secondly, provide more details on how we conducted the expert interviews in order to capture the MRs for an artifact- oriented RE model. 3.1 Literature Review We followed the guidelines set by Vom Brocke et al. (2009) and Webster and Watson (2002) for conducting the literature review. We used the keyword “requirements engineering” in the databases, as summarized in Table 1, including a backwards and forwards search. We considered the following parameters in order to determine whether a given paper should be included: published after 2000; peer-reviewed; in English or German language; and an article from a journal or an A-ranked conference. The article also needed to cover MRs for the design of an artifact-oriented RE model, or it needed to introduce an artifact-oriented model and address a holistic RE. We started the literature review in October 2014 and finished in December 2014. With respect to the included conferences, we chose the “WI-Orientierungsliste” (WKWI, 2008) as a reference and only included three A-ranked conferences; the International Conference on Information Systems (ICIS), the European Conference on Information Systems (ECIS), and 115 Christian Ruf Wirtschaftsinformatik (WI). Given the mentioned restrictions, we selected 145 articles from 768 hits in the literature review. Source Hits Analyzed Relevant Keyword search AIS Electronic Library (AISel) 16 10 6 EBSCOhost 164 21 14 Emerald Insight 26 1 0 IEEE Xplore Digital Library 282 31 18 ScienceDirect 158 7 4 Springer Link 122 14 11 Backwards and forwards search – 61 52 TOTAL 768 145 105* *We present a representative sample regarding the identified articles in this research paper. A comprehensive and complete list of the identified articles can be provided on request. Table 1: Details of the Literature Review 3.2 Expert Interviews In order to validate and triangulate the findings from the literature review, we also conducted 7 expert interviews between November 2014 and January 2015 (Table 2 incorporates background information regarding the expert interviews). The 7 interviewees have extensive domain experience (3 years or more) and work for multinational and Swiss companies. Interviewee Industry Domain Experience Expert 1 Senior Consultant Consulting and Agency 5 years Expert 2 Senior Principal Consultant Consulting and Agency 16 years Expert 3 Senior Consultant Consulting and Agency 3 years Expert 4 Senior Consultant Consulting and Agency 4 years Expert 5 Product Manager IT Service Provider 4 years Expert 6 Senior Supply Chain Director Fast Moving Consumer Goods 10 years Expert 7 Consultant Consulting 4 years Table 2: Background Information of the Experts These interviews lasted between 35 and 58 minutes and followed a semi-structured guideline. First, the interviewer asked questions about the background and experience of each expert. Second, the expert provided details on the current RE practices in the respective organization. Finally, we were interested in how the expert would improve the RE process in the future. We put the interview transcripts in a central database. Furthermore, two researchers coded the 116 Towards an Artefact-Oriented Requirements Engineering Model transcripts with the coding schema independently. The coding schema entailed 7 MRs, which we identified in the literature (Table 3 in Section 4). We measured the intercoder reliability with the overall percentage agreement and the Cohen’s Kappa coefficient (Cohen, 1968), two well-acknowledged indicators (Dewey, 1983). The percentage agreement between the two coders was 88% and the Cohen’s Kappa result was 0.65, which is above the threshold of 0.6 as suggested by the equally arbitrary guidelines from Fleiss, Lewin, and Paik (2013). Subsequent to the independent coding process, the researchers negotiated the discrepancies in a workshop and examined each of the interviews thoroughly. Finally, they discussed whether the code matched the quotations from the interviews and consequently supported the MRs identified in the existing body of knowledge. The subsequent Section 4 highlights the results from both the literature review and the expert interviews. 4 Results 4.1 Model Requirements (MRs) Table 3 provides a summary of the seven MRs and the respective sources (literature and expert interviews). Due to the page limit of this research paper, we only quote a representative sample of the identified articles and cite selective statements from the expert interviews here. On request, we are happy to provide the entire transcripts and results from the literature review as well as the expert interviews. (MR1) Goal orientation: The RE model should link requirements to customer and service provider goals. The most important goal for a service provider is to address the needs of its customers and stakeholders (Berkovich, München, & Leimeister, 2009), which consequently means involving customers in the RE process. Hence, capturing customer goals is critical for an RE model. However, service providers do not pursue purely altruistic goals; they also want to benefit financially from launching products and services. Therefore, besides a strong customer orientation, the service provider goals are equally important (Nuseibeh & Easterbrook, 2000). Moreover, by linking requirements back to the overall customer and service provider goals, the RE team is able to prioritize requirements (Lee et al., 2013). A representative quote from Expert 2 also supports this first MR: “There was a high-level goal which guided us in the requirements engineering process: with one workplace we add value to the company...” All of the 7 experts mentioned the importance of this MR. (MR2) Documentation and traceability: The RE model should facilitate a thorough documentation process and consequently enable traceability. Documentation and traceability describe the process of following the artifact throughout the development process (Méndez Fernández & Wagner, 2014). A quotation from Expert 4 supports the importance of (MR2): “I find it crucial that documentation is carried out throughout the entire RE process. Sometimes our customers or project team members do not get how important documentation 117 Christian Ruf really is.” 6 out of the 7 interviewed experts also specifically mentioned the importance of documentation and traceability. Source wie evRe 1 2 3 4 5 6 7 Model requirements (MRs) Literatur Expert Expert Expert Expert Expert Expert Expert (MR1) Goal orientation x x x x x x x x (MR2) Documentation x x x x x x - x (MR3) Integration x x - x - - x x (MR4) Agility x x x x x - - x (MR5) Adaptability x - x - - x x - (MR6) Continuity x x x x x x x x (MR7) Responsibilities x x x x x x x x x = addressed, - = not addressed Table 3: The Identified MRs from the Literature Review and the Expert Interviews (MR3) Integration: The RE model should integrate the RE for products, services, and systems. We argue that the RE discipline should integrate three specific views – namely software, service, and product development (Grau, 2012) – into an RE process for product service systems (PSS). The literature also suggests that “technology and service design decisions become deeply intertwined” (Patrício et al., 2009, p. 210). Other researchers confirm the necessity of combining RE with software product line engineering (Lee et al., 2013). Hence, the (MR3) acknowledges the integration of different disciplines (Nuseibeh & Easterbrook, 2000), and particularly, RE practices from the domain of product, service, and software engineering. With such an integration, the RE process elicits and specifies requirements for PSS, or hybrid products, as suggested by the literature (Berkovich et al., 2012). Expert 1 elaborated on (MR3) with the following statement: “We needed to change the ERP integration, the payment, customer processes and so on and so forth”. Regarding (MR3), 4 out of the 7 experts mentioned the importance of integrating product, service, and software development. (MR4) Agility: The RE model should allow for fast throughput time. A large number of RE techniques and models visualize the RE process as a sequence or an iterative approach, but in reality, requirements are captured, analyzed, negotiated, and prioritized in a parallel order. Agility refers to a parallel RE and development process (Hickey & Davis, 2004) that results in a fast throughput time (Broy, 2006a). Expert 6 described this parallel development and RE process as follows: “We organized workshops to validate requirements and started with the development at the same time …” In total, 4 experts confirmed this MR. 118 Towards an Artefact-Oriented Requirements Engineering Model (MR5) Adaptability: The RE model should be adaptable to different organizational and project contexts. Each organization, each business unit, and each project has unique and different characteristics. Researchers acknowledge that RE needs to be adaptive to such different contexts (Grau, 2012; Sarker & Sarker, 2009). In order to address such different project setups, establishing a shared understanding at the beginning of the project (Hanisch & Corbitt, 2007) clearly represents a success factor for an artifact-oriented RE model. Regarding (MR5), Expert 5 provided a representative quote: “I find it very challenging to adapt the same model for different organizations with different industry backgrounds. There are so many different factors which might lead to different requirement engineering processes.” Adaptability as an MR was mentioned by three of the domain experts. (MR6) Continuity: The RE model should provide support for the continuous evaluation of the elicited requirements throughout, and beyond, the project. Continuously evolving requirements not only require agile RE, as discussed with (MR4), but also the continuous evaluation of artifacts (Ramesh, Cao, & Baskerville, 2010). More specifically, they require multi-disciplinary teams and the evaluation of requirements on a continuous basis (Cox, Niazi, & Verner, 2009). Regardless of the chosen approach, the evaluated requirements need to comply with the IEEE recommended practice for requirements specification (IEEE, 1998). Consequently, the requirements should be correct, unambiguous, complete, consistent, ranked for importance, verifiable, modifiable, and traceable. Expert 7 gave an example regarding the continuous evaluation of requirements: “During the development, we challenged the requirements with various iterations. Do we really need this requirement, does it really make sense?” Apart from Expert 7, all of the other interviewees emphasized the importance of continuous evaluation processes. (MR7) Responsibilities: The RE model should help to define responsibilities and roles throughout the RE process. RE teams and development project team members often organize themselves and are self-directing, particularly in agile project contexts. The motivation and drive of employees, along with the required skillset of team members, are prerequisites for this type of informal project organization (Pink, 2011). However, researchers also state that formal project organizations and the definition of role responsibilities are prerequisites for successful RE (Méndez Fernández & Wagner, 2014). Other researchers confirm this and elaborate that the definition of roles and responsibilities, particularly that of the requirements analyst, is critical (Klendauer et al., 2012). Expert 3 also stated the necessity of defining roles and responsibilities in an agile project: “Our customer chose an agile Scrum process. This is quite risky, given that the project involved 16 different apps from 6 different service providers… Everything is coordinated through the product manager…” Overall, this MR was acknowledged by all of the experts. 4.2 Artifact-Oriented Requirements Engineering (RE) Models In total, we identified and validated the 7 MRs an artifact-oriented RE model needs to address. Based on the literature review, we also identified 11 already published artifact-oriented 119 Christian Ruf models. A comparison with the 7 MRs suggests a potential research gap and gives rise to future endeavors in this field of artifact orientation. A domain-independent requirements engineering approach (AMDiRE). The AMDiRE artifact model builds on the REM (RE Reference Model) (Méndez Fernández & Penzenstadler, 2014). Despite addressing (MR1), (MR2), (MR5), and (MR7), the model fails to address (MR4) agile practices, (MR6) continuity and (MR3) integration. Scenario and goal-based system development method (COSMOD-RE). The COSMOD-RE (sCenario and gOal based SysteM development methOD – RE) introduces co-design or concurrent RE and artifact development (Pohl & Sikora, 2007). Accordingly, this approach is particularly successful for developing innovative and software-intensive systems. However, the COSMOD-RE only addresses (MR1), (MR2), and (MR4). Goal-Oriented Requirements Engineering (GORE). There are several goal-oriented modeling languages, such as KAOS (Keep All Objectives Satisfied) (Dardenne, van Lamsweerde, & Fickas, 1993) or Tropos (Bresciani et al., 2004), for example. Within our analysis, we combined these goal-oriented RE approaches because of their comparability. However, these GORE models lack the inclusion of (MR7) and hence, do not define roles and responsibilities. Furthermore, (MR4), agile practices, is not supported. Requirements Abstraction Model (RAM). The RAM (Gorschek & Wohlin, 2006) facilitates a goal-oriented approach for capturing requirements. The authors introduced four levels in order to specify requirements on the goal, feature, function, and component levels. The model is widely acknowledged, but meets neither (MR3) integration of product, service, systems, nor (MR4) agile RE practices. The model also lacks (MR5) adaptability and does not include (MR7) responsibilities. Requirements Data Model (RDMod). The RDMod follows an iterative process for capturing requirements for PSS (Berkovich et al., 2012). Thus, the model particularly addresses (MR3) integration. Moreover, it combines requirements artifacts from the RAM and the REM. The authors also include the four requirements abstraction levels: goals, features, functions, and components. However, as with previous artifact-oriented models, (MR4) is not addressed. Requirements Engineering Reference Model (REM). The REM is an approach that includes various artifacts (Broy et al., 2007). Similar to the RAM, the model does not meet (MR3) and (MR4). REMsES. The REMsES (Requirements Engineering und Management für softwareintensiver Eingebetteter Systeme) resulted from a practical engineering project in the automotive industry (Méndez Fernández & Penzenstadler, 2014). Other than agile practices (MR4), the model includes all 7 MRs. Scrum-based model for software products. Scrum is an agile software development method that was developed together with the agile manifesto (Beck et al., 2001). Hence, it particularly 120 Towards an Artefact-Oriented Requirements Engineering Model addresses (MR4) agility. However, (MR3) the integration of products, services, and systems is not supported with this model. Software and Systems Requirements Engineering (SSRE). Researchers have introduced the SSRE for analyzing different RE practices in an extensive literature review (Parviainen et al., 2003). The SSRE covers the entire RE process and suggests different artifacts throughout the development process. However, several MRs are not met, including (MR1), (MR3), (MR5), and (MR7). Volere Model. The Volere RE Model proposes various templates for eliciting, analyzing, and documenting requirements (Robertson & Robertson, 2013). Two MRs – agile practices (MR4) and responsibilities (MR7) – are not mentioned by the Volere model. V Model. Researchers and practitioners have discussed a particular soft- or hardware development process that includes an RE elicitation and management process in the V Model (Hoffmann, 2012). This artifact model only addresses (MR2) documentation and (MR6) continuous evaluation. After having examined each of the identified artifact-oriented RE models, we concluded that none of the models addresses all 7 MRs. We discuss the implications of these findings in the subsequent section. Artifact-oriented RE models eldo m eld -RE ased o E D d -b M el iR o sES m d M o RE lere Model requirements (MRs) AMD COSMO GORE RAM RD REM REM Scru SS Vo V M (MR1) Goal orientation x x x x x x x x - x - (MR2) Documentation x x x x x x x x x x x (MR3) Integration - - x - x - x - - x - (MR4) Agility - x - - - - - x x - - (MR5) Adaptability x - x - - x x x - x - (MR6) Continuity - - x x x x x x x x x (MR7) Responsibilities x - - - - x x x - - - x = addressed, - = not addressed Table 4: Comparing MRs with Existing Artifact-Oriented RE Models 5 Discussion From the results in the previous section two recommendations were derived. First, we believe that (MR4) agility is a crucial requirement, which is acknowledged by both researchers and 121 Christian Ruf practitioners. However only a few of the artifact-oriented models truly support this requirement. In particular, the combination of traditional MRs, such as (MR1) goal orientation and (MR2) documentation and traceability, with agile RE practices seems critical. Hence, we argue that future research should address this combination of goal orientation, documentation, and traceability, with the agile domain more thoroughly. Second, future research should address the (MR3) integration of PSS specifically. The analysis of the expert interviews supports that argument. Several consultants who work on projects that introduce new mobile applications, for example, described their projects mainly as software projects. However, implicitly, they discussed various problems with the integration of such software projects in the organizational environment. Obviously, due to the company size, communication across different departments may become a challenge. Furthermore, such software projects implicitly change business processes and legacy systems, which is not usually addressed thoroughly in the RE process. Not only practitioners, but also the existing body of knowledge only partially addresses the integration of products, services, and systems in the RE process (only 4 models meet this MR). Hence, the holistic RE approach, or combining product, service, and software engineering, poses the second major challenge that a future artifact- oriented RE model should solve. 6 Conclusions Based on a literature review and 7 expert interviews we identified and validated 7 model requirements (MRs) for an artifact-oriented requirements engineering (RE) model that should facilitate successful product, service and software engineering endeavors. After having compared the 7 MRs with existing artifact-oriented models, we identified a potential research gap which consequently legitimates future work in this domain. In particular, the combination of agile and traditional RE practices, such as documentation, traceability, or goal orientation, is a first area of future research. The second area addresses more holistic RE which should include the domains of product, service, and software engineering. Having validated the results with 7 experts, the findings should not only contribute to the existing body of knowledge, but should also be relevant for all practitioners. Despite promising findings and contributions, we also need to discuss some limitations. First, most of the interviewees had a background in the consulting and agency industry. Hence, the findings might be biased due to the limited variety of backgrounds in terms of positions and industry. Second, we also argue that 7 expert interviews is not a sufficient number in order to provide entirely reliable results. Consequently, we propose that despite our chosen triangulation approach with data from the literature review and the expert interviews, the results should be further validated. Third, despite having conducted an exhaustive literature review, the limited selection of databases in the field of information management and software engineering literature might present a bias. We expect that more MRs or artifact- oriented RE models might be discussed in the product engineering discipline which we only 122 Towards an Artefact-Oriented Requirements Engineering Model marginally addressed in our literature review. We tried to limit this bias by involving experts to triangulate the findings. Acknowledgements We would like to thank Alexander Müdespacher, Patrick Schenker, and Silvan Schumacher for collaborating on the research project and for supporting us in the data collection and data analysis process. Furthermore, we also appreciate the feedback from the reviewers which we tried to address in the final submission. References Beck, K., Beedle, M., Van Bennekum, A., Cockburn, A., Cunningham, W., Fowler, M., … others. (2001). Manifesto for agile software development. Beecham, S., Hall, T., Britton, C., Cottee, M., & Rainer, A. (2005). Using an expert panel to validate a requirements process improvement model. Journal of Systems & Software, 76(3), 251–275. Retrieved from 10.1016/j.jss.2004.06.004 Berkovich, M., Leimeister, J. M., Hoffmann, A., & Krcmar, H. (2012). A requirements data model for product service systems. Requirements Engineering. doi:10.1007/s00766-012- 0164-1 Berkovich, M., München, T. U., & Leimeister, J. M. (2009). An Empirical Exploration of Requirements Engineering for Hybrid Products. In ECIS 2009 Proceedings. (p. Paper 67). Bresciani, P., Perini, A., Giorgini, P., Giunchiglia, F., & Mylopoulos, J. (2004). Tropos: An Agent- Oriented Software Development Methodology. Autonomous Agents and Multi-Agent Systems, 8(3), 203–236. doi:10.1023/B:AGNT.0000018806.20944.ef Broy, M. (2006a). Challenges in automotive software engineering. In Proceeding of the 28th international conference on Software engineering - ICSE ’06 (p. 33). New York, New York, USA: ACM Press. doi:10.1145/1134285.1134292 Broy, M. (2006b). Requirements Engineering as a Key to Holistic Software Quality. (A. Levi, E. Savaş, H. Yenigün, S. Balcısoy, & Y. Saygın, Eds.) (Vol. 4263). Berlin, Heidelberg: Springer Berlin Heidelberg. doi:10.1007/11902140 Broy, M., Geisberger, E., Kazmeier, J., Rudorfer, A., & Beetz, K. (2007). Ein Requirements- Engineering-Referenzmodell. Informatik-Spektrum, 30(3), 127–142. doi:10.1007/s00287- 007-0149-5 Cohen, J. (1968). Weighted kappa: nominal scale agreement with provision for scaled disagreement or partial credit. Psychological Bulletin, 70, 213–220. doi:10.1037/h0026256 123 Christian Ruf Cox, K., Niazi, M., & Verner, J. (2009). Empirical study of Sommerville and Sawyer’s requirements engineering practices. IET Software, 3(5), 339. doi:10.1049/iet- sen.2008.0076 Dardenne, A., van Lamsweerde, A., & Fickas, S. (1993). Goal-directed requirements acquisition. Science of Computer Programming, 20(1–2), 3–50. doi:http://dx.doi.org/10.1016/0167- 6423(93)90021-G Dewey, M. E. (1983). Coefficients of agreement. British Journal of Psychiatry, 143, 487–489. doi:10.1192/bjp.143.5.487 Fleiss, J. L., Lewin, B., & Paik, M. C. (2013). Statistical methods for rates and proportions. John Wiley & Sons. Gorschek, T., & Wohlin, C. (2006). Requirements Abstraction Model. Requirements Engineering, 11(1), 79–101. doi:10.1007/s00766-005-0020-7 Grady, J. O. (2010). System requirements analysis. Academic Press. Grau, R. (2012). Requirements Engineering in Agile Software Development. In A. Maedche, A. Botzenhardt, & L. Neer (Eds.), Software for People SE - 6 (pp. 97–119). Springer Berlin Heidelberg. doi:10.1007/978-3-642-31371-4_6 Hall, T., Beecham, S., & Rainer, A. (2002). Requirements problems in twelve software companies: an empirical analysis. Software, IEE Proceedings -. doi:10.1049/ip- sen:20020694 Hanisch, J., & Corbitt, B. (2007). Impediments to requirements engineering during global software development. European Journal of Information Systems, 16(6), 793–805. doi:10.1057/palgrave.ejis.3000723 Hickey, A. M., & Davis, A. M. (2004). A Unified Model of Requirements Elicitation. Journal of Management Information Systems, 20(4), 65–84. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=buh&AN=12841174&site=ehost -live Hoffmann, H. (2012). Streamlining the development of complex systems through model-based systems engineering. Digital Avionics Systems Conference (DASC), 2012 IEEE/AIAA 31st, 6E6–1–6E6–8. doi:10.1109/DASC.2012.6382404 IEEE. (1998). IEEE Recommended Practice for Software Requirements Specifications. IEEE Std 830-1998. doi:10.1109/IEEESTD.1998.88286 Jiang, L., Eberlein, A., Far, B. H., & Mousavi, M. (2007). A methodology for the selection of requirements engineering techniques. Software & Systems Modeling, 7(3), 303–328. doi:10.1007/s10270-007-0055-y 124 Towards an Artefact-Oriented Requirements Engineering Model Kassab, M., Neill, C., & Laplante, P. (2014). State of practice in requirements engineering: contemporary data. Innovations in Systems and Software Engineering, 10(4), 235–241. doi:10.1007/s11334-014-0232-4 Klendauer, R., Berkovich, M., Gelvin, R., Leimeister, J. M., & Krcmar, H. (2012). Towards a competency model for requirements analysts. Information Systems Journal, 22(6), 475– 503. Retrieved from 10.1111/j.1365-2575.2011.00395.x Lee, J., Kang, K. C., Sawyer, P., & Lee, H. (2013). A holistic approach to feature modeling for product line requirements engineering. Requirements Engineering, 19(4), 377–395. doi:10.1007/s00766-013-0183-6 Loucopoulos, P., & Kavakli, E. (1995). Enterprise Modelling and the Teleological Approach to Requirements Engineering. International Journal of Cooperative Information Systems, 04(01), 45–79. doi:10.1142/S0218843095000032 Méndez Fernández, D., & Penzenstadler, B. (2014). Artefact-based requirements engineering: the AMDiRE approach. Requirements Engineering. doi:10.1007/s00766-014-0206-y Méndez Fernández, D., & Wagner, S. (2014). Naming the pain in requirements engineering: A design for a global family of surveys and first results from Germany. Information and Software Technology. doi:10.1016/j.infsof.2014.05.008 Nuseibeh, B., & Easterbrook, S. (2000). Requirements Engineering: A Roadmap. In Proceedings of the Conference on The Future of Software Engineering (pp. 35–46). New York, NY, USA: ACM. doi:10.1145/336512.336523 Parviainen, P., Hulkko, H., Kaariainen, J., Takalo, J., & Tihinen, M. (2003). Requirements engineering inventory of technologies. VTT PUBLICATIONS. Patrício, L., Falcão e Cunha, J., & Fisk, R. (2009). Requirements engineering for multi-channel services: the SEB method and its application to a multi-channel bank. Requirements Engineering, 14(3), 209–227. doi:10.1007/s00766-009-0082-z Penzenstadler, B., Mendez Fernandez, D., & Eckhardt, J. (2013). Understanding the Impact of Artefact-Based RE -- Design of a Replication Study. Empirical Software Engineering and Measurement, 2013 ACM / IEEE International Symposium on. doi:10.1109/ESEM.2013.46 Pink, D. (2011). Drive: The Surprising Truth About What Motivates Us. Riverhead Books. Pohl, K. (2008). Requirements Engineering: Grundlagen, Prinzipien, Techniken (Vol. 2). dpunkt Verlag GmbH. Pohl, K., & Sikora, E. (2007). COSMOD-RE: Supporting the Co-Design of Requirements and Architectural Artifacts. Requirements Engineering Conference, 2007. RE ’07. 15th IEEE International. doi:10.1109/RE.2007.25 125 Christian Ruf Ramesh, B., Cao, L., & Baskerville, R. (2010). Agile requirements engineering practices and challenges: an empirical study. Information Systems Journal, 20(5), 449–480. Retrieved from 10.1111/j.1365-2575.2007.00259.x Robertson, S., & Robertson, J. (2013). Mastering the Requirements Engineering Process - Getting Requirements Right (3rd ed.). New Jersey: Addison-Wesley. Sarker, S., & Sarker, S. (2009). Exploring Agility in Distributed Information Systems Development Teams: An Interpretive Study in an Offshoring Context. Information Systems Research, 20(3), 440–461. doi:10.2307/23015474 Sommerville, I., & Kotonya, G. (1998). Requirements engineering: processes and techniques. John Wiley & Sons, Inc. Vom Brocke, J., Simons, A., Niehaves, B., Reimer, K., Riemer, K., Plattfaut, R., & Cleven, A. Reconstructing the Giant: On the Importance of Rigour in Documenting the Literature Search Process (2009). ECIS. Webster, J., & Watson, R. T. (2002). Analyzing the Past to Prepare for the Future: Writing a Literature Review. MIS Quarterly, 26(2), xiii–xxiii. WKWI. (2008, June 20). WI-Orientierungslisten. WIRTSCHAFTSINFORMATIK. doi:10.1365/s11576-008-0040-2 126 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Financial Market Surveillance Decision Support: An Explanatory Design Theory Irina Alić University of Göttingen, Germany irina.alic@wiwi.uni-goet ingen.de Abstract In this paper, an explanatory design theory for Financial Market Surveillance Systems is presented, which addresses both user requirements and regulatory demands. The identified general requirements and generated general components of the proposed design theory provides a theoretical foundation for design of implementation of highly flexible and real-time surveillance systems for capital markets. Keywords: Financial Markets, Surveillance Systems, Explanatory Design Theory 1 Introduction Information sources, such as financial blogs and tweets, seduce nonprofessional investors into investing in potentially suspicious financial instruments (SEC, 2012). Many investors struggle with their involvement in faulty investments. For the identification of market abuse, Information Systems (IS) for market surveillance include the detection of notable market abuse patterns in structured data (Eren & Ozsoylev, 2006). However, as of yet, there has been no research that integrates both structured and unstructured user-generated content with the information provided by the regulatory authority in a system that supports financial institutions in their surveillance tasks. The aim of the proposed design theory is to address this research gap by formulating design recommendations for an IS that supports market surveillance decision making. The research presented in this paper is based on a three-year research project that provided the opportunity to develop an IT artefact to detect market manipulation. From October 2010 through September 2013, the market surveillance system was developed, implemented and evaluated in close researcher-practitioner collaboration (“Project FIRST,” 2013). The domain experts and regulatory authorities intervened as needed to align the design theory with their surveillance issues. 127 Irina Alić This paper intends to contribute to the explanatory picture of market surveillance by providing insights into an explanatory design theory for financial markets as well as to support regulatory authority decision making by proposing a design solution for market surveillance. Thus, the study is led by the following research question: What are the general requirements and general components of financial market surveillance systems that are capable to detect market manipulation initiated via social media? This paper is organized as follows: the next section provides a brief research background regarding market manipulation and design theories for IS, followed by proposed study design. Finally, a design theory is described, followed by a conclusion. 2 Research Background 2.1 Market Manipulation Allen and Gale (1992) investigated different manipulation schemes, distinguishing between three groups of manipulation strategies. The first group consists of trade-based manipulations used as strategies for buying and selling that do not result in changes to beneficial interests or market risks. The second group is made up of information-based manipulation strategies, where false and misleading information is published in order to manipulate prices. The third group is made up of action-based manipulation strategies, in which compromising actions are undertaken by the management in order to affect the value of the company. Market manipulation related to the illegal disclosure of untrue information by the sender via unstructured data has been explored (van Bommel, 2003). The “pump and dump” market manipulation is one of the most widespread fraud schemes (Securities and Exchange Commission [SEC], 2012), manipulating share prices by first buying a specific stock and then spreading untrue positive information about the company in order to push share prices to an artificial level. The Internet (e.g., financial news platforms and blog forums) is used to spread the misleading information. Profit is then made by selling the stock at this artificial price level (Aggarwal & Wu, 2006). Affecting the share price of penny stock companies is therefore much easier than of large cap companies whose shares are traded by professional institutional investors. In summary, to detect the various types of market manipulation, a corresponding surveillance system needs to handle traditional data (e.g., time series) as well as the non-traditional data (e.g., news, blogs, and twitter platforms). 2.2 Design Theories for Information Systems Several studies on theory-building approaches in Design Science Research (DSR) have been published in recent years. In (Hevner, March, Park, & Ram, 2004), seven guidelines were proposed to assist design science researchers in both contributing to IS theory and creating and evaluating as-of-yet unknown and innovative information technology (IT) artefacts. Particularly relevant to this study was the recently-developed 128 Financial Market Surveillance Decision Support: An Explanatory Design Theory IS design theory proposed by Baskerville & Pries-Heje, (2010), which was an explanatory model of design IT artefacts. The theory distinguishes between general components and general requirements where the components are justified by the requirements. The explanatory design theory explains why a set of requirements is satisfied by a set of components. Hence, only two essential parts are needed for a complete explanatory design theory: general requirements and general solution components. Nevertheless, an evaluation with the domain experts and regulatory authority will be provided. 3 Study Design 3.1 Action Design Research The general requirements and components of interest are identified and the action design research (ADR) methodology for design science research problems is utilized to merge science and practice (Sein, Henfridsson, Purao, Rossi, & Lindgren, 2011). The method bridges the gap between research and practice (Baskerville & Myers, 2004) and is appropriate for collaborative projects between scientists and practitioners who wish to develop or improve solutions for real practical problems (Marshall, Willson, Salas, & McKay, 2010). Thus, ADR is appropriate for this project because it is expected to provide a solution to a real-world problem while reflecting on lessons learned (i.e., by formalizing the design theory). ADR is by its nature an intervention, in this research, not in a unique organizational setting, but on the European regulatory background where financial authorities face the problem of market abuse and the need to counteract such abuses. In order to satisfy the reliability of this research the findings were steadily counterchecked with practitioner of the project consortium including stakeholders from a European financial supervisory authority. More precisely, during sequentially held consortium meetings, developed IT components where presented and the practitioner provided feedback if they provide a solution to the problem. 3.2 Research Stages In our case, the ADR stages are maintained iteratively in cycles of theory and practice steps (Baskerville & Wood-Harper, 1996) and in close collaboration between the participants, leading to the generation of general requirements and general components that constitute an explanatory design theory. The ADR stages are detailed as follows (Sein et al., 2011): Stage 1: Problem formulation The project task is to develop an IT artefact to detect market manipulation. Thus, the main driver for development is the support of market surveillance tasks via systematic collection and analysis of any data that can be utilized for decision support. The project 129 Irina Alić involved other researchers and practitioners from both the financial and IT domains; the market surveillance project team consisted of 15 partners (five scientists from Slovenia, Spain, and Germany; two practitioners from the German and Italian regulatory authorities; and eight practitioners from Germany and Italy). The role of the researchers was to consider the problem in order to assess the situation from a scientific perspective and contribute accordingly to the knowledge base. The practitioners worked in a market surveillance context as financial domain experts. Communication was maintained over a project-web service platform that contained all project-relevant documents (e.g. models, prototypes, documents). The platform was extensively used by both practitioners and researchers. In this first stage of the project, user needs were identified and problem awareness for a specific goal was generated1. From a theoretical perspective, the literature steam on decision support systems (DSS) was examined, the initial questions to be discussed with regulatory authorities and practitioners were settled on, and possible methods were debated. The first meeting was set for this discussion. Stage 2: Building, intervention, and evaluation (BIE) In these process steps, collaboration between practitioners and scientists was motivated by specifying the activities that should lead to the desired solution for the problem. In doing so, the researchers initiated the first semi-structured interview, which included the following questions: What is to be accessed? What is the decision about? Who is the decision maker? Who is affected by the decision? In several further meetings and telephone conferences, the tacit knowledge regarding how to assess the market abuse driver was explored. The data collected in collaborative meetings was analysed instantly within the team of practitioners, users, and scientists. In each meeting, the initial question served as both a starting point for discussions and a focus point for the resultant discussion on gaining a better understanding of market abuse. The attributes were used to enhance understanding of the phenomenon (Hadasch, Mueller, & Maedche, 2012). Over the course of the project, the entire team met in person several times in annual meetings, each of which lasted three days. In these meetings, development stages were presented, possible improvements and ideas were suggested, and subsequent steps were discussed. The market surveillance team additionally met in person twice a year. In addition, several telephone conferences were conducted. Initially, the system was designed as a prototype qualitative model (Alić, Siering, & Bohanec, 2013) allowing the derivation of initial design principles. The prototype was evaluated in two ways. First as a simulation where artificial data was utilized to simulate and prove the usability of the prototype and second as a verification of whether 1 FIRST Consortium D1.2 Use case requirements specification, http://www.project- first.eu/public_deliverables. 130 Financial Market Surveillance Decision Support: An Explanatory Design Theory the model addressed the problem. In the subsequent phases, further developments were continuously made, discussed, and evaluated, resulting in a final IS. Stage 3: Reflection and learning This was the continuous stage, conducted synchronously with the two first stages. Across all three stages, possible problem solutions were re-conceptualized, ensuring greater generalizability of learning. During the entire project, the permanent involvement of a regulatory authority member and the evaluation phases resulted in the development of general requirements and general components. Stage 4: Formalization of learning The learning was incorporated into the outcome, representing a generalized solution to the problem (Sein et al., 2011). In this stage, nine general requirements and five general components were formulated as the design theory for a market surveillance system. Table 1 presents the summary of ADR cycles in the project. ADR Stages and Principles Outcome Problem Formulation Principle 1: The main driver for this research was the Recognition: Practice- need to support market supervisory Based on recognized inspired authorities in market surveillance tasks. shortcomings the IT research Artefact should Principle 2: General theoretical background related to operate on: Theory- model-driven DSS (Turban, Sharda, & -structured time series ingrained Delen, 2010) data Artefact -unstructured user- generated content data -and information provided by the regulatory authority. Building, Intervention and Evaluation Principle 3: Infrastructure for the retrieval, storage and The prototype was Reciprocal knowledge extraction from social network designed as a shaping was expected to be an ongoing problem. qualitative model. The developed prototypes were steadily counterchecked with the regulatory authority. Principle 4: The role of the researchers was to assess the The prototype was Mutually situation from a scientific perspective. They iteratively developed influential also acted as the artefact developers. and evaluated within roles 131 Irina Alić Principle 5: During the development, the artefact (i.e. the the team resulting in a Authentic instantiated prototype) was continuously final IS (Alić et al., and evaluated within the project team including 2013) concurrent the regulatory authority members. The final evaluation IS was evaluated by the potential end-users from financial institutions. Reflection and learning Principle 6: Constant intervention and evaluation lead to Refined version of the Guided re-conceptualization of possible design design. Emergence components. Formalization of learning Principle 7: Formulation of financial market surveillance A set of general Generalized constituting explanatory design theory: requirements and outcomes interconnection between theory components general components. and goals to apply the knowledge to the problem class. Table 1. ADR Stages based on Sein et al. (2011) 4 An Explanatory Design Theory for Market Surveillance Decision Support This section provides the general descriptions, units of analysis, and requirements in the construction of the desired system as the results of ADR stages. Further, it explains artefact classifications in order to greater conceptualize generalized components. The meaning of the word “requirements” as it is used by (Baskerville & Pries-Heje, 2010), refers to a “condition or capability needed by a user to solve a problem or achieve an objective.” 4.1 General Requirements Through meetings and telephone conferences with the experts, a set of general requirements was established. One of the practitioners pointed out that: “The target users are surveillance staff members who are employed by a regulatory authority.” Other experts on the team highlighted the importance of daily observations: “The surveillance staff members need to prove daily if some bad guys are out there.” As a consequence, the DSS focused on compliance staff members and their daily work activities in the context of market surveillance. DSR on DSS has shown that most systems are designed to support IS practitioners and managerial users as a single user (Arnott & Pervan, 2012). Focusing on classes of systems that support decision-making processes of regulatory authorities, compliance officers in financial institutions can be expected to benefit from market surveillance DSS. The importance of this research is 132 Financial Market Surveillance Decision Support: An Explanatory Design Theory grounded on the nature of financial DSS and the consequences that such newly- introduced methodologies and artefacts can have upon its users. Hence, the general requirements of the system were assumed to be as follows: The task: The market surveillance officers attempt to ensure the proper functioning of capital markets in accordance with the regulation rules (R1). The decision support: The market surveillance officers are supported in their daily efforts to maintain observations of market participants’ abusive behaviours (R2). Compliance offices are not profitable cost centres (Cumming, 2008), so the user needs to ensure market surveillance is as time-efficient as possible in order to reduce costs. As a result, the following requirements are defined: The signalling: If an anomaly occurs, an alert needs to be generated (R3). The surveillance: Monitoring the market and the market’s behaviour implies timely analysis of a large number of financial instruments (R4). The data monitored is primarily structured (e.g., in a time series). Detecting trade-based manipulation by finding suspicious trading patterns in structured data has already been well examined and employed in market surveillance IS (Cao & Ou, 2008). Regarding the one behind the manipulation, one expert states: "A bad guy is engaged in the market, is interested in selling after he buys low, and starts to spreads highly positive news on the social net." The detection of information-based manipulation in recognition of suspicious information published on social media, together with the detection of trade- based manipulation, was therefore mandatory for this research. The systems combine structured data with unstructured social media data to aid in decision making: The data: The ability to deal with heterogeneous data (R5). Regulatory authorities often recommend transparency of adaptive management while emphasizing specific processes (Linkov et al., 2006) such as the detection of suspicious patterns in a historical time series of data, the investigation of the transaction, and the escalation to the regulatory authority if necessary (Buta & Barletta, 1991; Lucas, 1993). According to one project expert, financial institutions have to provide the regulatory authority with “detailed information on every potentially abusive case”. This implies the following general requirements: The rules: Must be comprehensive (R6). Documentation of rules: Alerts need to be processed and stored for investigative purposes (R7). The subsequent general requirements are for the precise detection of abusive cases (suspicious behaviour) and the provision of signals if suspicious behaviour appears: The history: The user must have the ability to prove the background of the case that caused an alert (R8). 133 Irina Alić So as not to overwhelm users with false alarms, as is the case when the rules are too sensitive, the user needs to be authorized to change the rules to a more balanced level. The ability to modify the elements in order to both receive all relevant abusive cases and reduce the appearance of false alerts is further expected with the cost-reducing measures. This implies the final general requirement: Ability to modify the model configurations: The values of the rules can be changed by the user (R9). Hence, the unit of analysis in the proposed research was the financial market surveillance decision support system. This system provides all the relevant information necessary to support the regulatory decision making processes. The requirements were evaluated within the team with the purpose of ensuring design theory generalizability, which applies to the class of surveillance systems instead of an instance (Müller- Wienbergen & Müller, 2011). In addition, the developed solutions were presented to the European Commission by the project leaders, presenting achievements and discussing possible modifications of the solutions. 4.2 General Components General requirements derived from interviews with the practitioners in several cycles provided guidance in order to develop suitable IT solutions. Through abstraction and learning general IT components were identified on this basis. In the following, the abstract architecture of the proposed explanatory design theory is presented. The data sources that will be considered in the market surveillance task are retrieved from the internal sources of the specific organization and from external data sources. The external structured data is usually provided by data vendors via proprietary IS and other delivery systems. The unstructured textual data is collected from the regulatory authority’s web sites. Further unstructured data considered in this project was user- generated content collected from several social networks such as blogs. The regulatory data and user-generated content data have not been fully acknowledged in prior research. Thus, a promising research approach may be achieved by assessing all three of these data sources (regulatory-, vendor- and user-generated content). The value-added components for modern surveillance solutions are: Internal and external data capturing systems (C1) 2. The acquisition of a web data stream can be realized with web APIs, (e.g., Twitter™ API). Such stream-based workflows (up-to-date with the stream) can be built on data mining models, allowing client queries at any time (Saveski & Grcar, 2011). The unstructured data relevant for market surveillance retrieved from external sources, such as blogs, tweets, news web pages, and regulatory web pages, is stored here. This data is 2 FIRST Consortium D3.1 Semantic resources and data acquisition; D3.3 Large-scale ontology reuse and evolution, http://www.project-first.eu/public_deliverables. 134 Financial Market Surveillance Decision Support: An Explanatory Design Theory characterized as highly informative (Zhang & Skiena, 2010), and can be used to assess the investors’ opinions (Klein, Altuntas, Riekert, & Dinev, 2013). The data retrieved from data vendors, such as structured financial time series data, can also be processed and analysed using data mining techniques (Gopal, Marsden, & Vanthienen, 2011). The general requirements for rules 2, 4, and 5 are thus satisfied. Consequently, the component that provides these services can be taken into account is: Data storage and analysis (C2)3. For the huge amounts of unstructured data, techniques for extracting and adapting information from the text are necessary (Park & Song, 2011). Thus the component comprises the preparation of unstructured data for further use in the workflow process. For this purpose, the scientific literature offers two different approaches, namely ontology-based methods and data mining methods (Klein, Altuntas, Häusser, & Kessler, 2011). Ontology is the formal specification of the vocabulary and its relationships in the domain (Gruber, 1993). The data mining method, particularly the text mining method, deals with the transformation of the natural text into numerical vector values (Feldman & Sanger, 2007). For the purpose of sentiment analysis, one of the sophisticated techniques is the 'active learning principle' where the output is represented by the model for sentiment classification, (e.g., positive or negative financial tweets) (Saveski & Grcar, 2011). In order to maintain the time-critical surveillance tasks of compliance officers, the methods for automatic sentiment classification are obligatory, satisfying general requirements 1 and 2. The component is therefore the further value-added component for modern surveillance solutions: Processing of unstructured data (C3)4. The data applied from the data processing unit serves as input to the knowledge repository, allowing the user to assess the data. Furthermore, as the repository meets general requirement 8 by comprising the information from internal databases and further external data sources, it stores all involved data in the alert signal. The data is further utilized by several models and rules and is stored in the repository. The models to which this research refers are quantitative data mining models, qualitative multi- attribute models, and further market surveillance rules that can detect market anomalies or abusive behaviour (“Project FIRST,” 2013). Qualitative multi-attribute models were developed in the interviews with experts and were suitable for the evaluation and analysis of decision alternatives (Bohanec, Žnidaršič, Rajkovič, Bratko,& Zupan, 2013). The data mining models that handle forecasting from large unstructured and structured data sets for the detection of notable or suspicious patterns were also developed. Thus, the following component satisfies general requirements 3, 6, 7, 8, and 9: 3 FIRST Consortium D6.5 Highly Scalable Interactive Visualization of Textual Streams v2, http://www.project-first.eu/public_deliverables. 4 FIRST Consortium D4.3 Large-scale Semantic Information Extraction Components; http://www.project-first.eu/public_deliverables. 135 Irina Alić Glass-box model of the knowledge repository (C4)5. The component should ensure an enhanced understanding of the occurring phenomena and facilitate the decision making processes for the compliance officer. With visualization of the text mining results, along with the qualitative multi-attribute results, the user is deeply involved in the processes of alert generation. The most appropriate visualization can be represented as a decision tree (Liu & Salvendy, 2007) viewed graphically as a set of connected decision nodes and leafs. While the nodes carry the attribute values, the user can use their tacit knowledge regarding pattern recognition and change the attribute values if necessary. This ensures a better understanding of the data samples. This component fulfils general requirements 6, 8, and 9 by employing rule- based methodologies for comprehensibility of rules, vivid representation of the history of occurrence, and ease in rule modification: Graphical user interface (C5)6. For flexibility, financial market surveillance DSS needs to be modular, and the solution can be integrated into existing systems. Table 2 summarizes the contributions this study makes to the scientific knowledge. General Requirements General Components (R1) Proper functioning of capital markets in (C1) Internal and external accordance with the regulation rules. data capturing systems. (R2) The user is supported in his daily efforts to (C2) Data storage and maintain observations of market participants’ abusive analysis. behaviours. (C3) Processing of (R3) If an anomaly occurs, an alert will be generated. unstructured data. (R4) Timely analysis of large number of financial (C4) Glass-box model of instruments. the knowledge repository. (R5) Use of heterogeneous data. (C5) Graphical user (R6) Comprehensive rules. interface. (R7) The rules can be configured by the user. (R8) Storage of alerts for investigative purposes. (R9) The user has the ability to prove the background of the case which caused an alert. Table 2. Design theory for financial market surveillance DSS 5 FIRST Consortium D6.2 Machine Learning and Qualitative Models; http://www.project- first.eu/public_deliverables. 6 FIRST Consortium D2.1 Technical requirements and state-of-the-art; http://www.project- first.eu/public_deliverables. 136 Financial Market Surveillance Decision Support: An Explanatory Design Theory 5 Conclusion The goal of this research on explanatory design theory development was to support decision making for market surveillance enforcement. The approach of theory development was based on the development of an instantiated IT artefact addressing identified user requirements. The emerging qualitative data exploration of semi- structured interviews with team members was carried out with the goal of determining important decision attributes where the exploration was predicated by ADR. Further, an explanatory theory-building method was applied. From a practical perspective, the general requirements and components represent the design theory that provides guidance for the development of market surveillance IS. Furthermore, from the cost perspective, where market surveillance is emphasized as a time consuming cost centre, this study provided insights into the development of more efficient surveillance systems. From a theoretical perspective, this research contributes to the literature on financial market surveillance by enhancing future development strategies of explanatory design theories to solve a class of problems. The theory development approach was based on prescriptive research, and accordingly, it built on the suggestions for development. This research is limited by the fact that it is based on interviews with European domain experts and regulatory authorities. It could be argued that non-European experts have a different point of view of market surveillance. Additionally, this research considers only English articles. Future research could be enhanced by adding non-EU regulatory authorities and by utilizing non-English data sources. To reduce bias during the project phase, the researchers tried to remain in close contact via email, Skype™, team views, and face-to-face meetings with the experts. Even so, there could be limitations in researcher bias due to the fact that the researchers’ goals and those of the expert’s sometimes differed, leading to restriction in generalization. Acknowledgments The presented research program received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013) within the context of the Project FIRST, large scale information extraction and integration infrastructure for supporting financial decision making, under grant agreement no. 257928. The author thanks all of the members of the FIRST project consortium for their contributions to the IT artifact developments described in this paper. References Aggarwal, R., & Wu, G. (2006). Stock Market Manipulations. The Journal of Business, 79(4), 1915– 1953. Alić, I., Siering, M., & Bohanec, M. (2013). Hot Stock or Not? A Qualitative Multi-Attribute Model to Detect Financial Market Manipulation. In 26th Bled eConference eInnovations: Challenges and Impacts for Individuals, Organizations and Society (pp. 1–15). 137 Irina Alić Allen, F., & Gale, D. (1992). Stock-price manipulation. Review of Financial Studies. Arnott, D., & Pervan, G. (2012). Design Science in Decision Support Systems Research: An Assessment using the Hevner, March, Park, and Ram Guidelines. Journal of the Association for Information Systems, 13(11), 923–949. Baskerville, R., & Myers, M. D. (2004). Special issue on action research in information systems: making is research relevant to practice--foreword. Mis Quarterly, 28(3), 329–335. Baskerville, R., & Pries-Heje, J. (2010). Explanatory Design Theory. Business & Information Systems Engineering, 2(5), 271–282. doi:10.1007/s12599-010-0118-4 Baskerville, R., & Wood-Harper, A. T. (1998). Diversity in information systems action research methods. European Journal of Information Systems, 7(2), 90–107. doi:10.1057/palgrave.ejis.3000298 Baskerville, R., & Wood-Harper, T. (1996). A critical perspective on action research as a method for information systems research. Journal of Information Technology, (1996). Bohanec, M., Žnidaršič, M., Rajkovič•, V., Bratko, I., & Zupan, B. (2013). DEX Methodology: Three Decades of Qualitative Multi-Attribute Modeling. Informatica, 37, 49–54. Buta, P., & Barletta, R. (1991). Case-based reasoning for market surveillance. Proceedings First International Conference on Artificial Intelligence Applications on Wall Street, 116–121. doi:10.1109/AIAWS.1991.236563 Cao, L., & Ou, Y. (2008). Market Microstructure Patterns Powering Trading and Surveillance Agents. Journal of Universal Computer Sciences, 14(14), 2288–2308. Cumming, D. (2008). Global Market Surveillance. American Law and Economics Review, 10(2), 454– 506. doi:10.1093/aler/ahn009 Eren, N., & Ozsoylev, H. N. (2006). Hype and dump manipulation. In Eropean Finance Association 2007 Ljubljana. Feldman, R., & Sanger, J. (2007). The text mining handbook: advanced approaches in analyzing unstructured data. Cambridge University Press. Gopal, R., Marsden, J. R., & Vanthienen, J. (2011). Information mining — Reflections on recent advancements and the road ahead in data, text, and media mining. Decision Support Systems, 51(4), 727–731. doi:10.1016/j.dss.2011.01.008 Gruber, T. R. (1993). A Translation Approach to Portable Ontology Specifications by A Translation Approach to Portable Ontology Specifications. Knowledge Acquisition, 5(April). Hadasch, F., Mueller, B., & Maedche, A. (2012). Exploring Antecedent Environmental and Organizational Factors to User-caused Information Leaks: A Qualitative Study. In European Journal on Information Systems (ECIS). Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design Science in Information Systems Research. MIS Quarterly, 28(1), 75–105. Klein, A., Altuntas, O., Häusser, T., & Kessler, W. (2011). Extracting Investor Sentiment from Weblog Texts: A Knowledge-based Approach. In Computing (CEC), 2011. Klein, A., Altuntas, O., Riekert, M., & Dinev, V. (2013). A Combined Approach for Extracting Financial Instrument-Specific Investor Sentiment from Weblogs. Wirtschaftsinformatik, (March), 691–705. Linkov, I., Satterstrom, F. K., Kiker, G., Batchelor, C., Bridges, T., & Ferguson, E. (2006). From comparative risk assessment to multi-criteria decision analysis and adaptive management: recent developments and applications. Environment International, 32(8), 1072–93. doi:10.1016/j.envint.2006.06.013 Liu, Y., & Salvendy, G. (2007). Design and evaluation of visualization support to facilitate decision trees classification. International Journal of Human-Computer Studies, 65(2), 95–110. doi:10.1016/j.ijhcs.2006.07.005 Lucas Jr, H. C. (1993). Market expert surveillance system. Communications of the ACM, 36(12), 27–34. Marshall, P., Willson, P., Salas, K. de, & McKay, J. (2010). Action research in practice: issues and challenges in a financial services case study. Qualitative Report, 15(1), 76–93. 138 Financial Market Surveillance Decision Support: An Explanatory Design Theory Müller-Wienbergen, F., & Müller, O. (2011). Leaving the beaten tracks in creative work–A design theory for systems that support convergent and divergent thinking. Journal of the Association for Information Systems, 12(11), 714–740. Park, B.-K., & Song, I.-Y. (2011). Toward total business intelligence incorporating structured and unstructured data. In 2nd International Workshop on Business intelligence (BEWEB) (pp. 12–19). Project FIRST. (2013). Dissemination Material: http://www.project-first.eu/public_deliverables. Saveski, M., & Grc•ar, M. (2011). Web Services for Stream Mining: A Stream-Based Active Learning Use Case. In European Conference on Machine Learning and Pronciples and Practice of Knowledge discovery in Databases, eCML PKDD 2011 (pp. 1–11). SEC, G. (2012). SEC Suspends Trading in Common Stock of Three Hundred Seventy- Nine Compnies Quoted on OTC May 14 , 2012 (No. No . 66980). SEC (Vol. May). Sein, M. K., Henfridsson, O., Purao, S., Rossi, M., & Lindgren, R. (2011). Action design research. MIS Quarterly, 35(1), 37–56. Turban, E., Sharda, R., & Delen, D. (2010). Decision Support and Business Intelligence Systems. Prentice Hall (9th edition) Van Bommel, J. (2003). Rumors. The Journal of Finance, LVIII(4). Zhang, W., & Skiena, S. (2010). Trading Strategies to Exploit Blog and News Sentiment. In 4th Int’l AAAI Conference on Weblogs and Social Media (ICWSM). 139 BACK 28th Bled eConference #eWellBeing June 7 - 105, 2015; Bled, Slovenia Cloud Oriented Business Process Outsourcing using Business Rule Management Jeroen van Grondelle1 Jeroen van Grondelle Digital Ecosystems, HU University of Applied Sciences Utrecht, The Netherlands jeroen@vangrondelle.com Rudolf Liefers Liefers Consultancy, The Netherlands rudolf@adviesvanliefers.nl Johan Versendaal HU University of Applied Sciences Utrecht, Open University of The Netherlands, The Netherlands johan.versendaal@hu.nl Abstract The development of cloud services is rapidly transforming IT outsourcing. Highly standardized services are offered in elastic ways. Pricing models are shifting away from up front investments, allowing pay per use. In business process outsourcing (BPO), these business models are still less common and BPO services are often still implemented on a per customer basis, often heavily based on customer’s existing practices. This paper presents a framework for using the field of business rule management (BRM) to develop BPO offerings that exhibit cloud properties. The framework specifies ways in which business rules can be used to parameterize the different aspects of a BPO service. The framework is applied in three practical scenarios, which are evaluated for their cloud characteristics by interviewing experts from outsourcing customers, outsourcing suppliers and consultants. Keywords: Business Process Outsourcing, Cloud Computing, Business Rule Management, Declarativity, Governance 1 Introduction Cloud services are transforming the field of IT Outsourcing. Cloud providers offer highly standardized services that can only be personalized to a limited extent. In return, 1 This research is supported by the Dutch Platform for Outsourcing (www.platformoutsourcing.nl) through a research grant. 1 Jeroen van Grondelle, Rudolf Liefers and Johan Versendaal the customer receives highly elastic services that scale with his individual demand and the customer gets to pay based on actual use only (cf. Mell and Grance, 2011). They help shifting cost from capital expenditures (CAPEX) to operational expenditures (OPEX), and even aligning cost with actual demand (Armbrust et al., 2010). The cloud model is still little used in Business Process Outsourcing (BPO) services. Conventional BPO practices often focus on continuing customer’s existing processes and applications, and in the past often also on on-boarding customer’s existing employees. Due to the resulting one-of-a-kind nature, supplier’s employees are assigned to and trained for a specific customer, and can typically not be transparently reassigned without going through some transition process. As a result, pay-per-use may not be feasible, as the costs incurred for each customer are not flexible and temporary overcapacity for one customer can not dynamically be used to service other customers where there is a (temporary) under capacity. The field of BRM (Hay et al., 2000; Ross, 2013) seems a promising field to underpin BPO services that do exhibit such cloud properties. The concept of declarativity, where the criteria that have to be met are specified and the procedural way in which to reach that state are inferred, elegantly meets important aspects of the cloud philosophy. In the cloud, a customer similarly has limited or no influence on how the service is delivered, and can only specify some of the properties the service will have. Therefore, our research question in this paper is: How can concepts, techniques and methods from the field of business rule management be applied to develop business process outsourcing offerings that exhibit cloud properties? BPO offerings with cloud properties could have further practical applications. Cloud oriented BPO offerings could make BPO accessible to smaller organizations, including SME, that might struggle with upfront investments of conventional BPO. Such services could also be part of dynamic sourcing strategies in larger organizations that deal with volatile demand and that use BPO suppliers dynamically to deal with peak demands. This research is performed within the methodical framework of Design Research (Hevner, 2004). Based on explorative interviews with experts, a framework is proposed that helps to develop cloud oriented BPO offerings using existing concepts and techniques from business rule management. A number of concrete scenarios that fit within this framework is proposed, and these scenarios are assessed for feasibility and cloud properties by experts from the field of sourcing. Based on the design science research methodology by Peffers (2007), the remainder of the paper is organized as follows. In the next section, the framework for cloud oriented BPO using BRM is drafted. This framework is exemplified in Section 3 by three potential scenarios. Section 4 reports on an initial evaluation of the feasibility and the exhibition of cloud properties of these scenarios by interviewing experts from the outsourcing community. 2 Framework for Cloud Oriented BPO In order to develop a framework for cloud oriented BPO, we have held explorative interviews with 8 experts from the field of outsourcing. They represent organizations that outsource, supply outsourcing services and consult organizations on their sourcing strategies. 141 Cloud Oriented Business Process Outsourcing using Business Rule Management The interviews were conducted in a lowly structured, explorative way. Topics discussed in every interview include their involvement in outsourcing in general and BPO more specifically, their strategies and practices in governing these sourcing relationships. Also, their appreciation of the cloud trend, and its impact on BPO offerings and governance practices was discussed. Specifically, their perception of possible convergence and/or reduction of the degrees of freedom in service parameterization by the cloud was discussed. From the interviews it is clear that there is a real interest in business process outsourcing models that have low upfront investment, and are elastic in capacity. We have met suppliers that already realize their BPO services in very structured, parameterized ways. Predominantly however, BPO services are implemented on a per customer basis, with their current practices as starting point, realizing little or no elasticity in for example human resource capacity across customers and requiring high initial investments. 2.1 Definition of Cloud Oriented BPO The association with the cloud metaphor frequently introduces misunderstandings in the explorative interviews. For instance, the relative recent term of Business Process as a Service (Cantara and Lheureux, 2013) used in many cloud descriptions today often refers to highly verticalized IT solutions for generic business functions, rather than the broader service concepts in BPO that typically include human resources. Also, the one- size-fits-all association of cloud services for some still contrasts with the kind of customer specific solutions they see delivered in BPO. The cloud metaphor is introduced in our research not to make sure the IT aspects of the BPO offerings are implemented using cloud technology, but to develop BPO offerings that themselves exhibit cloud properties such as high elasticity and pay-as-you-go pricing models. To underpin our framework, we therefore use the term cloud oriented BPO rather than cloud based BPO. We have generalized the NIST definition of Cloud Computing, to include the human actors that are often part of BPO: “Cloud oriented BPO is a model for enabling ubiquitous, convenient, on- demand network access to a shared pool of human resources, and potentially supporting computing resources, that can be rapidly provisioned and released with minimal management effort or service provider interaction.” The five characteristics of cloud services introduced by NIST seem relevant in the BPO domain too. We have generalized them similarly in Table 1. NIST Characteristic Interpretation in BPO Context On Demand Self Service A BPO service that allows customers to specify and change functionality and capacity themselves, and minimal interaction with supplier is needed to implement those. Broad Network Access A BPO service with a high channel independence and location independence, both in terms of customer/supplier interaction as potentially the location of the agents that perform the BPO. Resource Pooling A BPO service, where agent capacity is shared across workloads from different customers, by rapid transfer between teams and ultimately by working on different customer’s cases transparently. 142 Jeroen van Grondelle, Rudolf Liefers and Johan Versendaal Rapid Elasticity A BPO service that scales up and down fast with customer capacity demand, including aligning the human workforce assigned to a customer accordingly. Measured Service A BPO service that is priced based on measured usage. This may include direct labor cost per hour for dedicated agents, but typically requires production related pay (number of customer contacts, price per on-boarded employee, etc.) or based on results (generated revenue or cost reductions). Table 1: NIST Cloud Characteristics generalized towards a BPO Context 2.2 Mapping BRM Concepts and Techniques to the BPO Domain A recurring topic in the interviews and an important challenge when offering cloud oriented BPO services is the parameterization of generic service components into a service that meets the individual requirements and goals of a specific customer. 2.2.1 Methods and Techniques from BRM The field of BRM has developed a number of tools and techniques that make it a candidate to underpin cloud oriented BPO solutions, which is depicted in Figure 1. Structured rule representation techniques like Rulespeak (Ross, 1996) or OMG’s Semantics for Business Vocabulary and Rules (Object Management Group, 2008) can be used to capture the specific requirements of the outsourcing organization, allowing organization to make unique choices within a shared vocabulary chosen upfront. The executable nature of business rules helps combinations of people and machines to service larger numbers of customers, applying the right combination of rules for each individual case. BRM’s rule lifecycle management approaches help organizations to capture, validate, assess impact, enact, evaluate and change those rules, often offering tool support across the lifecycle (Boyer and Mili, 2011). Figure 1: Mapping BRM Techniques to BPO Governance Aspects 143 Cloud Oriented Business Process Outsourcing using Business Rule Management 2.2.2 Leveraging Declarativity On a more conceptual level, BRM can help change the prescriptive specification practices common to the BPO sector. One of the most important concepts in BRM is that of declarativity: organizations are asked to express their choices and requirements in terms of goals and constraints that must be met in the operation, rather than in terms of procedural recipes hów those requirements are to be met (Van Grondelle et al., 2013). This allows the operation leeway when it comes to how to reach goals and stay within constraints at execution time. Rule engines leverage that leeway by automatically computing smart action plans, without sticking to fixed, pre determined procedures. Human experts use that same freedom to really leverage their expertise and deal with exceptional cases (Pesic and Van der Aalst, 2006; Van Grondelle et al., 2013). In BPO, it could allow customers to only express requirements on the aspects that really matter to them. The leeway this introduces could enable suppliers to develop generic services that create economies of scale across customers, while still staying within the individual constraints of its different customers. 2.3 Three Aspects of BPO Parameterization From coding the explorative interview responses, we have identified three service aspects that were repeatedly mentioned to typically require customer specific choices when specifying a BPO service. 2.3.1 Product Characteristics Often, the BPO is part of the delivery of the customer’s products and service to its end- customers. In those cases, the characteristics of those products typically affect the BPO service. Examples are the rules on eligibility to purchase a product or service or the entitlement to certain service when enrolled. In the interviews, pensions and insurance claims where mentioned as examples of this category. 2.3.2 Procedural Aspects When the former aspect was discussed it was observed in interviews that BPO customers have legitimate needs to influence the procedural aspects themselves too. Policies on for instance quality control, compliance or the treatment of high risk cases may require certain steps to be taken, additionally or at specific moments in the process. 2.3.3 Quality of Service Like in IT outsourcing, quality of service (QoS) is an important aspect of a BPO service. In the explorative interviews experts observed that concepts such as availability and responsivity translate into the business process domain well, in the form of for instance activity completion times and success rates. 2.4 Resulting Framework This results in a three dimensional framework that guides how BRM can be applied to develop BPO offerings that exhibit cloud properties. A graphical representation is presented in Figure 2. This framework is operationalized into concrete scenarios in the next section. Jeroen van Grondelle, Rudolf Liefers and Johan Versendaal Figure 2: Three dimensions to applying BRM to develop Cloud oriented BPO Offerings 3 Three Scenarios for Cloud oriented BPO using BRM To be able to evaluate our framework, we have developed three concrete scenarios of BPO using BRM based on our framework. They each are based on existing research and/or actual applications of business rules, although mostly outside the field of BPO. For each scenario, we outline the role distribution between BPO supplier and customer, provide an example, provide background existing practices this scenario is based on, and summarize the analysis how the scenario realizes the NIST cloud properties. Figure 3: Evaluation Coverage of the Framework by the Scenarios The scenarios each illustrate one of the aspects of BPO parameterization of the framework, as is shown in Figure 3. Although mixed scenarios are certainly anticipated within the framework, such scenarios are not included in this study for the sake of clarity when presenting them to BPO experts in evaluation. Also, the current scenarios do not explicitly address how BRM’s lifecycle management concepts can be applied to manage the rules introduced in the scenarios. 3.1 Scenario 1: Parameterization of Product Characteristics using Decisions Often, product related decisions guide to a large degree the operational processes in an organization. For instance, when processing insurance claims, or HR or grant applications, the decision who is entitled to what is a key aspect of the different operational processes. They prescribe much of the processing of the applications themselves, but also highly influence online self services or the answering of customer Cloud Oriented Business Process Outsourcing using Business Rule Management questions in a call centre. Many sales or marketing processes are guided by best next action type of decisions related to these products too. An emerging practice in the field of business process management (BPM) is the identification of decisions within business processes, and instead of modeling them within the business process itself using process metaphors, isolating them from the business process and using some rule formalism to model them instead. This has led to the rise of the field of Enterprise or Operational Decision Management (Taylor, 2011). A number of tools for specification and execution of decision models is available commercially. A number of (open) standards like the decision model (Von Halle, 2009) and OMG’s Decision Modeling Notation (Object Management Group, 2014) have been developed. Standardization of how to integrate decision models into business process models modeled in Business Process Management Notation (Object Management Group, 2008) is underway. 3.1.1 BPO Scenario In this scenario, the BPO provider develops a generic business process, in which a number of decisions is taken that reflect the characteristics of the product. The BPO customer gets to specify the rules based on which these decisions are taken for the service delivered to his end-customers. Figure 4: A Generic Process, parameterized with Decision Rules 3.1.2 Example A white label pension provider may have the agents and systems in place to administer the pension claims of the members of multiple funds. He may have the processes in place for enrolling in the fund, dealing with changing employment and reaching pension age. Pension funds that delegate their administration to this provider get to specify the rules based on which membership to the fund is granted, contributions are calculated and entitlements are established. 3.1.3 Mapping to the NIST properties This model potentially exhibits a number of the NIST properties of cloud services. As the business process itself is essentially left unchanged, and only some guiding Jeroen van Grondelle, Rudolf Liefers and Johan Versendaal decisions within the process are redefined on a per-customer basis, it enables a high degree of self-provisioning. Also, sharing agents across customer cases is quite feasible, resulting in rapid elasticity, as the agents essentially act in familiar roles within a familiar process. It is important to have the customers specify the decisions in a formalism that communicates well for people too. In case of questions, or when handling exceptions, the agents will need to understand the specific decision rules used in this instance and be able to explain them to customers. Finally, in terms of offering a measured service, the variable decisions often represent classes of cases that affect handling costs, but of which the mix can be part of the specification and pricing. In the example in Figure 4, a BPO provider could for instance offer different prices for high risk and low risk cases, or offer a single price, under the condition that no more than 5% of cases qualify as high risk. 3.2 Scenario 2: Specify Procedural Aspects using Declarative Rules Within the field of BPM, there is a class of declarative process modeling formalisms that focus on capturing the constraints a process flow must meet, rather than prescribing the flow itself (Pesic and Van Der Aalst, 2006; Goedertier et al., 2007; Van Grondelle and Gülpers, 2011). They typically provide rule-oriented formalisms that express criteria when activities may and/or must be performed. Typically, the resulting execution flows are only inferred at execution time. 3.2.1 BPO Scenario In this scenario, the BPO supplier offers his process offering in terms of a set of generically useful activities or tasks he can provide for his customer’s end-customers as is shown in Figure 5. The BPO customer get to specify the rules that determine which activity is performed in which end-customer case, and potentially in which order. Figure 5: Composing a Customer-specific Business Process 3.2.2 Example A medical supplier may have all activities (process fragments) in place to support the application for and delivery of medical aid material for customers on behalf of insurers or public agencies. His customers may specify the rules in which cases a complete, Cloud Oriented Business Process Outsourcing using Business Rule Management formal eligibility check needs to be performed and in which cases medical aids are supplied on request and the formal paperwork is processed afterwards. Additionally, he may provide rules for different regimes of checking applications depending on the value and fraud risk of specific cases. 3.2.3 Mapping to the NIST properties Often, the standardized activities can be executed by either machines or agents in ways that exhibit some of the NIST properties. Generic application services can be developed to support certain activities, irrespective of the fact that they are performed only in some cases and in different orders. Generic human tasks, similarly, can often be assigned to agents by task type, without the reason why it needs to be performed being relevant to the agent. Delivering such a variable process as a measured service benefits from a set of generic activities and tasks that is used across customers. Billing per activity performed is possible, but does not lead to predictable costs. Integrated pricing can be based on historical averages, and also on simulations on the rules as they are drafted by/for a new customer. 3.3 Scenario 3: Delivering “the Same” Service at Different QoS In IT clouds, the field of scheduling is applied to match supply and demand and deal with mismatches. A well studied problem is scheduling the order in which a sequence of jobs need to be processed by a machine to minimize the number of late jobs or the total tardiness based on a due time per job. In a weighted variant of this problem, a weight is assigned to each job encoding the priority of the job. The scheduling problem than tries to minimize the total weighted tardiness across the jobs, favoring the high priority jobs to some extent to prevent their high contribution to the average tardiness. The parameterization of SLA’s into job due times and priority weights is a strong oversimplification of the types of constructs agreed upon in SLA’s. An approach to manage SLA’s based on knowledge representation has been developed by Paschke and Bichler. They are able to encode large sets of real worlds SLA’s and monitor whether they are met at runtime. A technique that encodes SLA of intermediate complexity, but that does at runtime maximize revenue by (re) assigning jobs to resources based on SLA reasoning is developed by Macias et al.. 3.3.1 BPO Scenario The final scenario, it is not the functionality (product characteristics) or behavior (procedural aspects) of the service provided, but the quality of service it is delivered at. The BPO provider may offer a generic (or otherwise parameterized) service, and offer it at different qualities of service. The BPO consumer may subscribe to this service and express the quality of service he requires. The performance criteria agreed on could for instance include response times, numbers of transactions delivered or success rate, and these criteria could or could not have financial consequences in terms of bonus/malus or performance related pay. In cases of capacity shortage in delivering the service to its different customers, the BPO supplier Jeroen van Grondelle, Rudolf Liefers and Johan Versendaal will typically weigh the consequences of the different ways to assign his resources to the cross customer workload. 3.3.2 Example A call center may perform outbound calls to customer who have made a purchase on the fifth day after the sale was made. One customer may demand that the call is made on that precise day and pay accordingly. Another customer may agree on calling between day 4 and 7, and negotiate a lower price in return. 3.3.3 Mapping to the NIST properties This scenario exhibits a number of the NIST properties. As the planning occurs across customers, and the service itself is not very variable across customers, a high elasticity in terms of agents can be reached. This is combined with a very individual regime for the QoS on a per customer basis. 4 Evaluation In the initial evaluation, the framework and its operationalization into the scenarios have been presented to 6 experts in the field of sourcing, and in semi-structured interviews they were questioned about both relevance and rigor aspects of the results. 4.1 Setup In all interviews, the framework and the scenarios and examples were presented using the definition, the example and visual explanation present in this paper. Depending on the background of the expert, the business rule management concepts were explained in more depth interactively where needed. After that, a number of questions were asked based on an interview guide prepared upfront. Each expert was interviewed about four aspects: 1) what their professional role is in BPO sourcing situations, 2) to what extent they agree with the challenges identified earlier in the research, 3) what their first response is on the presented mapping between the field of BRM and the field of sourcing governance, and 4) whether the scenarios seem realistic and feasible, and would address the challenges by exhibiting cloud properties. Topics 1 and 2 were discussed using structured questions, except for their role, for which no limitative list was used in this stage of the research. Their agreement with the challenges for BPO was asked using a 5-point Likert scale. On topic 3, the mapping between BRM and sourcing governance, the experts were asked for a brief reflection, of which the interviewer took notes. Topic 4 was discussed using a fixed set of questions, asked for each of the three scenarios. These were mainly Likert scale type of questions, apart for the open questions what the most important impact on governance of the sourcing relation might be if the scenarios were to be adopted. The questions on agreement with the challenges were asked to validate relevance of the research, in the practical sense used in the design science framework introduced by Hevner (2004). The feasibility and extent to which the challenges were addressed in the scenarios were asked to validate whether the developed artifact and its operationalization answered our research question. Cloud Oriented Business Process Outsourcing using Business Rule Management 4.2 Observations In this phase of the evaluation, the one to one and a half hour interviews were conducted with 1 outsourcing customer, 4 outsourcing suppliers and 1 consultant. Although the number of interviewed experts is far too low to draw any firm, quantitative conclusions, a number of interesting observations can be made from the interviews. 4.2.1 Relevance of Cloud for the BPO market The participants were asked their agreement with the statements in Table 2, with their agreement expressed using a 5-point Likert scale. Completely Partly Completely Agree Agree Neutral Partly Disagree Disagree 1. Bespoke BPO is challenged by Cloud/Utility 3 3 trends 2. Bespoke BPO only feasible for large 2 4 organizations 3. A market for highly standardized BPO is emerging to support SME and independent 2 3 1 professionals 4. In many outsourcing situations over- specification by the customer leads to under- 1 5 leveraging the expertise of the supplier 5. When outsourcing multiple services to different cloud providers, governance can become 2 2 1 unmanageable2 Table 2: Participant’s Agreement with the Statements on Cloud and BPO The participants agreed, in part or completely, with all five statements. One participant indicated that the SME sector was unknown to him and scored the associated statement neutral. Another participant has concrete ideas and experiences how agile methods can work when integrating cloud services, and scored disagreement in part to statement 5. On that same statement, two respondents mentioned the increasing importance and effect of open standards in reducing the impact of integration. One also mentioned that the governance challenge may be lower for BPO cloud services, as their verticalized scope will typically lead to reduced number of “touch points” and less horizontal, technical integration. 4.2.2 Mapping of BRM to Outsourcing Governance In the midsection of the interview, our conceptual framework for developing BPO with cloud properties using BRM was presented to the respondents. The feature mapping of BRM and outsourcing governance, as depicted in Figure 1 in this paper, was discussed. It was met with recognition, yet at different levels, mainly due to different levels of familiarity with the field of BRM. Most respondents representing suppliers could mention similar parameterization aspects in their respective application stacks. One respondent made a link to the concept of cognitive computing, which for him was a linking pin to enabling automated agents. 2 Only answered by 5 out of 6 participants 150 Jeroen van Grondelle, Rudolf Liefers and Johan Versendaal The second concept presented was that of declarativity. The notion that customers only express the requirements and goals that need to be met, and not the procedural steps how to meet them triggered a lot of response. Most suppliers expressed in some form the experience that only after complying with the initial, overspecified specification they increasingly gained the trust to propose improvements or alternative approaches. One respondent saw a distinction between outsourcing for cost reduction purposes and outsourcing to leave current practices altogether. His example involved banks that to some degree have to reinvent themselves. In the process, they outsource big aspects of their work to reduce legacy in business processes and technology, and are inclined to steer on results and award autonomy to the supplier in reaching that goal. 4.2.3 Assessment of the Scenarios Finally, the operationalization of the artifact was presented in the form of the three scenarios in this paper. Already Unseen, but Hard to Done Feasible Imagine Parameterization of Product Characteristics using Decisions 6 Specify Procedural Aspects using Declarative Rules 2 1 3 Delivering “the Same” Service at Different QoS 4 2 Table 3: Participant’s Assessment of the Scenarios As is seen in Table 3, Scenario 1 was recognized by all participants, and often implemented in rather structured ways using for instance service templates. Scenario 2 was rather abstract, according to 3 participants, and as such it was hard to imagine what such a scenario in their field of business would look like. On the other hand, authorization rules were mentioned by two participants as a means to in- or exclude certain steps for certain customers or cases. Scenario 3 was recognized by 4 respondents, in the sense that they had quality of service related agreements with their customers, and that these agreements reflected on the service fulfillment. The examples they provided could well be supported by scenario 3, but were typically at this time not part of an integral, dynamic parameterization of their service. One of the participants described a practice where an external team monitors performance with respect to SLA’s from the different teams, and gives the team leads queues to reconfigure when KPI’s are at risk. That process resonates very well with scenario 3. Reconfiguration was however done within the customer specific team, for instance by switching team members from email team to the telephone team. Two respondents mentioned with respect to scenario 3 the risk that also occurs in for instance ITIL prioritization of tickets: Prio’s 1 and 2 are processed, but tickets of lower priority end up in a reservoir of unprocessed requests. One participant outlined how scenario 3 could be used for temporary dealing with under capacity, but would lead to similar unprocessed work in case of structural under capacity. Another participant described how this scenario would support mixed workloads, where low margin work with flexible QoS constraints could create volume to support the high margin work with high QoS constraints. 151 Cloud Oriented Business Process Outsourcing using Business Rule Management 4.2.4 Additional Observations A number of interesting observations were made by the respondents, independent of the direct interview topics. First, it was observed that in those cases where BPO offerings are in fact based on generic templates or reference models, that genericity is typically kept under the hood. Customers are still asked for their unique requirements, which are then internally fit to the generic models and templates. This genericity is not used to qualify as an expert or as a means to guarantee predictable results ort economies of scale. Also, many respondents mentioned initiatives to reach elasticity of different kinds, but almost without exception that elasticity is limited to single accounts when human agents are involved. Training agents for specific accounts is still mentioned often as the main reason for this. In one interview, a direct reference was made to the previous observation: As the genericity is under the hood, agents could in principle be expected to work across accounts, but the local terminology and implicitness of the underlying models prevents fast transferal of agents between accounts. Finally, a more cultural dimension was mentioned repeatedly in both explorative as evaluation interviews. Outsourcing a process, especially for the first time, was said to require a degree of ‘letting go’. Contracts based on results and performance left some customers with a sense of a lack of control. Not much input is needed of them in daily operations, as the BPO supplier acts autonomously within the agreed parameters. One respondent shared anecdotes on how as a service level manager he would detect this happing when customers requested extra site visits or additional bespoke reporting. 5 Conclusions and Future Work Our research question in this paper is how concepts, techniques and methods from the field of business rule management can be applied to develop business process outsourcing offerings that exhibit cloud properties. Based on the the challenges identified in the explorative interviews, and the measured agreement with those challenges in the evaluative interviews, we believe that there is a real interest in the development of such cloud oriented BPO offerings. BRM’s techniques and methods map well to the different aspects of BPO governance of specifying a service definition, monitoring and reporting on it at execution time, and having a methodical framework how to deal with improvements and external change. Furthermore, BRM’s foundation in the concept of declarativity could help establish practices where BPO customers only express their requirements and goal, and that way leave freedom to their BPO suppliers to leverage their expertise in meeting those. This concept resonated extremely well with the interviewed experts in evaluation. The three parameterization aspects of BPO identified, map well to concrete applications of business rules in other sectors. The BPO scenarios in which we translated those use cases to the field of BPO have been evaluated by initial, qualitative interviews. The first is widely recognized by the experts, although currently often not methodically founded in a comprehensive field such as BRM. The other two are either deemed imaginable, or recognized as the conceptualization of small, local initiatives already applied. 152 Jeroen van Grondelle, Rudolf Liefers and Johan Versendaal We are in the process of conducting wider, more thorough evaluation amongst the outsourcing community in the Netherlands. In addition, we are interested whether governance challenges observed in cloud IT, such as the emergence of shadow IT and increased coordination and integration efforts between individually purchased cloud offerings, may transpose to cloud oriented BPO as well. If so, additional research is needed to establish whether the declarative approach used in our study might help mitigate these. Acknowledgements The authors gratefully acknowledge the support of this research by the Dutch Platform for Outsourcing (www.platformoutsourcing.nl) through a research grant. Also, the authors wish to thank the experts and organizations that participated in the interviews. Literature Armbrust, M., Stoica, I., Zaharia, M., Fox, A., Griffith, R., Joseph, A. D., … Rabkin, A. (2010). A view of cloud computing. Communications of the ACM, 53, 50. Boyer, J., & Mili, H. (2011). Agile Business Rule Development. Springer. Cantara, M., & Benoit J. Lheureux. (n.d.). SaaS, BPaaS and CSB Need Three Capabilities to Innovate or Transform the Business. Gartner. Goedertier, S., Haesen, R., & Vanthienen, J. (2007). EM-BrA2CE v0. 1: A vocabulary and execution model for declarative business process modeling. Halle, B. van, & Goldberg, L. (2009). The Decision Model. CRC Press. Hay, D., & Healy, K. A. (2000). Defining business rules: what are they really? . Final Report. Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design Science in Information Systems Research. MIS Quarterly, 28(1), 75–105. Macías, M., Fitó, J. O., & Guitart, J. (2010). Rule-based SLA management for revenue maximisation in Cloud Computing Markets. In Proceedings of CNSM 2010. Mell, P., & Grance, T. (2011). The NIST Definition of Cloud Computing. National Institute of Standards and Technology, Information Technology Laboratory (Vol. 145). Object Management Group. (2008a). Business Process Model and Notation (BPMN). http://www.omg.org/spec/BPMN/ Object Management Group. (2008b). Semantics for Business Vocabulary and Rules (SBVR). http://www.omg.org/spec/SBVR/ Object Management Group. (2014). Decision Model and Notation (DMN). http://www.omg.org/spec/DMN/ Paschke, A., & Bichler, M. (2008). Knowledge representation concepts for automated SLA management. Decision Support Systems, 46, 187–205. Peffers, K., Tuunanen, T., Rothenberger, M. A., & Chatterjee, S. (2007). A Design Science Research Methodology for Information Systems Research. Journal of Management Information Systems, 24(3), 45–77. 153 Cloud Oriented Business Process Outsourcing using Business Rule Management Pesic, M., & Van Der Aalst, W. M. P. (2006). A Declarative Approach for Flexible Business Processes Management. In J. Eder & S. Dustdar (Eds.), Business Process Management Workshops (Vol. 4103, pp. 169–180). Springer. Ross, R. G. (1996). Rulespeak. http://www.rulespeak.com/. Ross, R. G. (2003). Principles of the business rule approach. Boston: Addison-Wesley. Taylor, J. (2011). Decision Management Systems. IBM Press. Van Grondelle, J., & Gülpers, M. (2011). Specifying Flexible Business Processes using Pre and Post Conditions. In P. Johannesson, J. Krogstie, & A. L. Opdahl (Eds.), Practice of Enterprise Modeling (Vol. 92, pp. 38–51). Springer. Van Grondelle, J., Zoet, M., & Vermeer, F. (2013). Characterizing Declarativity across Business Process Formalisms. In Proceedings of IIMA 2013. Zoet, M., Versendaal, J., Ravesteyn, P., & Welke, R. (2011). Alignment of Business Process Management and Business Rules. In European Conference on Information Systems (ECIS). 154 28th Bled eConference #WellBeing June 7 - 10, 2015; Bled, Slovenia A Classification of Modification Categories for Business Rules Martijn Zoet HU University of Applied Sciences Utrecht, Nijenoord 1, 3552 AS Utrecht, Netherlands, martijn.zoet@hu.nl Koen Smit HU University of Applied Sciences Utrecht, Nijenoord 1, 3552 AS Utrecht, Netherlands, koen.smit@hu.nl; Utrecht University Graduate School of Natural Sciences, Princetonplein 5, 3584 CC Utrecht, The Netherlands, k.smit@students.uu.nl Sam Leewis HU University of Applied Sciences Utrecht, Nijenoord 1, 3552 AS Utrecht, Netherlands, sam.leeuwis@hu.nl Abstract Business rules play a critical role in an organization’s daily activities. With the increased use of business rules (solutions) the interest in modelling guidelines that address the manageability of business rules has increased as well. However, current research on modelling guidelines is mainly based on a theoretical view of modifications that can occur to a business rule set. Research on actual modifications that occur in practice is limited. The goal of this study is to identify modifications that can occur to a business rule set and underlying business rules. To accomplish this goal we conducted a grounded theory study on 229 rules set, as applied from March 2006 till June 2014, by the National Health Service. In total 3495 modifications have been analysed from which we defined eleven modification categories that can occur to a business rule set. The classification provides a framework for the analysis and design of business rules management architectures. Keywords: Business Rules Management, Business Rules Modifications, Business Rule Architectures, Change Management. 155 Zoet, Smit, and Leewis 1 Introduction Laws, regulation, protocols, standards, are each example of rules that organizations are forced to act in accordance with (Shao and Pound 1999; Bajec and Krisper 2005; Tarantino, 2008). Each of the previous mentioned form of rules is applied to guide/constrain entities, such as individuals, teams and organizations to act in accordance with internal or external provided criteria. Take, for example, a general practice. From a regulatory and legislative point of view, business rules are used to restrict access to patient information, force general practitioners to be more transparent in their decision- making and constrain the incentive system general practices can apply (Blomgren and Sunden, 2008; King and Green, 2012). In addition to externally provided criteria, organizations themselves also create additional rules, which they want teams and individuals to comply to. For example a general practitioner states rules on how a specific decision must be made. To prevent individuals and teams in an organization deviating from desired behaviour, laws regulation, protocols and standards are translated to business rules. A business rule is (Morgan 2002): “a statement that defines or constrains some aspect of the business intending to assert business structure or to control the behaviour of the business. ” In addition to faster changing and increased amounts of laws, regulation, protocols and standards implemented, trends like higher demanding customers and, faster changing customer’s demands give rise to an increase in the amount of business rules as well as an increased pace of modifications to these business rules. Thereby increasing the need to decompose and structure business rules to accommodate for expected or unexpected modifications and making it possible to rapidly modify them when necessary. Scientific research with respect to business rules decomposition and structuring to address modifiability in terms of anomalies such as insertion, updates and deletion is scarce (Vanthienen and Snoeck 1993; Von Halle and Goldberg, 2010; Anonymous et al., 2012). Current research that is conducted mostly applies experimental research methods and applies theoretical modifications that can occur to a business rule set. This paper extends understanding of business rules modification by addressing the type of modifications that can occur to a business rule (set). Dissimilar to previous research we do not approach this from a theoretical point of view, but analyse eight years of actual modifications to a business rule set. Within this scope, the research question addressed is: “Which modifications can impact a business rule set?” Answering this question will help practitioners better manage business rules that support analytical activities in business processes. The remainder of this paper proceeds as follows. The next section provides a context by describing business rules, separation of concerns, and theory on modification that can occur to business rules. The third section describes the data collection and data analysis. Section four presents the analysis and results of the grounded theory study. The final section summarizes the study’s core findings, contributions as well as its limitations. 156 Classification of Modification Categories for Business Rules 2 Literature Evolution of information systems is characterized by functional or non-functional modifications that occur to the information system. Modifications are necessary because of changes in 1) the operating environment, 2) the implementation technology, and/or 3) in stakeholder needs. In this work, we adopt the concept of modifiability as defined by Bass et al, (2012): “The ability to incorporate anomalies to an information system made possible by the minimal number of changes.” An information system cannot be engineered to adept to every possible modification. Qumer and Henderson (2006, p3) state that a system must be able to accommodate “changes rapidly, following the shortest time span, using economical, simple and quality instruments in a dynamic environment and applying updated prior knowledge and experience to learn from the internal and external environment.” From a technical and economic perspective it is impossible to build a system that can cope with every modification possible. To increase the number of modifications an information system can cope with, multiple design principles have been proposed and validated. One important principle in information systems and computer sciences which enables organizations to manage change is separation of concerns (Versendaal, 1991, Van der Aalst, 1996, Weske, 2007). The advantages of applying the separation of concerns principle are simplified development and simplified maintenance. Development and maintenance are simplified because concerns are separated and therefore can be modified independently of each other without having to know the other concern’s details. Although several variants of separation of concerns have been proposed, various authors agree on a general evolution of information technology architecture which is depicted in Figure 1. This general evolution follows the decoupling of operating systems from applications, database from applications, the user interface from the application and in the 90’s the workflow from the application. With each of the concerns separated, research streams started to focus on modifications within the individual concerns answering questions like: “which modifications can occur to a database?” , “how to cope with change to databases? ”, “which modifications can occur to user interface?” , and “which modifications can occur to workflows? ” In the workflow (Business Process Management) community this research has led to the classification of different type of business processes, e.g. workflow processes, adaptive case management and, straight through processes. Based on the change behaviour of the process a different design paradigm is applied to design and execute the business process. For example a process which is highly structured applies workflow management while a process which is late-structured applies adaptive case management (Van der Aalst, 1996). This example illustrates that organizations need to make a decision on what set of anticipated modifications should be defined to cope with to be able to utilize a stable product and/or service (Mannaert and Verelst, 2009). 157 Zoet, Smit, and Leewis Figure 1: Evolution of Information Technology Architecture (Van der Aalst, 1996) The next wave of separation followed around the 2000’s where research and practice started to propose the separation of business rules from the application and create a separate layer (Chapin et al, 2001; Boyer and Mili, 2011, Graham, 2006). Chapin et al. (2001) states that among the other concerns (application, databases, user interface and workflow) business rule modifications are the most frequent and have the highest impact on software and business processes. Additionally, the authors identified that the other concerns rely extensively on the support of business rules and that modifications to business rules are commonly the most significant in terms of effort required, thereby indicating the need to properly manage modifications to business rules. Scientific research with respect to business rules modeling guidelines that address manageability in terms of anomalies such as insertion, updates and deletion is scarce (Vanthienen and Snoeck 1993; Zoet et al. 2011). Some research regarding this subject can be identified in the knowledge management community (e.g. Vanthienen and Snoeck 1993), the business rules management community (e.g. Zoet et al, 2011) and the software engineering community (Chapin et al, 2001). Chapin et al. (2001) proposes that modifications to business rules are either 1) Reductive, 2) Corrective, or 3) Enhancive of nature. The first modification archetype, Reductive, comprises reducing the business logic implemented. The second modification archetype, Corrective, comprises refinement and making more specific of implemented business rules. The third modification archetype, Enhancive, comprises changing and adding upon the repertoire of software implemented business rules to enlarge or extend their scope. Although Chapin et al. (2001) proposes a theoretical set of modification archetypes they do not elaborate in detail how they affect business rules and how to manage / design business rules in such a way that one can cope with change. Vanthienen and Snoeck (1993) propose in their study, based on relational theory and database normalization, guidelines to factor knowledge thereby improving maintainability. VanThienen and Snoeck’s (1993) research showed that normalization has a positive effect on the average number of business rules affected when anomalies occur. Thus, when anomalies such as updates, inserts and deletes occur, the number of business rules affected in third normal form is less than the number of rules affected in first normal form. However, their research is based on decision tables instead of business rules in general. Building on the work of VanThienen and Snoeck (1993), Zoet et al. (2011) developed a normalization procedure based on representational difference analysis of existing business rules 158 Classification of Modification Categories for Business Rules modelling languages, relational theory and database normalization. The procedure consists of three steps: 1) apply first normalization form, 2) apply second normalization form and 3) apply third normalization form. This research strengthens the conclusions drawn by VanThienen and Snoeck (1993) that normalization has a positive effect on the average number of business rules affected when anomalies occur. A contribution from practice which has the same focus is The Decision Model (Goldberg, 2010). Von Halle and Goldberg's (2010) normalization procedure also is based on the ideas proposed by VanThienen and Snoeck (1993), showing similarities with the solution proposed by Zoet et al. (2011). An important difference between the method proposed by Von Halle and Goldberg (2010) and Zoet et al. (2011) is that the latter supports multiple business rules formalism like decision tables, event condition action languages while Von Halle and Goldborg (2010) focus only on decision tables. Previous research provides conceptual and theoretical understanding of modifications that can occur to business rules. However, these studies applied controlled experiments based on small case studies and/or theorized modifications that can occur to business rules. Thereby focusing on generalization from construct or theory to collected data and generalization from theory to theory (Lee and Baskerville, 2003). We feel that this represents a notable gap, and we argue that there is a need to generalize from collected data to constructs and theory. Differently stated, collecting modifications which occurred to business rule (sets) and generalize this to a theoretical framework. A research method to generalize from data to constructs and theory is grounded theory (Glaser, 1978), which therefore will be adapted for this research. 3 Data collection and analysis The goal of this research is to identify and define the most common set of anticipated modifications (Manneart and Verelst, 2009) that impact the design of a business rule set. To accomplish this goal a research approach is needed that can: 1) identify modifications applied to the business rule and 2) identify similarities and dissimilarities between types of modifications. An additional criterion is that the set of anticipated modifications is grounded in practice. Each of these goals are realized when applying grounded theory. The purpose of grounded theory is to (Glaser, 1978): “explain with the fewest possible concepts, and with the greatest possible scope, as much variation as possible in the behaviour and problem under study.” Theory states that the first selection of respondents and documentation is based on the phenomenon studied at a group of individuals, organization, information technology, or community that best represents this phenomenon (Glaser, 1978). Our choice for a case was based on theoretical and pragmatic criteria. Our theoretical criterion was: “the case site should deal with business rules, regulation, laws or policies that change frequently.” Our pragmatic criterion was: “the case site should have kept different versions of the business rules, regulation, laws or policies.” Based on these criteria the British National Health Service (NHS) was selected. 159 Zoet, Smit, and Leewis 3.1 Data Collection The NHS is built up from four different health care systems, England, Northern Ireland, Scotland, and Wales. These regions combined provide healthcare services for over 64.1 million UK residents. The NHS employs more than 1.6 million people, which makes it one of the top five workforces in the world in terms of scale. Over one million patients every 36 hours make use of NHS services. A significant part of healthcare management in the UK by the NHS focuses on the management of chronic diseases. In April 2004 the NHS introduced the Quality and Outcomes Framework (QOF) as part of the new General Medical Services (GMS) contract. The QOF is a Pay-for-Performance-scheme covering a range of clinical, organizational, and patient areas in primary care. It is established to reward practices for the provision of high quality care and helps fund further improvements in the delivery of clinical care. The QOF includes the measurement of different domains, however, due to the scope of this study only the clinical and public health domains are considered. The NHS manages the QOF which is a Pay-for-Performance-scheme in that comprises to 25 clinical conditions. For each individual condition they create business rules to select when a clinic must be paid for the treatment of the patient (Gilliam and Siriwardena, 2011). The business rule sets are updated twice a year to accommodate the introduction of new insights revealed by empirical research and/or changes in law and regulations. At the time of writing, the combination of these domains contain 25 clinical conditions, with a large amount of underlying indicators, which make up for 80 percent of the commonly encountered health issues in primary care (Gilliam and Siriwardena, 2011). Examples of clinical conditions as part of the QOF are: Heart Failure (HF), Diabetes Mellitus (DM), and Chronic Obstructive Pulmonary Disease (COPD). Of the 25 clinical conditions, 16 have been analysed. The selection of the 16 clinical diseases has been done semi-randomly. First we selected the two clinical conditions with the largest set of business rules: Coronary heart disease and Diabetes Mellitus. After which fourteen additional diseases have been randomly selected: Chronic Obstructive Pulmonary Disease (COPD), Cancer, Asthma, Obesity, Atrial Fibrillation, Chronic Kidney Disease (CKD), Cardiovascular Disease (CVD), Blood Pressure, Contraception, Osteoporosis, Peripheral Arterial Disease (PAD), Cervical Screening, Cytology, and Dementia. For each disease the different versions of the business rules have been collected. At the time of writing the QOF is at version 29. However, version 1 till 8 and 20 cannot be retrieved, not even by the NHS itself. Therefore our analysis included versions 9 till 19 and 21 till 29. In total, the data collected comprises 229 versions (documents) of clinical conditions, from which the publication ranges from March 2006 until June 2014. In total, 16 out of 25 clinical conditions have been fully coded. 3.2 Data Analysis The goal of the first phase of coding (open coding) was to establish a coding scheme. To develop the coding scheme, first, each individual researcher read and coded two consecutive versions of a randomly selected clinical condition. In open coding the unit 160 Classification of Modification Categories for Business Rules of analysis are business rule sets and individual business rules (Boyatizs, 1998). For examples of open coding in our study see Table 1. After both researchers finished, the coded parts were discussed and compared to understand the process and agree on the elements that had to be coded. The result of this first cycle was a coding scheme. The goal of the second cycle of coding was to refine the coding scheme. Therefore two researchers, one researcher from the first cycle and one new researcher, coded multiple consecutive versions of multiple clinical conditions. The clinical conditions were randomly selected from the pool of clinical conditions. After both researchers finished, the coded parts were discussed among the three researchers, including the researchers from the first round. In these sessions coding was compared to understand the process and agree on the elements that had to be coded. The result of this second cycle was an improved coding scheme. The goal of the third cycle was to code the remainder of the 229 versions of clinical conditions and identify the modifications. This cycle was performed by two researchers. The third researcher acted as reliability coder which randomly selected modifications and compared his coding to those of the other two researchers. An extract of the coding scheme is shown in first row of Figure 2. Open coding resulted in 3495 references classified to eleven modification categories: A) create decision, B) delete decision, C) update decision, D) create business rule, E) delete business rule, F) create condition, G) delete condition, H) update condition, I) create fact value, J) delete fact value, and K) update fact value. An overview of all modifications per modification category is provided in Figure 2. Table 1: Examples of open coding: clinical condition COPD (Health and Social Care Information Centre, 2007) Text Fragments Version A Text Fragments Version B Open Coding Clinical indicator COPD8 Clinical indicator COPD13 Update decision If COPDSPIR_DAT >= (COPD_DAT – 3 months) AND If COPDSPIR_DAT >= Delete business rule If COPDSPIR_DAT <= (COPD_DAT (COPD_DAT – 3 months) + 12 months) Read codes v2: (8I2M., 8I3b., 8I6L., 8I6d.) Read codes v2: (8I2M., 8I3b., 8I6L.) SNOMED-CT: (415571003, SNOMED-CT: (415571003, 415572005, 415570002, Create fact value 415572005, 415570002) 279261000000103) CTV3: (XaJz4, XaK27, XaK2A) CTV3: (XaJz4, XaK27, XaK2A, XaMh9) 161 Zoet, Smit, and Leewis The second phase of coding is axial coding. To support this process Glasser (1978) formulated 18 coding families. Glaser (1992) stresses that researchers should not blindly apply each individual coding family to data at hand. The application for a specific coding family must emerge first from the research question and secondly from the data. The purpose of applying coding families in our research was to determine mutual exclusivity between and completeness of the modifications that can be applied to business rules (sets). To test for mutual exclusivity and completeness we therefore applied coding families that searched for end stages, clusters, conceptual ordering, conformity, and structural ordering: the ordering and elaboration family and means-goal family (Glaser, 1978). Applying the mentioned coding families served as a basis for the business rule modifiability framework, which is depicted in Table 2. el e u l r ur n n e n n n n o oi e e ul o o o ss e ss oi it ti ul ul a si si e t i a v i si i n n i d d a c ic c si d n n v v tc e e e u si n o o t tc a d u c c d d b b oc c a af f e e e f e t e t e e e et t e t a t a ta t t e a et t a e el d e a l d a e d r e p er l e e p e l p C e r e C D U r - D U D C U - D - - - - - - C - - A B C D E F G H -I J K Version 9 1 1 3 72 49 13 Version 10 15 12 10 Version 11 21 40 12 Version 12 8 3 2 9 8 14 Version 13 3 10 Version 14 12 4 1 7 4 2 9 16 4 10 Version 15 13 150 10 Version 16 157 39 310 10 Version 17 2 12 Version 18 19 10 Version 19 16 20 9 26 64 81 5 4 7 11 Version 21 1 4 106 16 12 Version 22 4 2 7 25 86 83 12 2 59 28 20 Version 23 1 99 12 12 Version 24 107 2 12 Version 25 33 32 52 19 28 10 21 18 29 33 93 Version 26 4 77 4 25 Version 27 8 113 7 44 Version 28 2 16 10 13 70 27 2 534 30 15 26 Version 29 8 16 14 Total 67 74 88 111 252 206 195 576 849 697 380 Grand total 3495 Figure 2: Amount of modifications per modification category Furthermore, it is interesting to report on what caused the large amount of modifications for some versions of the business rule sets. For example, we know that the large amount of modifications concerning the modification type Delete fact value in version 16 are caused by the phase out of a medical information system containing those fact values. However, it is beyond the scope of this study to fully elaborate on these causes. More research on the causes of the large amount of modifications for some versions can be found in the work of Gilliam and Siriwardena (2011). 162 Classification of Modification Categories for Business Rules 4 Results In this section the identified modification categories are presented elaborated upon. To ground the modification categories, our research includes an example of a business rule set within the context of the QOF which is provided in Figure 3 and Figure 4. Table 2: Business rules modification framework Business Fact Decision Condition Rule value Create CD CBR CC CFV Update UD UC UFV Delete DD DBR DC DFV A. The first modification is identified as: “create decision.” This modification adds an additional decision or sub-decision to the already existing set of business rules. This includes all underlying variables such as new business rules and new fact values. This particular modification category is observed 67 times out of 3495 observations. B. The second modification is identified as “delete decision. ” This modification deletes a decision that, for example became obsolete. This includes all underlying variables such as new business rules and new fact values. This particular modification category is observed 74 times out of 3495 observations. C. The third modification is identified as “Update decision. ” This modification solely updates the name (label) of a specific concept without changing underlying logic. An example regarding the QOF is a decision currently labelled as: Amount of achievement points obtained, which is modified into: Amount of achievement percentage obtained. This particular modification category is observed 88 times out of 3495 observations. D. The fourth modification is identified as “create business rule. ” This modification creates a new business rule within the business rule set of a given decision, including one or more conditions and one conclusion. This particular modification category is observed 111 times out of 3495 observations. E. The fifth modification is identified as “delete business rule. ” This modification deletes an existing business rule within the business rule set of a given decision, including one or more conditions and one conclusion. This particular modification category is observed 252 times out of 3495 observations. F. The sixth modification is identified as “create condition. ” This modification creates a new condition to be used by existing or new conclusions. An example regarding the QOF is the addition of a ratio to calculate the conclusion final points achieved. The condition relative achievement ratio is added in the calculation to balance inequalities of register list sizes of general practices. This particular modification category is observed 206 times out of 3495 observations. G. The seventh modification is identified as “delete condition. ” This modification deletes an existing condition from a given ruleset. An example regarding the QOF is the deletion of the condition higher threshold. In the new situation, GP’s will or will not achieve the minimum threshold and will not be able to attain bonus achievement over a 163 Zoet, Smit, and Leewis certain achievement percentage anymore. This particular modification category is observed 195 times out of 3495 observations. H. The eight modification is identified as “Update condition. ” This modification solely updates the name (label) of a condition. An example regarding the QOF is a condition currently labelled as: REF_DAT, which is modified into: ACHIEVEMENT DAT. This particular modification category is observed 576 times out of 3495 observations. Figure 3: Example business rule document of the QOF 1/2 (Health and Social Care Information Centre, 2014) I. The ninth modification is identified as “create fact value. ” This modification creates a new fact value for its parent condition or conclusion. An example regarding the QOF is the addition of a fact value under a new condition labelled as maximum raw points achieved. The fact value added operates as an upper threshold and is set to 550. This particular modification category is observed 849 times out of 3495 observations. J. The tenth modification is identified as “delete fact value. ” This modification deletes an existing fact value from its parent condition or conclusion. An example regarding the QOF is deleting a fact value from the conclusion patient registration status. From the four available conclusions this ruleset can generate, the fact value previously registered is deleted, leaving the possibility to generate three conclusions. This particular modification category is observed 697 times out of 3495 observations. K. The eleventh modification is identified as “Update fact value. ” A fact value is a possible value or fixed value of its parent condition. An example regarding the QOF is renaming the fact values of the condition Upper threshold from 70 achievement 164 Classification of Modification Categories for Business Rules percentage to 80 achievement percentage. This particular modification category is observed 380 times out of 3495 observations. Figure 4: Example business rule document of the QOF 2/2 (Health and Social Care Information Centre, 2014) The eleven identified modifications have a hierarchical structure. In this structure the highest level of is a decision followed by business rules, conditions and fact values. The existence of a hierarchy indicates a cause and effect relationship between the different elements. For example, when a new decision is created the possibility exist that also new business rules, conditions and fact values must be created. The data shows this is not always the case since underlying hierarchical elements are reused. Due to size constraints we decided to omit a full overview of this phenomenon. 5 Conclusion & discussion Business rules are widely applied, standalone and embedded in smart objects. Therefore they have become a separate concern in information system design. As a result they also have to be managed separately. From a technical and economic perspective it is impossible to build an information system that can cope with every modification possible. Therefore a choice has to be made which defined set of anticipated modifications the system must be stable to cope with (Mannaert and Verelst, 2009) . The purpose of this research is to define the set of anticipated modifications a business rule set must be able to cope with. To be able to this we addressed the following research question: “Which modifications can impact a business rule set. ” In order to answer this question, we conducted a grounded theory study on modifications occurring in the business rules applied for payment to primary care organizations in the United Kingdom by the NHS, the QOF payment schemes. In total we analysed 3495 modifications that occurred during the last eight years resulting in a set of modification types that can occur to business rules (sets). From the data, we identified eleven types of modifications: A) create decision, B) delete decision, C) update decision, D) create business rule, E) delete business rule, F) create condition, G) delete condition, H) update condition, I) create fact value, J), delete fact value, and K) update fact value. From a research perspective, our study provides a generalization from collected data to constructs and theory (Lee and Baskerville, 2003). Thereby it provides a fundament for further research which can focus on building business rule architectures that can optimally cope with the identified modifications. From a practical perspective, our study provides an overview of the modifications that can occur to business rules which can help organizations to construct test scenarios that help information systems to cope with future modifications. Several limitations may affect our results. The first limitation is the related to sample size. While the sample size of business rules modifications (3495) is representative, the 165 Zoet, Smit, and Leewis modification types are all derived from one organization, which may limit generalization. The second limitation is related to the first, our sampling strategy. Our research was applied to business rule sets from the medical industry. And while the medical industry is known for the relatively high amount of utilization of business rules, several other industries are interesting to include as well; for example the financial or governmental industries. The omission of modifications to business rules from other industries may also limit generalization. Adding business rule sets from other industries will be a part of further research. 6 References Appleton, D. (1984). Business Rules: The Missing Link. Datamation, 145-150. Arnott, D., & Pervan, G. (2005). A critical analysis of decision support systems research. Journal of Information Technology, 20(2), 67-87. Bajec, M., & Krisper, M. (2005). A methodology and tool support for managing business rules in organisations. Information Systems, 30, 423-443. Bass, L., Clements, P., & Kazman, R. (2012). Software Architecture in Practice (3rd Edition) New York: Addison-Wesley Professional. Blomgren, M., & Sunden, E. (2008). Constructing a European Healthcare Market: The private healthcare company Capio and the strategic aspect of the drive for transparency. . Social Science & Medicine, 67(10), 1512-1520. Boyatzis, R. (1998). Transforming qualitative information: Thematic analysis and code development. Thousand Oaks, CA: Sage. Boyer, J., & Mili, H. (2011). Agile Business Rules Development: Process, Architecture and JRules Examples. Heidelberg: Springer. Gillam, S., & Siriwardena, N. (2011). The Quality and Outcomes Framework: QOF- transforming general practice: Radcliffe Publishing. Glaser, B. (1978). Teoretical Sensitivity: Advances in the Methodology of Grounded Theory. Mill Valley, CA: Sociology Press. Glaser, B., & Strauss, A. (1967). Discovery of Grounded Theory. Strategies for Qualitative Research. Mill Valley: Sociology Press Graham, I. (2006). Business Rules Management and Service Oriented Architecture. New York: Wiley. Health and Social Care Information Centre. (2007). Chronic Obstructive Pulmonary Disease Rule Set Version 10. http://webarchive.nationalarchives.gov.uk/20110426140805/http://pcc.nhs.uk/bu siness-rules-v10-0: Health and Social Care Information Centre. Health and Social Care Information Centre. (2014). Chronic Obstructive Pulmonary Disease Rule Set Version 29 http://www.hscic.gov.uk/qofbrv29: Health and Social Care Information Centre.,. King, R., & Green, P. (2012). Governance of primary healthcare practices: Australian insights. Business Horizons, 55(6), 593-608. Kovacic, A. (2004). Business renovation: business rules (still) the missing link. . Business Process Management Journal, 10(2), 158-170. 166 Classification of Modification Categories for Business Rules Lee, A., & Baskerville, R. (2003). Generalizing Generalizability in Information Systems Research. Information Systems Research, 14(3), 221-223. Liao, S.-H. (2004). Expert System Methodologies and Applications - A Decade Review from 1995 to 2004. Expert Systems with Applications, 28(1), 93-103. Mannaert, H., & Verelst, J. (2009). Normalized Systems: Re-creating Information Technology Based on Laws for Software Evolvability. Antwerpen: Koppa Digitale Media. Morgan, T. (2002). Business rules and information systems: aligning IT with business goals. London: Addision-Wesley. Nelson, M. L., Peterson, J., Rariden, R. L., & Sen, R. (2010). Transitioning to a business rule management service model: Case studies from the property and casualty insurance industry. Information & Management, 47(1), 30-41. Nelson, M. L., Rariden, R. L., & Sen, R. (2008). A Lifecycle Approach towards Business Rules Management . Proceedings of the 41st Hawaii International Conference on System Sciences, Hawaii. Qumer, A., & Henderson-Seller, B. (2006). Measuring agility and adoptability of agile methods: a 4-dimensional analytical tool . Proceedings of the IADIS International Conference Applied Computing, Barcelona. Shao, J., & Pound, C. J. (1999). Extracting business rules from information systems. BT Technology Journal, 179-186. Tarantino, A. (2008). Governance, Risk, and Compliance Handbook. New Jersey: Wiley. Taylor, J. (2013). The Decision Management Manifesto: An Introduction. In Decision Management Solutions (Ed.). decisionmanagementsolutions.com/decision- management-manifesto Van der Aalst, W. (1996). Three Good Reasons for Using a Petri-net-based Workflow Management System . Proceedings of the International Working Conference on Information and Process Integration in Enterprises, Cambridge. Van Thienen, J., & Snoeck, M. (1993). Knowledge factoring using normalisation theory. Paper presented at the International Symposium on the Management of Industrial and Corporate Knowledge, Compiegne. Vanthienen, J., & Wets, G. (1994). Restructuring and optimizing knowledge representations . Proceedings of the Tools with Artificial Intelligence, 1994. Proceedings., Sixth International Conference on. Versendaal, J. (1991). Seperation of the User Interface and Application. PhD, Technische Universiteit Delft, Rotterdam. Von Halle, B., & Goldberg, L. (2010). The Decision Model: A Business Logic Framework Linking Business and Technology. Boca Raton, FL: Auerbach Publications. Weske, M. (2007). Business Process Management - Concepts, Languages, Architectures. New York: Springer. Zoet, M., Versendaal, J., Ravesteyn, P., & Welke, R. (2011). Alignment of Business Process Management and Business Rules . Proceedings of the European Conference on Information Systems. 167 Zoet, Smit, and Leewis 168 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia The Effect of e-Mental Health Services on Saudi General Mental Health Bader Binhadyan Ministry of Education, Saudi Arabia binhadyan@gmail.com Konrad Peszynski RMIT, Australia konrad.peszynski@rmit.edu.au Nilmini Wickramasinghe Epworth Research Institute, Australia nilmini.wickramasinghe@epworth.org.au Abstract Mental Health is a state of wellbeing that plays a key role in affecting the quality of life of an individual. However, globally the level of treatment and focus it receives in terms of funding and priority as a healthcare issue differs significantly. More recently mental health in Saudi Arabia has begun to receive better attention from authorities and researchers. In particular, e-mental health services and solutions are growing rapidly and are showing promise to facilitate mental health services and delivery by providing better access, and early intervention and treatment for people with mental illness. This study has been designed to assist the current mental health services in Saudi Arabia and focuses on e-mental health, which is both timely and important. In the last decade Australia has become one of the leading countries in providing e-mental health services. The research-in-progress outlined in this paper introduces possibilities and challenges in transforming the e-mental health services of Australia to the Saudi Arabian healthcare context. Keywords: E-mental health, Saudi Arabian healthcare, Wellbeing, Social Network, Mobile health, Australian healthcare. 169 Binhadyan, Peszynski, Wickramasinghe Introduction Wellbeing is defined as a state of being with others, which emerges where human needs are met, where one can act meaningfully to pursue one's goals, and where one can enjoy a satisfactory quality of life (McGregor, 2008). To enjoy a satisfactory quality of life is directly related to mental health status. Mental health is defined as “state of well-being in which every individual realizes his or her own potential, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to her or his community” (WHO, 2014). Looking at the definitions of wellbeing and mental health, good mental health status is a key factor of wellbeing. Therefore, mental health services have to be available, easy to access, and affordable for people in need. In the last decade, technologies, such as the Internet and smartphones, are growing in popularity for mental health services and delivery (Lal & Adair, 2014). These technologies can assist mental health providers and governments to facilitate their mental health services by improving efficiency, accessibility and the opportunities for early intervention and treatment for many people with different mental illnesses (Christensen & Hickie, 2010; Christensen et al., 2009; e-Mental Health Alliance, 2014; Jorm, Morgan, & Malhi, 2013). They can also improve the barriers to treatments such as the stigma associated with visiting psychology clinics and poor mental health literacy, geography in general (Burns, Davenport, Durkin, Luscombe, & Hickie, 2010), and gender segregation, spiritual or religious beliefs in Saudi Arabia specifically (Alkabba, Hussein, Albar, Bahnassy, & Qadi, 2012; Christensen & Petrie, 2013b; Hammad, Kysia, Rabah, Hassoun, & Connelly, 1999; Koenig et al., 2014). Recently, the mental health sector in Saudi Arabia has started receiving attention from Saudi authorities and mental health providers to develop a strategic plan to facilitate this sector and improve the mental health wellbeing in Saudi Arabia (Al-Habeeb & Qureshi, 2010; Koenig et al., 2014). However, there are a number of barriers to mental health services and treatment, which can be specific for that region such as religious healing practices, gender segregation, social and legal aspects (al-Shahri, 2002; Hammad et al., 1999; Koenig et al., 2014). Although Saudi Arabia has invested billions of dollars to improve the quality and the delivery of e-health in the last ten years (Altuwaijri, 2010), less focus has been given to e-mental health. This is a key void and one this paper sets out to address. The emphasis of this research-in-progress design will be on assisting the current state of the mental health sector of Saudi Arabia, and how e-mental health services can better the mental health services and reduce the barriers that affect mental health services and delivery in Saudi Arabia. The Australian e-mental health services programme was chosen to be the case study to assist with the investigation. The research question guiding this study is: How are e-mental health services implemented to facilitate the current mental health in Saudi Arabia? 170 The Effect of e-Mental Health Services on Saudi General Mental Health Background To better investigate the expected effect of e-mental health services on Saudi mental health services and delivery, it is necessary to understand the Saudi Arabian culture. Islam shapes the culture, social, health practice and politics in Saudi Arabia (Al-Saggaf, 2004; Koenig et al., 2014). This is because Islam is not only a religious ideology, but a complete system that offers detailed prescriptions for the entire way of life (Al-Saggaf, 2004; Aldraehim, Edwards, Watson, & Chan, 2012; AlMunajjed, 1997). Therefore, the Saudi health system is strongly grounded in religion and culture, which needs to be taken into consideration when examining and treating patients and planning health services (Koenig et al., 2014). In mental health for instance, religious healing practice, sex segregation, or women’s legal and social aspects are the main factors that might be barriers to treatments, such as affecting mental health services access or decreasing the mental health literacy (Al-Saggaf, 2004; al-Shahri, 2002; Koenig & Al Shohaib, 2014; Koenig et al., 2014). These factors will be discussed later in this paper Recently, the mental health sector in Saudi Arabia is undergoing a large transformation and new policies and acts have been applied by the Saudi authorities to improve mental health services and deliveries (Koenig et al., 2014; Qureshi, Al-Habeeb, & Koenig, 2013). This section provides an overview of the mental health sector in Saudi Arabia and the related aspects to that country. Finally, a brief introduction of e-mental health services and their benefits and disadvantages is presented. 1.1 Saudi Mental Health Sector Recently, the mental health sector has started to gain attention from Saudi authorities; to improve accessibility to mental health services and treatments as well as implanting better mental health policy and procedure (Almalki, Fitzgerald, & Clark, 2011), the recent milestone developments are shown in Table 1. Year Development References 2006 National mental health policy (Koenig et al., 2014) Special mental health programmes in the general medical system 2007 1st Saudi Arabian Mental and Social Health Atlas (Al-Habeeb & Qureshi, 2010) 2010 2nd Saudi Arabian Mental and Social Health Atlas (Koenig et al., 2014) Table 1: Recent mental health development milestone in Saudi Arabia Between 2006 and 2012, the Ministry of Health (MoH) reported that the total number of outpatients seeking mental health services at public hospitals increased by 59.4% (from 310,848 to 495,484 cases), and the total number of inpatients increased by 12.9% (Moh, 2011, 2012). Moh (2012) used the International Classification of Diseases (ICD-10) to identify disease groups and reported the following in 2012:  48.93% of the total number of visitors to mental health clinics were women 171 Binhadyan, Peszynski, Wickramasinghe  53.19% were between the age of 15-40  Depression 35%  Anxiety 36% Saudi Mental Health Factors Besides the common barriers to mental health services and treatments such as stigma, location, therapist availability and geographic location (e-Mental Health Alliance, 2014; Gulliver, Griffiths, & Christensen, 2010; Lal & Adair, 2014), religious faith healing beliefs, sex segregation and/or a number of women’s legal and social aspects are some of the factors that have been found that may impact mental health service access or delivery (Al-Saggaf, 2004; al-Shahri, 2002; Koenig & Al Shohaib, 2014; Koenig et al., 2014; Saleh, 2014). Table 2 illustrates these factors. Factor Deception Religious Faith Healing Koenig and Al Shohaib (2014) argue there is a positive impact of religious faith healing on a person’s wellbeing and improves their hope and self-esteem and provides a sense of belonging. However, this practice can delay diagnosis and treatment for serious mental illness, and can increase the criticism that these delays may lead to increased anxiety and guilt in the patient (Koenig & Al Shohaib, 2014). Sex Segregation Islam implies sex segregation, which means that women are prohibited to mix with unrelated men and this applies to the work environment, education, and hospitals (Al- Saggaf, 2004). Usually men demand a female doctor to examine their female relatives, or a female will refuse to be seen by a male doctor (al-Shahri, 2002). As a result, mental health treatment may be affected. Women's Legal and Social aspects Divorce has a direct impact on Saudi women's mental health wellbeing and Saudi Arabia has the highest divorce rate among the Gulf Cooperation Council countries (35%) and is also above the world average rate of 22% (Saleh, 2014). Because women in Saudi Arabia are not permitted to drive and there is insufficient public transport (Alghamdi & Beloff, 2014), the access to mental health services for these women might become a challenge. Table 2: Factors that may impact mental health service access in Saudi Arabia The negative impact of these factors are some of the barriers to mental health access and treatment in Saudi Arabia and as mentioned, the number of people seeking mental health attention is increasing, which can be a challenge that mental health providers may face in the future. Therefore, the need for e-solutions that will facilitate the mental health sector is required. E-mental health services indicate significant outcomes to improve the barriers that affect traditional mental health services. 172 The Effect of e-Mental Health Services on Saudi General Mental Health E-mental Health E-mental health is defined as providing treatment and/or support to people with different mental disorders through sensible technologies (as seen in Table 3) (Anthony, Nagel, & Goss, 2010; Christensen & Petrie, 2013b; e-Mental Health Alliance, 2014; Whittaker et al., 2012). E-mental health services have the ability to improve accessibility, reduce cost, provide flexibility, and better consumer interactivity and engagement (Lal & Adair, 2014). Some of the tools are shown in Table 3. E-mental Health tools  Short Message Service (SMS)  Email  Website/apps  Chat or instant messaging (IM) tools  Social Media  Video/Audio via the Internet  Smart phones  Tablets Table 3: Tools that are used in mental health services E-mental health services have the ability to overcome issues existing in the current mental health sector; however, there disadvantages of using e-mental health. Table 4 shows these benefits and disadvantages of e-mental health. Benefits Disadvantages  Improve lack of access due to location, time or financial  Lack of quality control (Lal & Adair, difficulties (Booth et al., 2004) 2014)  Reduce stigma incurred by seeing a therapist (Burns et  Limited only for people with low to al., 2010; Christensen & Hickie, 2010) moderate mental illnesses (Lal &  Adair, 2014) Improve mental health literacy   Limited to people who are familiar with Improve the therapist’s time and efficacy (Jorm et al., using technology (Lal & Adair, 2014) 2013; Jorm, Wright, & Morgan, 2007). Table 4: E-mental health benfists and disadvatages One of the countries considered a leader in e-mental health services is Australia, which will be introduced later in this paper. Method and research design E-mental health in Saudi Arabia has not been previously explored and it lacks defined characteristics. As such, an exploratory qualitative research method is the most suitable method and a single case study will be used (Yin, 2008). At this stage of the research, the answer to the research question of “How are e-mental health services implemented to facilitate the current mental health in Saudi Arabia?” will be examined and explored. Creswell (2013) argues that the use of case study as a tool of investigation is found in many fields, which allows the research to develop an in-depth analysis of a field. According to Stake (1995) there are three types of case study research: intrinsic, instrumental, and collective case studies. To gain more insight and knowledge into this research topic, an instrumental case study has been chosen. 173 Binhadyan, Peszynski, Wickramasinghe In general, case study methodology is not strictly planned but researchers are guided by what they see in the field, given a planned field study with specific steps for data collection and analysis (Fidel, 1984). This flexibility provides researchers with the ability to deal with unexpected events and results. However, to reduce risks that might appear during data collection and analysis of this research, a framework developed by Eisenhardt (1989) will be employed. This framework involves the following eight steps: Getting started In this step, after reviewing and outlining the literature, the research question will be developed and defined. A priori constructs will be examined for further measuring. At this stage of the research, the research question is developed and the related literatures are articulated. Selecting the case study The Australian mental health services programme has been selected as the single case study that will be examined against Saudi Arabia’s mental health sector. 1.1.1 E-mental Health in Australia In the last decade Australia has become one of the leading countries in providing e- mental health services along with Sweden and the Netherlands. Furthermore, 50% of the recent publications in e-mental health are Australian resources (Christensen & Petrie, 2013a). These services target young people in Australia using Internet technologies (Australian Government, 2012). The majority of e-mental health programs target depression, anxiety and suicidal thoughts. There are five types of e-mental health services in Australia as shown in Table 1. Number Mental health services type Example 1 Health promotion, wellness promotion and Beyondblue www.beyondblue.org.au psycho-education 2 Prevention and early intervention Kids Helpline www.kidshelp.com.au MoodGYM moodgym.anu.edu.au 3 Crisis intervention and suicide prevention Lifeline www.lifeline.org.au 4 Treatment myCompass www.mycompass.org.au/ 5 Recovery and mutual peer support BlueBoard www.blueboard.anu.edu.au Table 5: Australian mental health services types Crafting the instrument and Protocols Observation and document review is the main method adopted to collect data for this research. Observation is a fundamental and highly important method in all qualitative inquiry (Yin, 2008). Because Australian mental health services are online (e-Mental Health Alliance, 2014), observation and document review will be the most stable method at this stage of the research. Tools, such as Nvivo and MS Visio, will be used to facilitate this stage of the research to explore the relationships and identify differences 174 The Effect of e-Mental Health Services on Saudi General Mental Health in both contexts. Some of the aspects that will be observed and taken into consideration include the culture and how the role of e-mental health will be different in a Saudi Arabian context. Entering the Field This step can behave as a data collection and the initial data analysis sub-process. In this stage, the data will be collected. Data collection and analysis overlap, and research can go back and forth until the main theory emerges. Data Analysis The case study approach allows the researcher to move back and forth between the data collection and analysis, and it will continue until the main theory starts to develop. The data is analysed by using within-case analysis (Yin, 2008). The data collected will be inspected for common characteristics that represent categories or themes (Boyatzis, 1998). For instance, the important data or information collected will be divided into categories that will later involve a group of sub-categories after more in-depth analysis occurs. Hypotheses formulation The relationship between the sub-categories will be validated and the categories will be refined. Most likely in this stage, all findings will be considered and the overall theory can be shaped and tested. The relationships and findings will be verified among the unique and specific constructs and aspects identified; for instance, culture and language. Some of the expected hypotheses will be introduced in the discussion section of this paper. Enfolding literature In this stage, the findings will be compared with similar and conflicting literature. Inspecting literature that conflicts with the findings will support the confidence in the research result and will represent an opportunity (Eisenhardt, 1989). “ The result can be deeper insight into both the emergent theory and the conflicting literature, as well as sharpening of the limits to generalization of the focal research”(Eisenhardt, 1989, p. 538). Recommendation Based on the outcomes of the data analysis, findings and the literature, a recommendation will be proposed. Discussion This research-in-progress aims to explore what Saudi Arabia is able to adapt from Australian e-mental health services. This will be done through observing and analysing the current status of mental health services in Saudi Arabia and examining the cultural 175 Binhadyan, Peszynski, Wickramasinghe similarity and differences between Saudi Arabia and Australia. E-mental health services will assist in overcoming cultural and religious barriers to access mental health services or treatment. Islamic faith healing practices influence the mental health of individuals either in a positive or negative way. Mental health professionals cannot ignore these practices. The outcome of using e-mental services is to increase the practitioners’ awareness of the importance of religious faith healing and their patients’ mental health and it might provide a common ground between the practitioners and their patients, or perhaps ease the integration of Islamic beliefs and practices into the psychology treatment model. Abdel-Khalek (2009); Thomas and Ashraf (2011) cite that using spiritually modified cognitive therapies have a significant influence in decreasing symptoms in Saudi people with depression. In addition, by reviewing previous studies conducted on e-mental health’s benefits and potentials (Christensen & Hickie, 2010; Christensen & Petrie, 2013b; e-Mental Health Alliance, 2014; Jorm et al., 2013; Lal & Adair, 2014), it is believed that e-mental health services can be used as a strategic tool to improve mental health literacy to facilitate early intervention for mental disorders and as a cost effective tool to increase awareness among patients, their families, and religious faith healers to reduce any delays in treatment for clinical mental illness. Gender is “a critical determinant of health, including mental health. It influences the power and control men and women have over the determinants of their mental health, including their socioeconomic position, roles, rank and social status, access to resources and treatment in society” (Astbury, 2001, p. 4). Furthermore, depression is the most diagnosed mental illness globally and it is found in women more than men (Piccinelli & Wilkinson, 2000). In Saudi Arabia, 50.19% of visitors to psychiatric clinics were women (Moh, 2011). Sex segregation was almost not existent in Saudi online communities and e-services (Al-Saggaf, 2004; Alghamdi & Beloff, 2014). Furthermore, direct communication and interaction between the two genders in Saudi Arabia has improved since internet technologies were introduced (Madini & de Nooy, 2014). The benefits of e-mental health services will assist in eliminating sex segregation, which will allow some mental health services to be available to everyone at the same time. One of the benefits that has been profoundly improved by e-mental health is accessibility (Lal & Adair, 2014). In Saudi Arabia, women are not permitted to drive and by going online, they do not need to drive to access mental health services. Studies show that women in Saudi Arabia who have sought online help and joined online social networks to help them overcome mental distresses associated with divorce, such as loneliness and social stigma has increased their self-esteem and reduced depression and anxiety (Saleh, 2014). 176 The Effect of e-Mental Health Services on Saudi General Mental Health An important element that has to be measured and explored in future research is the user acceptance of using e-mental health in Saudi Arabia. Conclusion This research-in-progress aims to investigate how e-mental health services can facilitate mental health services in Saudi Arabia and how these services can impact mental health wellbeing for Saudi people. Further, it also looks into the possibilities and challenges of transferring lessons from the Australian context into the Saudi Arabian healthcare context. E-mental health services have the potential to improve the current state of mental health in Saudi Arabia, and to reduce the barriers that affect mental health services and delivery for this country. This investigation will motivate researchers and mental health providers to conduct research and analysis that seeks to enhance healthcare and health results, and support the introduction of an e-mental health vision in Saudi Arabia. Furthermore, the research will also contribute to e-health practices in Saudi Arabia as well as support policy development associated with e-mental health. This paper outlined the initial theory and conceptual framework for this research. The next steps include data collection and analysis in the chosen case study. In this way, a fuller appreciation of the impact of e- mental health on facilitating and supporting wellbeing will become evident. References Abdel-Khalek, A. M. (2009). Religiosity, subjective well-being, and depression in Saudi children and adolescents. Mental Health, Religion & Culture, 12(8), 803-815. Al-Habeeb, A., & Qureshi, N. (2010). Mental and Social Health Atlas I in Saudi Arabia. EMHJ, 16(5). Al-Saggaf, Y. (2004). The effect of online community on offline community in Saudi Arabia. The Electronic Journal of Information Systems in Developing Countries, 16. al-Shahri, M. Z. (2002). Culturally sensitive caring for Saudi patients. Journal of Transcultural Nursing, 13(2), 133-138. Aldraehim, M. S., Edwards, S. L., Watson, J. A., & Chan, T. (2012). Cultural impact on e-service use in Saudi Arabia: the role of nepotism. International Journal for Infonomics (IJI), 5(3/4), 655-662. Alghamdi, S., & Beloff, N. (2014). Towards a comprehensive model for e-Government adoption and utilisation analysis: The case of Saudi Arabia. Paper presented at the Computer Science and Information Systems (FedCSIS), 2014 Federated Conference on. Alkabba, A. F., Hussein, G. M., Albar, A. A., Bahnassy, A. A., & Qadi, M. (2012). The major medical ethical challenges facing the public and healthcare providers in Saudi Arabia. Journal of Family and Community Medicine, 19(1), 1. Almalki, M., Fitzgerald, G., & Clark, M. (2011). Health care system in Saudi Arabia: an overview. Eastern Mediterranean health journal, 17(10). AlMunajjed, M. (1997). Women in Saudi Arabia Today. UK: Macmillan. Altuwaijri, M. (2010). Supporting the Saudi e-health initiative: the Master of Health Informatics programme at KSAU-HS. EMHJ, 16(1). Anthony, K., Nagel, D. M., & Goss, S. (2010). The use of technology in mental health: Applications, ethics and practice. Springfield, Ill: Charles C Thomas Pub Limited. 177 Binhadyan, Peszynski, Wickramasinghe Astbury, J. (2001). Gender disparities in mental health. WORLD HEALTH ORGANIZATION. Australian Government. (2012). E-mental health strategy for Australia. Canberra Retrieved from http://www.health.gov.au/internet/main/publishing.nsf/Content/7C7B0BFEB985D0EB CA257BF0001BB0A6/$File/emstrat.pdf. Booth, M. L., Bernard, D., Quine, S., Kang, M. S., Usherwood, T., Alperstein, G., & Bennett, D. L. (2004). Access to health care among Australian adolescents young people's perspectives and their sociodemographic distribution. Journal of Adolescent Health, 34(1), 97-103. Boyatzis, R. E. (1998). Transforming qualitative information: Thematic analysis and code development: Sage. Burns, J. M., Davenport, T. A., Durkin, L. A., Luscombe, G. M., & Hickie, I. B. (2010). The internet as a setting for mental health service utilisation by young people. Medical Journal of Australia, 192(11), S22. Christensen, H., & Hickie, I. B. (2010). E-mental health: a new era in delivery of mental health services. Medical Journal of Australia, 192(11), S2. Christensen, H., & Petrie, K. (2013a). Information technology as the key to accelerating advances in mental health care. Australian and New Zealand Journal of Psychiatry, 47(2), 114-116. Christensen, H., & Petrie, K. (2013b). State of the e-mental health field in Australia: Where are we now? Australian and New Zealand Journal of Psychiatry, 47(2), 117-120. Christensen, H., Proudfoot, J., Andrews, G., Klein, B., Kavanagh, D., O'Neil, D., . . . Graham, K. (2009). E-mental health: a 2020 vision and strategy for Australia. Creswell, J. W. (2013). Research design: Qualitative, quantitative, and mixed methods approaches (4th ed.). London: Sage Publications, Incorporated. e-Mental Health Alliance. (2014). E-Mental Health Services in Australia 2014: Current And Future. online: e-Mental Health Alliance,. Eisenhardt, K. M. (1989). Building theories from case study research. Academy of management review, 14(4), 532-550. Fidel, R. (1984). The case study method: a case study. Library and Information Science Research, 6(3), 273-288. Gulliver, A., Griffiths, K. M., & Christensen, H. (2010). Perceived barriers and facilitators to mental health help-seeking in young people: a systematic review. BMC psychiatry, 10(1), 113. Hammad, A., Kysia, R., Rabah, R., Hassoun, R., & Connelly, M. (1999). Guide to Arab culture: Health care delivery to the Arab American community. Retrieved October, 15, 2010. Jorm, A. F., Morgan, A. J., & Malhi, G. S. (2013). The future of e-mental health. Australian and New Zealand Journal of Psychiatry, 47(2), 104-106. Jorm, A. F., Wright, A., & Morgan, A. J. (2007). Where to seek help for a mental disorder? The Medical Journal of Australia, 187(10), 556-560. Koenig, H. G., & Al Shohaib, S. (2014). Understanding How Islam Influences Health Health and Well-Being in Islamic Societies (pp. 295-310): Springer. Koenig, H. G., Al Zaben, F., Sehlo, M. G., Khalifa, D. A., Al Ahwal, M. S., Qureshi, N. A., & Al- Habeeb, A. A. (2014). Mental Health Care in Saudi Arabia: Past, Present and Future. Open Journal of Psychiatry, 4(02), 113. Lal, S., & Adair, C. E. (2014). E-Mental Health: A Rapid Review of the Literature. Psychiatric Services, 65(1), 24-32. Madini, A. A., & de Nooy, J. (2014). Cross-gender communication in a Saudi Arabian Internet discussion forum Opportunities, attitudes, and reactions. Convergence: The International Journal of Research into New Media Technologies, 1354856514534574. 178 The Effect of e-Mental Health Services on Saudi General Mental Health McGregor, A. (2008). Well-being, Poverty and Conflict Briefing Paper 1/08, ESRC Research Group on Well-being in Developing Countries. Moh. (2011). Statistical Book for the Year 1432. http://www.moh.gov.sa: Retrieved from http://www.moh.gov.sa/en/Ministry/Statistics/book/Documents/1431.rar. Moh. (2012). Statistical Book for the Year 1433. http://www.moh.gov.sa: Retrieved from http://www.moh.gov.sa/en/Ministry/Statistics/book/Documents/1433.pdf. Piccinelli, M., & Wilkinson, G. (2000). Gender differences in depression Critical review. The British Journal of Psychiatry, 177(6), 486-492. Qureshi, N. A., Al-Habeeb, A. A., & Koenig, H. G. (2013). Mental health system in Saudi Arabia: an overview. Neuropsychiatric disease and treatment, 9, 1121. Saleh, R. H. (2014). The Supporting Role of Online Social Networks for Divorced Saudi Women. (Masters), University of Ottawa. Stake, R. E. (1995). The art of case study research (1st ed. Vol. 95004979). Thousand Oaks, CA: Sage publications Thomas, J., & Ashraf, S. (2011). Exploring the Islamic tradition for resonance and dissonance with cognitive therapy for depression. Mental Health, Religion & Culture, 14(2), 183- 190. Whittaker, R., Merry, S., Stasiak, K., McDowell, H., Doherty, I., Shepherd, M., . . . Rodgers, A. (2012). MEMO—A mobile phone depression prevention intervention for adolescents: Development process and postprogram findings on acceptability from a randomized controlled trial. Journal of medical Internet research, 14(1), e13. WHO. (2014). Mental health: a state of well-being. Retrieved 11/11, 2014, from http://www.who.int/features/factfiles/mental_health/en/ Yin, R. K. (2008). Case study research: Design and methods (4th ed.). Thousand Oaks, Ca: Sage. 179 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia The role of IT governance in generating business value from IT investments in healthcare: Lessons from an Australian experience Peter Haddad RMIT University and Epworth HealthCare, Australia peter.haddad@rmit.edu.au Nilmini Wickramasinghe Deakin University and Epworth HealthCare, Australia nilmini.wickramasinghe@epworth.org.au Abstract Digitizing the core processes of healthcare delivery is looked at as a solution to control the escalating costs without compromising quality or patient outcomes. However, to date the business value of such IT solutions remains elusive, especially in view of the high failure rate of many solutions coupled with the high user resistance. The uniqueness of the healthcare industry makes measuring the business value of IT a complex missions, yet it is the thesis of this research that such an activity is an essential first step if we are to realise the full potential of IT in healthcare. The role IT governance can play is of high importance to generate business value from IT investments in healthcare. This is investigated using an integrative model that is proffered to conceptualise the business value of IT in healthcare. This conceptual model is then used to guide an exploratory case study based at a leading private healthcare provider in Melbourne-Australia. Keywords: Well-being, Business Value, Value, Information Technology, IT Governance 1 Introduction Today, digitizing healthcare processes is a relatively common practice, and we see more healthcare providers aggressively moving to IT-enabled solutions, which need both up- front and ongoing investments for outcomes that no one can precisely predict (Weill & Ross, 2004). This trend has found a good appetite for researchers to study the impacts of different IT solutions. Although there are plethora of such studies, most of them share the same two types of limitations. First, the lack of a comprehensive framework 180 Haddad, P., Wickramasinghe, N. that looks at these systems in their contexts. Second, the scope of these studies is mostly limited on the impact of one system on limited measurements in the output. This paper represents a part of a larger research project to comprehensively assess business value of IT. It particularly investigates the role of IT governance in generating business value of IT in the context of healthcare. This paper is arranged as follows. First, it gives a brief summary of the terms ‘value’ and ‘business value’, and then some insights from the literature on the basic definitions and principles of IT governance, and how does it differ from IT management in the contemporary organisations. Then the conceptual model to assess business value of IT in healthcare along with the underpinning theories will be presented, followed by the case study and the findings and discussion. 1.1 Value and Business Value Healthcare commentary often revolves around universal availability and cost control, i.e. access and cost (Wickramasinghe & Schaffer, 2010). Further, value is often defined in terms of the expenditure outcome benefits, divided by the cost expenditure (Porter & Teisberg, 2006). The healthcare benefits, from a patient’s perspective, include the quality of healthcare outcomes, the safety of the delivery process, and the services associated with the delivery process (Rouse & Cortese, 2010; Wickramasinghe & Schaffer, 2010) The term 'business value of IT' is commonly used to refer to the organizational performance impacts of IT i.e. the impact of enterprise architecture (digitizing the operations in a firm) including cost reduction, profitability improvement, productivity enhancement, competitive advantage, inventory reduction, and other measures of performance (Melville, Kraemer, & Gurbaxani, 2004). It is important to emphasize that business value of IT is not a value by itself; rather, it is a model that suggests how value might be generated by implementing different IT solutions (Haddad, Gregory, & Wickramasinghe, 2014). 1.2 IT Governance The term IT governance is relatively new in the academic and professional contexts (Wim Van & Steven De, 2012). One of the main reasons it has emerged was the increasing need for higher level of accountability and responsibility. This, in turn, was a direct result of many failures in generating business value from IT investments (Weill & Ross, 2004). The Term IT governance has evolved from a mere mechanism to manage IT implementation to extend beyond IT contexts and cover the business domain. It was defined in the context of Hawaii International Conference on Systems Sciences (HICSS) as “organizational capacity exercised by the board, executive management and IT management to control the formulation and implementation of IT strategy and in this way ensure the fusion of business and IT” (Van Grembergen & DeHaes, 2008). The standardization organization ISO also issued ISO/IEC 38500 in 2008 as a worldwide new standard called “Corporate Governance of IT”. Weill and Ross (2004) earlier defined IT governance as “Specifying the decision rights and accountability framework 181 The role of IT governance in generating business value from IT investments in healthcare to encourage desirable behaviour in the use of IT”, identifying six key assets that enable achieving the strategies of an organization and generating business value: human assets, financial assets, physical assets, IP assets, information and IT assets, relationship assets. The leadership in IT governance is controversial; IT people argue they know how to manage IT implementations and even reinventing business processes to utilise IT systems and solutions. At the same time, business people and many researchers argue that the leadership of IT governance is the core responsibility of business people, differentiating between effective IT management (the effective delivery of IT services internally) and IT governance, whose aim is to better fit IT implementations into the business strategy (Wim Van & Steven De, 2012). The literature of IT governance in healthcare lags way behind compared with other industries. This may reflect the complexity of healthcare, as it has a third powerful player (the clinicians) beside IT and business players in other industries. This paper would serve as one of the few attempts to study IT governance in healthcare, aiming at exploring the best practices of an effective IT governance, identifying the main barriers and enablers for such a governance, and issuing a number of recommendations in this regard. 2 Research Design and Methodology This section presents the conceptual model and the methodology, alsong with justifying the selection of the case study. 2.1 The Conceptual Model In order to develop an integrative model (See Figure 1) that will assess the business value of IT in healthcare, all perspectives of healthcare value from the respective points of view of all key stakeholders must be considered (Haddad et al., 2014). To operationalize the IT resource, from a technical perspective, the conceptual model is based on the IT portfolio suggested by (Weill & Broadbent, 1998) who classify IT investments based on their business objectives into infrastructure IT, transactional IT, informational IT, and strategic IT. Table 1 summaries the differences between these investments. Table 1: IT Portfolio [Adopted from (Weill & Broadbent, 1998)] Objectives Description Infrastructure  The foundation of IT capacity, which is delivered as reliable services, shared throughout the firm and coordinated centrally, usually by the IT group.  Include both the technical and the managerial expertise required to provide reliable services.  Having the required infrastructure services in place significantly increases the speed with which new applications can be implemented to meet new strategies, thus increasing the firm’s strategic agility and flexibility. 182 Haddad, P., Wickramasinghe, N. Transactional  Process and automate the basic, repetitive transactions of the firm. These include systems that support order processing, inventory control, bank cash withdrawal, statement production, account receivable, accounts payable, and other transactional processing.  Transactional systems aim to cut costs by substituting capital for labor or to handle higher volumes of transactions with greater speed and less unit cost. These systems build on and depend on a reliable infrastructure capacity. Informational  Provide information for managing and controlling the firm.  Systems in this category typically support management control, decision making, communication and accounting. These systems can summarize and report this firm’s product and process performance across a wide range of areas.  Two examples of these systems come from Ford Australia (Electronic Corporate Memory), and from the consulting firm Bain & Company which developed Bain Resources Access for Value Addition (BRAVA). Strategic  The objective of strategic technology investment is quite different from those of the other parts of the portfolio.  Strategic investments are made to gain competitive advantage or to position the firm in the marketplace, most often by increasing market share or sales.  Firms with successful strategic IT initiatives have usually found a new use of IT for an industry at a particular point an time.  Two good examples of theses strategic initiatives are inventing automatic teller machines (ATMs), and designing a system that provides immediate 24-hour, seven-day-a-week loan approvals in car dealerships using expert systems technology. Both of these innovative systems have changed their industries forever. Finally, it is necessary to recognize the socio-technical perspective of these systems at four interrelated levels: (i) Clinical practices (people); (ii) Delivery operations (processes); (iii) System structure (organizations); and (iv) Healthcare ecosystem (society) (Rouse & Cortese, 2010) which all work together to provide a better patient experience. 183 The role of IT governance in generating business value from IT investments in healthcare Figure 1: The Proposed Conceptual Model 2.2 Case Study To test the validity of the proposed model, a case study method is adopted. As noted by (Yin, 2014), a case study method is appropriate when conducting an exploratory research study especially when the research question is how or why, as is the case in the current study. The case study was chosen based on the size of the hospital and the extent to which IT is utilized in terms of reach and range. The case study is one of the largest not-for-profit private health care groups in Victoria- Australia, renowned for excellence in diagnosis, treatment, care and rehabilitation. This hospital is well-known as an innovator in Australia’s health system, embracing the latest in evidence-based medicine to pioneer treatments and services for patients. During the last few years, more investment have been set for IS/IT solutions. According the Chief Financial Executive (CFO), these investments have exceeded 30% of the capital budget for the last financial year, and more is dedicated for IT investments during the 184 Haddad, P., Wickramasinghe, N. upcoming five years. The increasing amount of IT investments in this hospital makes it ideal as a case study to represent the healthcare sector, which started few years ago to follow this trend. Inputs were taken from three groups of participants: business, IT, and clinical personnel via both semi-structured interviews and an online survey. All interviews were professionally transcribed and qualitatively analysed using Nvivo qualitative analysis software package. Annual reports and other archival data were also used for data triangulation. All necessary ethics clearances were received before data collection commenced. 3. Findings About 3 years ago, and due to recognizing the growing importance of IT for healthcare delivery, the hospital in the case study commenced a large-scale initiative to digitize its core processes by having an increasing amount of the operational and capital budgets dedicated to IT. Investing in IT in this case study goes through a tough and sophisticated governance process. This starts with a business initiative, which needs to be submitted to the Project Management Office (PMO) within the IT department. A high-level assessment and a review are to be done on this initiative, to make sure it has all needed information. The idea then is explored. As needed, a project manager from the PMO would meet with the person that’s actually put up the proposal and just flesh it out a little bit. After having this done, the initiative would go up to the IT Steering Committee, which is chaired by the CIO and most of the executive directors and the CEO. The IT Steering Committee is to discuss whether the hospital is interested in such a project. This depends on its expected benefits and fitness within the business strategy and IT architecture of the hospital. In that case, more work needs to be done. This starts by deeper discussions between the PMO and the initiator of the project. Upon this, a business case is created. The path from this point depends on the budget required for this project; if it is less than a threshold (One million Australian dollar as of the time of data collection), then the business case is put in the Prioritisation List, on which all planned projects are listed based on their importance to the business. If the required budget is more than this threshold, then the business case is forwarded to the Finance Steering Committee and then to the Board, who will decide whether or not this business case will be sent to the Prioritisation List. Figure 2 depicts this process. 185 The role of IT governance in generating business value from IT investments in healthcare Figure 2: The Process of IT Governance at the Case Study If the business case is deemed appropriate and the suggested IT system passes this scrutiny then a ‘sponsor’ is assigned to the project. A sponsor in our case context would be a senior employee whose tasks are functionally aligned with the nature of the proposed IT system. Thus, to the best of our understanding, there have been two criterion to appoint a sponsor for an IT project: 1. The nature of the IT project (financial, clinical, administrative, IT, etc.) 2. The experience/ expertise required to sponsor each IT project. For example, if the suggested system addresses aspects of cancer care, the sponsor ideally will be an oncologist with a deep understanding of IT-based oncology management systems. If the system is meant to deal with the financial aspects of the business, then the sponsor would be the CFO, or the executive director of procurement and facilities, just depending on the nature of the project. These people would have had enough expertise dealing with similar investments. The sponsor has a relatively high level of authorisation within the dedicated budgets for their assigned projects. If they need additional resources more than 5% of the budget, they still can ask for it, but they have to go through another cycle of governance to demonstrate the reasons and the promises to the board. At the same time, they are fully accountable and the failure or success of their assigned IT project is their sole responsibility. Adopting this approach started informally in 2009 during a project to change the payroll systems, and over the last five years, it has become more formal and documented. This practice has shown an impact on the success of IT projects and generating business value of IT for this hospital. Apart from very few cases, people who have been interviewed found it difficult to identify a failed IT project since this approach was introduced. Besides the matching the nature of IT projects with experienced sponsors, the strong governance process gives the business the ability to predict possible failures and prevent it: “if something was going to fail, you’d see it coming a mile off. Each major project, each month, there is a one-page or a two-page update that goes to the Finance Committee. It says what the status of the project is, what are the key milestones, what are the upcoming activities the next month. There is a track, what we spent today against the 186 Haddad, P., Wickramasinghe, N. budget, what has been committed against budget. It is really transparent. It’s very obvious if something is going to go off track.” (FPM#1). All stakeholders at the case study recognise the importance of a good IT governance for successful IT investments in clinical and business domains: “Good governance structure has been something we’ve worked on in the last couple of years and I feel it absolutely necessary to actually work in this environment” (Exec#1). Although IT governance has this important role, generating business value of IT is not a direct result of a good IT governance. IT only enables the best opportunity to succeed in delivering IT services: “What we certainly know is without that [governance] structure, it becomes very hard to deliver that initial benefit of actually getting that new system in and transitioned over in a way those benefits or that accommodates the business and the business as usual work.” (Exec#2). The real business value of IT requires a strong governance process, but it ultimately is subject to the “decision of the chosen solution and program of work to begin with. It’s those two factors apply.” (Exec#3) The role of IT department in this practice is more advisor and supportive than leading: “we don’t want IT to say, “Here’s your new Internet,” and everyone is going, “Well, this is a pile of junk.”” (CFO). One of the most important roles IT department is expected to practice is change management, as well as technical support. The Requirements of a Successful IT Governance Successful IT governance has a number of requirements: 1. It is demanding in terms of human assets, and needs to be well resourced. This would mean enough personnel equipped with a diverse range of skills and expertise: “We resource it out properly so that we have people not doing it as part of their day jobs. We actually have a dedicated project manager, business analyst, and project team. That has been a real key”. (PMO#5). 2. It needs to continuously be nurtured within a collaborative atmosphere: “we expect that it’s going to be collaborative approach, so it’s not someone just running off and doing what they want to do for their site. There needs to be collaborative approach.” (CFO). 3. It needs a deep knowledge of business processes and organisational structure: “It needs to be at the front end in the sense that they need to basically have line of sight as to the processes” (PMO Program Manager). 4. It needs to be very well planned for up front. This will lead to a transparent project management and easy to track progress: “The fundamental failure up front leads to massive rework, inefficiencies and costs down the back-end and usually leads to immense frustration because it's based on "I 187 The role of IT governance in generating business value from IT investments in healthcare thought I asked for this" and there's no checkpoints along that whole journey” (PMO#3). 5. The right sponsor to the right project: Although there has been increasing concentration on matching the requirements of specific IT projects and the unique requirements for prospective sponsors, selecting the right sponsor needs to go beyond that, to cover the organisational loyalty. During one IT project, a number of cases of lack of planning and delays happened, even though the same strong IT governance was applied. Asking about the reason, we were advised that the sponsor was not an employee at the case study: “The role of the sponsor really in my view, it wouldn’t have mattered who that person was. It needs to come back to an Epworth executive and someone who’s employed and has accountability back to our board for delivering that outcome” (Exec#6). 4. Discussion The main finding we could extract from our case study is that a successful IT governance structure is a must in order to generate business value from IT investments, but it is not enough on its own. The chosen system IT system and its fit within the business strategy is the main factor in this regard. This is facilitated by a good IT governance structure though. The business people should practice the leadership role in IT governance, not IT, whose role should be advising and supporting the front-end role of business. IT department still need to deliver support and practice a mediating role between technology and business, but they should not drive IT governance. This is a priori theme in the literature. See for example (Van Grembergen & DeHaes, 2008; Weill & Ross, 2004). Now, we know that this also applies on the healthcare context. The collected data revealed a number of requirement a good IT governance structure would need in healthcare. Most of the requirements are human and organisational. From human perspective, a good IT governance requires to be well equipped with enough dedicated human resources, whose organisational loyalty should be to their organisations and not their own business objectives. i.e. they will have to be salaried employees for their hospitals and have accountability back to the board of their hospital (legal employer) to deliver the expected outcomes. From the organisational point of view, hospitals need to encourage collaborative atmospheres between three different groups of knowledge workers: clinicians, business, and IT personnel, and also they need to nurture upfront planning and reengineering of the organisational processes. Thus, IT governance can play a role as an enabler for organisational development, and should benefit from it in return. Future directions for this research will benefit from its current limitations. We have used a single case study to qualitatively investigate the role of IT governance in generating business value of IT in healthcare. Extending this research to quantitatively investigate this in multiple case studies is one of the main directions for future research. 188 Haddad, P., Wickramasinghe, N. 5. Conclusion There is a critical need for a systematic integrative conceptual model to assist the assessment of the business value of IT in healthcare, and to aid the understanding of the business value of the underlying enterprise architecture. To address this need, a suitably robust model was developed from various bodies of IS and business literature. The proffered model was then tested using case study methods in Victoria-Australia. For the purpose of this paper, IT governance and its role in generating business value from IT investments is tested. As a summary, a good IT governance structure is 1) demanding, 2) a must, and 3) promising to generate business value from IT investment in healthcare. In closing, we note that today wellbeing consists of financial wellbeing and wellbeing in terms of healthcare - when we take both together in healthcare we need to ensure that IT systems which are now being used to enable better healthcare delivery and consequently wellbeing ensue are well designed and sustainable and this in turn necessitates prudent IT governance strategies. To ensure sound IT governance strategies we have provided a suitable model to assist with the assessment of the business value of IT. In using this model it is possible for healthcare organisations to design and implement robust IT solutions that are both sustainable and support superior healthcare delivery. Future research will elaborate upon these findings further. References Haddad, P., Gregory, M., & Wickramasinghe, N. (2014). Business value of IT in health care. In L. A.-H. C. G. Nilmini Wickramasinghe & T. Joseph (Eds.): Springer. Melville, N., Kraemer, K., & Gurbaxani, V. (2004). Review: Information Technology and Organizational Performance: An Integrative Model of IT Business Value. MIS Quarterly, 28(2), 283-322. Porter, M. E., & Teisberg, E. O. (2006). Redefining health care : creating value-based competition on results. Boston, Mass.: Harvard Business School Press. Rouse, W. B., & Cortese, D. A. (2010). Engineering the system of healthcare delivery. Amsterdam: Amsterdam: IOS Press. Van Grembergen, W., & DeHaes, S. (2008). Implementing information technology governance : models, practices, and cases. Hershey, PA: IGI Pub. Weill, P., & Broadbent, M. (1998). Leveraging the new infrastructure : how market leaders capitalize on information technology. Boston, Mass.: Harvard Business School Press. Weill, P., & Ross, J. (2004). IT governance : how top performers manage IT decision rights for superior results. Boston, Mass.: Harvard Business School Press. Wickramasinghe, N., & Schaffer, J. (2010). Realizing Value Driven e-Health Solutions IMPROVING HEALTHCARE SERIES. Washington DC: IBM Center for the Business of Government. 189 The role of IT governance in generating business value from IT investments in healthcare Wim Van, G., & Steven De, H. (2012). A Research Journey into Enterprise Governance of IT, Business/IT Alignment and Value Creation Business Strategy and Applications in Enterprise IT Governance (pp. 1-13). Hershey, PA, USA: IGI Global. Yin, R. K. a. (2014). Case study research : design and methods (Fifth edition. ed.): SAGE. 190 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health1 Luuk P.A. Simons Delft University of Technology, Netherlands L.P.A.Simons@tudelft.nl David van Bodegom Leyden Academy of Vitality and Aging, LUMC, Netherlands Bodegom@leydenacademy.nl Adrie Dumaij Delft University of Technology, Netherlands A.C.M.Dumaij@tudelft.nl Catholijn M. Jonker Delft University of Technology, Netherlands C.M.Jonker@tudelft.nl Abstract Work site healthy lifestyle interventions hold promise for improving health and employability. As part of a larger employer vitality program and a work site RCT (Randomized Controlled Trial, n=59 intervention arm) to assess cardiac risk impacts, we conducted a design analysis on a hybrid eHealth solution. The control condition was a six weeks waiting list and then start of the hybrid eHealth support (n=57). Based on preliminary 6 week- and 3 month-results, the hybrid eHealth support generated statistically significant risk factors improvement (like LDL cholesterol). The waiting list condition yielded no significant improvements. The late start after the waiting list did yield significant improvements, but not as large as a direct start. The direct start also appears to yield higher satisfaction and intention to recommend. 1 We are very grateful for the contributions of Kees Schotsman, Niek Stolp, Eelco van Stokkum, Wendy van Leeuwen, Dick Hoeneveld, Lonneke Baas, Bas Gerritsen, Ralph Feenstra, Lotta Breed, Lucas van Vliet and Rudi Westendorp during setup and execution of this study. 191 Simons, Bodegom, Dumaij, Jonker Our analysis supports three types of conclusions. First, the hybrid eHealth intervention did significantly improve physical risk factor variables after 6 weeks. Motivation and measurement alone (waiting list) did not. Second, theory on timing of health support for patient appeared generalizable to employees: it did help to offer support at a moment of high motivation, instead of later. Third, a design analysis was conducted regarding service mix efficacy in relation to key requirements for designing ICT-enabled lifestyle interventions. This resulted in several recommendations and improved service adoption. Keywords: RCT, work site health, service design, multi-channel services, healthy lifestyle intervention 1 Introduction Cardio-metabolic syndrome hampers the health of almost half of the population by the age of 60. It was estimated that 43% of people >60 years of age have (cardio-)metabolic syndrome (Lakka 2002). This may put them at a 4.26-fold risk of death in an 11 year follow up compared to healthy men, and they are estimated to have a 3.7-fold risk for coronary artery disease and a 24.5-fold risk to develop diabetes-2 (Sattar 2003). The symptoms of metabolic syndrome can become apparent as cardiovascular disease, obesity and/or diabetes type 2. Metabolic syndrome increases burdens for the individual, as well as burdens on a societal and employer level. It has been estimated that cardiovascular disease leads to 10 additional sick days at work plus 1 month productivity loss while present at work (sickness presenteeism). For diabetes-2 these numbers are: 11 work days and 8 weeks sickness presenteeism (Steenbeek 2010). In general, cardio-metabolic syndrome hampers physical and mental energy, plus employability. Currently, the majority of the 60+ citizens in the potential work force in the Netherlands are not employed (CBS 2012). This is partly due to health and vitality concerns. For males, the 60-65 age group consumes the largest health care budget, mainly due to cardiovascular disease (Slobbe 2011). Given our aging population and rising health care costs, the need grows for a population which stays healthy and employable longer. The context of this study is provided by an employee vitality program. Since 2010 the Human Resource (HR) department of Delft University of Technology has piloted an extensive eSupported lifestyle program, which combines coach sessions with electronic dashboarding and self-management. The HR department and company physicians mainly aim at the following part of the employee population: those with increased cardio-metabolic risk (inclusion of >50 participants/year), with increased absenteeism (inclusion of >50 participants/year) plus a minority admission (inclusion of <30 participants/year) with various health issues or interests. In this HR setting, promising risk factor effects have been measured on a pre-/post- intervention basis (Simons 2013, 2014). Next, an RCT (Randomized Controlled Trial) has been designed, in order to further assess efficacy of the eSupported lifestyle program (Verweij 2011). The primary research question is: are physical risk factors impacted by the intervention? This RCT study design does not aim at the entire 192 Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health employee population. Rather, it aims at the employee subset which meet the eligibility criteria for cardio-metabolic risk, ability to participate and motivation. The primary study outcomes are total cholesterol and LDL cholesterol, aimed at cardio-metabolic risk. Besides, there are longer term outcomes. There is an HR interest in exploring impacts on productivity related measures like work engagement (UWES-9, Bakker 2009, Schaufeli 2006) or presenteeism and absenteeism (Iverson 2010). These longer term outcomes are outside the scope of this analysis of preliminary results. This paper does address another important issue: efficacy of the service mix deployed in the eSupported lifestyle intervention. We combine preliminary, short term results from the RCT measurements with a design analysis based on an evaluation framework of requirements for ICT-enabled healthy lifestyle interventions. 2 Theory The eSupported lifestyle program combines coach sessions with electronic dashboarding and self-management. Hybrid programs (face-to-face plus tele-support) have been indicated to be attractive (Demark-Wahnefried 2008) for participants because those participants have multiple preferences: face-to-face counselling is valued, but there are also perceived thresholds to face to face contacts due to travel- and logistics burdens for visiting a clinic. Hence follow-up contacts via telephone or internet are often preferred (Jones 2006). Finding the right mix between offline and online contacts is an ongoing design research challenge (Pekmezi 2011). To increase solution impact, a hybrid or multi-channel service mix is recommended (Sperling 2009, Simons 2002, 2006, 2010, 2010b), combining electronic and face-to- face interactions. For example, face to face ‘on site’ coaching had as benefits: a richer service experience with the coach, with other participants and with a health focused ‘service scape’; group support experiences (obtaining additional social support and co- creating service experiences together); learning from each other; health experiences in healthy food-, sports- and relaxation exercises. Disadvantages are: more (travel) time needed; less flexibility regarding when and where; and not everyone likes group sessions (Demark-Wahnefried 2007). Electronic and (semi-)automated coaching has as benefits: more time-efficient; more flexibility in when and where to have contact; very explicit monitoring of your own progress online; having status reports including ‘next steps’ commitments always online. Disadvantages are: the sensory-, emotional- and group experiences are more limited. Also, the ‘service scape’ in which people are immersed is only virtual, not physical. In summary, often a hybrid service mix has most to offer. In such a hybrid service mix, micro-learning tools accessible via smartphone, mail and/or web, potentially offer a number of advantages: they use a personal device that is available any time any place, they are efficient and can use idle time that is otherwise lost, and they are suited for just in time learning (Bruck 2012, Simons 2015). Key functionalities to increase health motivations and behaviours in this eSupported lifestyle program are (Simons 2010 and 2014): 193 Simons, Bodegom, Dumaij, Jonker • A personal online health dashboard with graphs of progress towards adherence targets on the various health behaviours; • Automated feedback (online and in email) on lifestyle aspects where relatively positive scores have been achieved (nutrition, physical activity, stress management or an overall score); • (Tele)coaching by a health coach, generating online reports on progress towards adherence targets in the personal dashboard; • The (tele)coaching sessions can be flexibly planned, based on convenience and participant preference: during in-clinic visits or phone based from home; • Options to ask questions to the coach: via messaging within the dashboard or via email; • Online schedule indicating upcoming events: group sessions, individual coach sessions (when and where), physical measurements, surveys; • A micro-learning Health Quiz accessible via smartphone, mail and/or web; • Reading materials in the mail; • Weekly tips via email on health, motivation and self-management; • Besides individual coaching, group sessions are also used in order to stimulate group support, mutual inspiration and encouragement, plus peer education. Secondly, reviews (Jones 2006, Demark-Wahnefried 2008, Pekmezi 2011) of lifestyle interventions suggested that multidisciplinary interventions have advantages (diet, physical activity and mental health). There are health advantages, since the total health impacts are larger. And an important motivational benefit is that participants experience progress faster (Simons 2010), regarding quality of life and self-efficacy. This increases intrinsic motivation and chances of long term compliance (Seligman 2012, Baumeister 2011). The lifestyle advice follows the guidelines of the Harvard Epidemiology and Nutrition Group for nutrition, physical activity and smoking cessation. Over the past decades, multiple studies from Harvard have illustrated that many diseases of affluence are largely preventable with only 5 or 6 healthy lifestyle guidelines. Willett (2004) repeats in an overview of these findings: 72% of colon cancer is preventable, 81% of coronary disease and 92% of diabetes type 2. The guidelines are (Willett 2004, 2011) to increase intake of fruits and vegetables (2,5 servings/day or more), to choose whole grains instead of refined grains, to have one daily serving of nuts and/or legumes, to limit intake of red meat and processed meat, to limit intake of trans and animal fats, and to have no more than 2 alcoholic beverages/day. Physical exercise guidelines are: 30 min/day moderate intensity activity (like walking or gardening) and 3x20 min/week intensive activity (Borg level 12-14). Stress management guidelines are: relaxation exercises for 30 min/day. If we look at the design challenge of persuasive technology (Fogg 2002, 2009) for health, it was theorized and tested elsewhere that this challenge is not just located in the ICT design, but also in the design of the overall service scape, including health effects 194 Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health and coach relationship (Simons 2014). It should generate positive, mutually reinforcing service experiences across communication channels and activate long term health motivation and -behaviours, in order to deliver long term results. This is reflected in the following design evaluation framework for health improvement ICT solutions (Simons 2014), see Figure 1. It helps evaluate the impact of ICT-enabled interventions on health effectiveness, coaching performance and ICT value adding. Health effectiveness: - Health literacy - Health behaviors - Health outcomes - Quality of life and well-being Coaching performance: ICT value adding: - Promoting health actions - Quality of motivators, triggers, experiences - Supporting self-efficacy - Simplicity: familiar interfaces, ease of use - Activating intrinsic motivation - Embedded in and enhancing coach relation Figure 1: Basic requirements when designing ICT-supported healthy lifestyle interventions Figure 1 addresses three evaluation domains. Firstly, health effectiveness not only includes health outcomes, but also health literacy (‘as a user I know how to best serve my health’), health behaviours and health well-being (meaning health related quality of life (Ware 1998) and the Seligman (2012) dimensions of well-being related to health). Preferably, health interventions have broader and deeper impacts rather than narrow ones, since the former will improve health well-being more significantly. Experiencing larger health well-being impacts forms an important intrinsic motivator for health behaviours in the longer term. Secondly, coaching performance not only includes promoting health actions (improving health readiness by moving from awareness to intentions to behaviours as in the HAPA and i-change models, Schwarzer 2010, Wiedeman 2011), but also activating intrinsic motivations, and supporting users in their self-efficacy (their day-to-day attempts and successes to turn their health behaviour experiments into health wellness experiences, Lipke 2009). Thirdly, ICT value adding includes (Fogg 2002, Fogg 2009): value adding via high quality triggers, motivators and service experiences (which often involves using a mix of channels, each for their strengths – Demark-Wahnefried 2007, De Vries 2008, Sperling 2009, Simons 2004, Simons 2006), simplicity (which means using ICT interfaces that are mainstream for the user group and being very attentive to ease of use - many initiatives underperform due to usability barriers, see Jimison 2008) and finally: embedding applications in an overall health provider or coach relationship (so that the meaning is enhanced of the coach relationship as well as the meaning of the data). For example, the foundations of coaching include ‘building rapport or relationship’, using different levels of listening based on empathy and intuition, see Starr 2008. This is best done by a person. Whereas data capturing, processing and feedback to users is preferably automated (Simons 2010b). 195 Simons, Bodegom, Dumaij, Jonker 3 Methods and Study Design Our study consists of an RCT (Randomized Controlled Trial) within a larger employer vitality program. Participants were recruited in 2014, on a voluntary basis, from the employee base of the Delft University of Technology. Slightly more participants entered the program than required on the basis of the power calculations for minimal sample size. After a 0-measurement of vitality and control variables and when meeting inclusion criteria, participants were randomly assigned to either a direct start of the lifestyle intervention (n=59), or to a waiting list with a start after a six weeks: the control group (n=57). Hence the control group consisted of participants who were re-measured six weeks later and then entered into the hybrid eHealth program. The study protocol was approved by the medical ethics committee of Leiden University Medical Centre. The first measurements and randomization started in January, there were ten start groups during the year and the final (waiting list control) group started their eHealth support program on November 27th 2014. On January 13th 2015 they were the final group for which the 6-weeks post-start physical measurements were conducted. At the time of writing, their 3-months survey results were not available yet; for the other nine start groups they were. Physical inclusion criteria were chosen on the basis of medical literature. The other in- /exclusion criteria were mainly concerned with feasibility and practicality: Can someone fully participate in the program and is there enough motivation? Eligibility following these criteria is checked by the company physicians, who know many of the employees. Besides, there are self-assessment questions in the 0-measurement for the prospective participants, regarding the degree of motivation and ability to participate. The inclusion criteria are: • Cardiovascular disease (including previous diagnoses, hypertension (>=140/90) or hypercholesteremia (cholesterol >= 6.0 or LDL >= 3.4), and/or diabetes-2 (including prediabetes risk: HbA1C >= 6.0) and/or being overweight (BMI >= 25); • Physically, mentally and socially capable of participating in an intensive lifestyle program. Exclusion criteria are: • Serious comorbidity or treatment side effects that hamper participation; • Psychiatric problems; • Risk factor measurement outcomes which require immediate medication changes; • Not enough motivation to participate (score < 3 ‘average’ on a 5-point scale). In this employee sub-population with cardio-metabolic risk, standard deviation for total cholesterol and LDL cholesterol is about 0.9 mmol/l. Hence, power calculations indicated that if the true difference in the experimental and control means is 0.5 mmol/l, we needed to study at least 52 experimental subjects and 52 control subjects to be able 196 Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health to reject the null hypothesis that the population means of the experimental and control groups are equal with probability (power) 0.8. The Type I error probability associated with this test of the null hypothesis is 0.05. Besides describing short term physical effects, a qualitative service design analysis is conducted in the results section, using the Figure 1 requirements framework from theory regarding the design of ICT-supported healthy lifestyle interventions. 4 Results In this section we combine preliminary, short term results from the RCT measurements with a design analysis based on the framework of Figure 1 (from the Theory section). Table 1 shows the differences between the waiting list (control) groups and the intervention groups that had a direct eHealth support start. Waiting list (control) participants (n=57): Direct start (intervention) participants (n=59): Descriptive: Descriptive: [as Waiting list, plus directly:] - Motivated volunteers. - Intake & personal action plan. - Taken ownership by applying for study. - Start workshop (full day) + coach sessions. - Physical measurements raised awareness. - Health behaviour logging. - Majority self-searched for measurement - Coach progress logs in dashboard interpretation and started health actions. - Heath Quiz + weekly start tips. Measurements: Measurements: - Avg LDL reduction at 6 weeks waiting = 0.08 +/- - [pre-start difference is not in study design] 0.17 mmol/l (95% CI) (n=52) - Avg LDL reduction at 6 weeks post-start = 0.16 - Avg LDL reduction at 6 weeks post-start = 0.35 +/- 0.13 mmol/l (95% CI) (n=48) +/- 0.16 mmol/l (95% CI) (n=53) - Avg satisfaction start week = 8.0 (n=54) - Avg satisfaction start week = 8.2 (n=55) - Avg satisfaction 3 months = 8.5 (n=33*) - Avg satisfaction 3 months = 8.6 (n=46) - Avg recommendation 3 months = 8.2 (n=33*) - Avg recommendation 3 months = 8.7 (n=46) Table 1: Comparison: Waiting list vs Direct start 6 weeks post-intervention start; *not all 3- month data available yet for Waiting list participants (CI= Confidence Interval; grades: 1-10) The descriptive elements indicate that the waiting list participants do have incentives to start health improvements after initial measurements, even though they are on the waiting list. They were motivated to start health improvement, that had actively stepped forward and enlisted themselves for the study (a process involving significant obligatory paper work) and they had received the results of their physical examination. After the 6 weeks waiting period we had intakes with these participants, where the majority indicated that had tried to interpret the results (usually Internet aided) and started attempts at healthier behaviours. By comparison, the ‘direct start’ participants had similar incentives, plus the full hybrid eHealth support package. Regarding physical measurements, the period of 6 weeks of waiting did not change LDL cholesterol (or any other health variables) significantly. By contrast, a direct start resulted in a 0.35 mmol/l LDL (‘high risk’) cholesterol reduction (significant, p<0.05), which is an improvement of about 10%. After delayed start (waiting list), LDL cholesterol reduced half that amount: 0.16 mmol/l (significant, p<0.05). Other health indicators also improved significantly (p<0.05) for the direct start group: BMI, total 197 Simons, Bodegom, Dumaij, Jonker cholesterol and diastolic blood pressure, but these variables have less statistical power to determine differences with the waiting list group, thus less suitable RCT endpoints. For all n=116 participants, the average initial values were (no statistical differences between groups): LDL = 3.7 mmol/l, total cholesterol = 5.7 mmol/l, BMI = 27, blood pressure = 127/82 mm Hg, with medication aid for some participants. Reductions in medication did take place, but not in the first 6 weeks of the intervention. Finally, Table 1 shows somewhat lower satisfaction scores for the waiting list participants (not statistically significant), and especially a lower score in the recommendation intention after 3 months. For the direct start participants the recommendation score is relatively high, at 8.7 (10-point scale) The 95% Confidence Interval was 0.29 (n=46), but we cannot conclude on statistical significance of the difference with the 8.2 waiting list group recommendation, given the still low n=33 of these preliminary data. However, these grade differences do reflect the comments we heard during intakes and coach sessions: that several participants had lost part of their motivation or worked hard at the wrong things during the waiting list period. Health Effectiveness Coaching Performance ICT Value Adding Health Literacy: Promoting health actions: Motivators, triggers, experience: ++ Health Quiz and start tips. + Suggestions in Health Quiz. + + Health quiz, start tips: (fun) - Waiting list effect: self-search - Waiting list: Some started in- experiences, triggers, hope, confusions. effective behaviour patterns. success experiences. Health behaviours: Supporting self-efficacy: Simplicity: - Waiting list effect: some + Health Quiz: improved +/- Old behaviour logging was a taking the wrong actions. portfolio of strategies (coping, burden (limited adoption); the Health outcomes: avoiding pitfalls) new version was better. - - Waiting list: poor short term Activating intrinsic motivation: Fit with coach processes: effects. + Start tips: 24 weeks + Health Quiz enhances coach Quality of Life: motivators on all health topics. insights and suggestions. + Participants sent thank you - - Waiting list effect: Part of the + The new behaviour logging mails replying to the start tips. start-motivation is gone. enhances behaviour insights. Table 2: Design evaluation on design requirements from Figure 1 (authors’ opinions, 5-point scale from - - to ++) Table 6 shows a qualitative design evaluation of effects observed by the authors during the 2014 RCT. On the one hand, several elements of eHealth support were added or changed, which led to improvements. On the other hand, there was a waiting list effect on several of our design requirements from Figure 1, which hampered performance on the design requirements. There were three forms of eHealth support added. Two at the start of 2014. First, a selection of 24 weekly start tips in the mail, to support growth in health awareness and competences. Second, a micro-learning Health Quiz starting one month after the initial workshop. An initial service design description in given in Simons (2014, 2015). In the 2014 implementation we added gaming elements (points for trying, extra points for correct answers, speed points for fast responses, for completing a level and reaching daily targets) and team play (team scores and top score lists), plus further simplification of the user process (participants automatically receive daily, clickable emails to enable 198 Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health answering Health questions; and weekly group progress statistics are mailed). A third improvement that was introduced at the end of 2014 was a simplified system for logging weekly health behaviours. So the final starting groups of 2014 benefitted from them and we could compare before-after differences. In summary, in terms of design requirements, the largest hampering effects from the waiting list procedure were: on average a decay in motivation of participants, self- search for measurement interpretation abounded but led to confusion and to adoption of some poor quality health beliefs, plus several participants started in-effective or unhealthy behaviour patterns. Also the waiting list led to lower health outcome improvements: not just after the first six weeks of waiting, but also after six weeks of hybrid eHealth support. The largest contributions from the Health Quiz were: improved health literacy and providing a continuous stream of motivators, triggers and success experiences (this enhances self-efficacy and further learning). The largest contribution from the 24 weekly start tips in the mail were: continued motivation support and providing triggers. The largest contributions from the improved weekly behaviour logging interface that started late in the year, from Dec 2014, are threefold. First, lower thresholds to logging (participants indicate that the new logging software is more enjoyable). Second, when people log a week’s behaviour, they enter about 50% more entries (exercise, mental balance, buddy contacts and foods/drinks). Third, participants look more closely at the progress graphs, which we contribute to freeing up extra mental processing capacity. 5 Discussion and Conclusion This preliminary analysis has several limitations. First, the 3 months survey data is not complete yet: due to the late starting date of the final group their 3 months data was not be available at the time of writing. Second, the study design for the RCT was aimed at testing eHealth intervention effects on physical risk parameters after 6 weeks. The design analysis was a qualitative addition to that study design. Third, regarding external validity, these study results may only apply to motivated individuals, who volunteer for lifestyle training. Four, thanks to the fact that the control group also entered the program, but after 6 weeks waiting list, we expected a limited ‘demotivation’ effect of being randomized into a control group. Still, some demotivation was observed, but not quantified. Still, on the positive side this study design did provide an opportunity to observe waiting list effects and to conduct a design analysis in relation to a number of eSupport changes that were made. Further details are provided in the remainder of this paper for specific situations. 5.1 Design Lessons and Implications for Practice Many employers offer (preventive) medical checkups, often without explicit follow up programs for health support. We have observed in the waiting list group, that a majority of people use the Internet and/or family/friends for: a) interpretation and b) possible health behaviour improvements. Unfortunately, this regularly leads to confusion and/or ineffective behaviours. Which partly explains why the waiting list results in our study 199 Simons, Bodegom, Dumaij, Jonker were minimal. Moreover, even just a 6-weeks-ineffectiveness period was enough to reduce about half of the positive risk factor effects of our hybrid eHealth support after participants did start program, at least in the short term (6 weeks after start). This suggests that offering employees active health support directly after measurements yields better results. We have to wait for the 1-year results to determine the longer term risk factor impacts. A second intervention design lesson also regards timing. In our 2014 RCT, we started our micro-learning Health Quiz plus weekly start tip mails only after 1 month, based on the rationale that the first intervention month is already packed with many support interactions (individual and group sessions, surveys, measurements, supporting materials) and we wanted to limit the work load. However, we observed that the start workshop is such a trigger for heightened motivation, health interests, health plans and a desire to learn more (see also the previous design lesson), that it seemed logical to start the Health Quiz and start tips mails directly. This was implemented in 2015 and the first (very preliminary) results do point to faster micro-learning Health Quiz course progression. Finally, logging health behaviours is very often perceived as a burden (Simons 2012, 2013) even though it may improve health behaviour self-awareness and insights. During the second half of 2014 an improved interface was developed for logging weekly health behaviours (physical activity, mental energy, buddy system and diet). In January 2015 it went ‘live’ and several groups experienced the improvements in comparison to the old interface. The extra uptake (see section 4 Results) and increased ‘mental space’ for learning effects instead of logging efforts, do confirm the ‘persuasive technology’ theories of limiting burdens as much as possible and the eagerness of people to grow and develop (Fogg 2002, 2009). 5.2 Implications for Theory The health support theory that suggests to start health improvement at the peak of motivation (Stull 2007) was confirmed in the sense that waiting list participants did not manage to catch up with the direct start participants in the short term (6 weeks after intervention start). They appear to have missed the opportunity to use their initial motivation. We have to wait for the long term results to know if this difference disappears in the longer run. Another interesting point, is the question what increases intrinsic motivation and helps to exploit it. When there is a health crisis (cancer or other diagnosis) this raises motivation and a majority of patients start implementing one or more health behaviour changes (Stull 2007). This is a negative (and unplanned) event that raises the sense of urgency. However, the hybrid eHealth support program appears to offer a more positive and more planned increase in intrinsic motivation and self-efficacy (Bandura 1997): that it pays rapid dividends to live more healthily. We believe this is an interesting erea for further research of increasing health self-management competences (Simons 2015) via training and positive reinforcement, following the theories of positive psychology (Seligman 2012) and ‘automatic’ healthy choices, perceptions, behaviours and self- assessments (Kahneman 2011). This appears to create a positive spiral of: increased 200 Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health awareness, effective behaviour experiments, increased quality of life and health results, increased competence, increased motivation, eagerness to learn more, and so on. In summary, we can conclude a few key points from our study. First, the hybrid eHealth intervention did significantly improve physical risk factor variables after 6 weeks, and that motivation and measurement alone (waiting list) do not. Second, the timing of the start does matter. Theory that suggests to start health improvement at the peak of motivation (Stull 2007) was confirmed in the sense that waiting list participants did not manage to catch up with the direct start participants in the short term (6 weeks after intervention start). This suggests that offering employees active health support directly after physical measurements (health check-ups) is more effective for improving health than the common practice of focusing on the employee check-up itself. References Bakker AB, Albrecht SL, Leiter MP (2011) Key questions regarding work engagement, European Journal of Work and Organizational Psychology, 20:1, p 4-28. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Baumeister RF, Tierney J. (2011). Willpower: Rediscovering the Greatest Human Strength. New York: Penguin Group. Bruck PA, Motiwalla L, Foerster F. 2012. Mobile Learning with Micro-content: A Framework and Evaluation. Paper presented at the 25th Bled eConference. from www.bledconference.org. CBS (2012) Centraal Bureau voor de Statistiek (Central bureau for Statistics NL, job market update), http://www.cbs.nl/nl-NL/menu/themas/arbeid-sociale- zekerheid/publicaties/arbeidsmarkt-vogelvlucht/structuur-arbeidsmarkt/2006- arbeidsmarkt-vv-participatie-art.htm Demark-Wahnefried, W., Clipp, E., Lipkus, I., Lobach, D. et al. (2007). Main Outcomes of the FRESH START Trial: A Sequentially Tailored, Diet and Exercise Mailed Print Intervention Among Breast and Prostate Cancer Survivors. J Clin Oncol, 25(19), pp. 2709-2718. Demark-Wahnefried W, Jones LW. Promoting a healthy lifestyle among cancer survivors. Hematology-Oncology Clinics of North America. 2008;22(2):319–342. De Vries, H. and Mudde, A. (1998). Predicting stage transitions for smoking cessation applying the Attitude – Social influence – Efficacy Model. Psychology & Health, 13, 369–385 De Vries, H., et al. (2008). The effectiveness of tailored feedback and action plans in an intervention addressing multiple health behaviors. Am J Health Promot. 22(6): p. 417-25. Fogg, B.J. (2002). Persuasive technology: using computers to change what we think and do." Ubiquity, December (2002): 5. 201 Simons, Bodegom, Dumaij, Jonker Fogg, B. J. (2009). A behavior model for persuasive design. Proceedings of the 4th international conference on persuasive technology. ACM, 2009. Jimison, H., Gorman, P., Woods, S., Nygren, P., Walker, M., et al. (2008). Barriers and drivers of health information technology use for the elderly, chronically ill, and underserved. Evid Rep Technol Assess (Full Rep) Nov;(175):1–1422. Jones LW, Demark-Wahnefried W. (2006) Diet, exercise, and complementary therapies after primary treatment for cancer. Lancet Oncol 7(12):1017–26. Kahneman, D. (2011). Thinking, fast and slow. London: Penguin. Lakka HM, Laaksonen DE, Lakka TA, Niskanen LK, Kumpusalo E, Tuomilehto J, and Salonen JT. (2002). The metabolic syndrome and total and cardiovascular disease mortality in middle-aged men. JAMA 288: 2709–2716. Lippke, S., Wiedemann, A. U., Ziegelmann, J. P., Reuter, T. and Schwarzer, R. (2009). Self-efficacy moderates the mediation of intentions into behavior via plans. American Journal of Health Behavior, 33(5), 521–529. Pekmezi DW, Demark-Wahnefried W. (2011). Updated evidence in support of diet and exercise interventions in cancer survivors. ActaOncol 50:167–78 Sattar N, Gaw A, Scherbakova O, Ford I, O’Reilly DSJ, Haffner SM, Isles C, Macfarlane PW, Packard CJ, Cobbe SM, and Shepherd J. (2003). Metabolic syndrome with and without C-reactive protein as a predictor of coronary heart disease and diabetes in the West of Scotland Coronary Prevention Study. Circulation 108: 414–419. Schaufeli WB, AB Bakker, M Salanova (2006) The measurement of work engagement with a short questionnaire, a cross-national study. Educational and Psychological Measurement. 66, 4, pp. 701-716. Schwarzer, R., et al. (2010). Translating intentions into nutrition behaviors via planning requires self-efficacy: evidence from Thailand and Germany. Int J Psychol. 45(4): p. 260-8. Seligman, MEP. (2012). Flourish: A visionary new understanding of happiness and well-being. Simon and Schuster, New York. Simons, LPA. (2006). Multi-channel services for click and mortars: development of a design method. PhD Thesis, Delft University of Technology. Simons, LPA and Bouwman, H. (2004). Designing a click and mortar channel mix. International Journal of Internet Marketing and Advertising 1(3): 229–250. Simons LPA, Foerster F., Bruck PA, Motiwalla L & Jonker CM. (2014). Microlearning mApp to Improve Long Term Health Behaviours: Design and Test of Multi- Channel Service Mix. Paper presented at the 27th Bled eConference. Bled, Slovenia, Proceedings. Retrieval from www.bledconference.org and http://aisel.aisnet.org/bled2014/4 202 Design Lessons from an RCT to Test Efficacy of a Hybrid eHealth Solution for Work Site Health Simons LPA, Foerster F., Bruck PA, Motiwalla L & Jonker CM. (2015). Microlearning mApp Raises Health Competence: Hybrid Service Design. Accepted for: Health and Technology. Simons, LPA, and Hampe, JF. (2010). Service Experience Design for Healthy Living Support; Comparing an In-House with an eHealth Solution. Paper presented at the 23rd Bled eConference. Bled, Slovenia, from www.bledconference.org. Simons, LPA and Hampe, JF. (2010b). Exploring e/mHealth Potential for Health Improvement; A Design Analysis for Future e/mHealth Impact. Paper presented at the 23rd Bled eConference. Bled, Slovenia, from www.bledconference.org. Simons, LPA, Hampe JF, and Guldemond NA. (2012). Designing Healthy Consumption Support: Mobile application use added to (e)Coach Solution. Paper presented at the 25th Bled eConference. Bled, Slovenia, from www.bledconference.org. Simons LPA, Hampe JF, Guldemond NA. (2013). Designing Healthy Living Support: Mobile applications added to hybrid (e)Coach Solution, Health and Technology, 3 (1), pp.1-11. Simons LPA, Hampe JF, Guldemond NA. (2014). ICT supported healthy lifestyle interventions: Design Lessons. Electronic Markets. 24 pp. 179-192. DOI 10.1007/s12525-014-0157-7. Simons, LPA, Steinfield C and Bouwman H. (2002) "Strategic positioning of the Web in a multi-channel market approach." Internet Research 12 (4): 339-347. Slobbe LCJ, Smit JM, Groen J, Poos MJJC, Kommer GJ (2011) Cost of Illness in the Netherlands 2007 : Trends in healthcare expenditure 1999-2010, Rijksinstituut voor Volksgezondheid en Milieu RIVM, Centraal Bureau voor de Statistiek CBS, part of the RIVM official reports collection, http://hdl.handle.net/10029/262360, downloaded from http://www.rivm.nl/bibliotheek/rapporten/270751023.pdf Sperling, R., L.P.A. Simons and H. Bouwman. (2009). Multi-Channel Service Concept Definition and Prototyping, International Journal of Electronic Business, 7 (3), pp.237–255. Starr, J. (2008). The coaching manual: the definitive guide to the process, principles and skills of personal coaching. New York, Prentice Hall. Steenbeek R, Hooftman W, Geuskens G, Wevers C. (2010). ‘Objectiveren van gezondheidsgerelateerde nonparticipatie en de vermijdbare bijdrage van de gezondheidszorg hieraan’ TNO, Hoofddorp. Stull, V., Snyder, D. and Demark-Wahnefried, W. (2007). Lifestyle Interventions in Cancer Survivors: Designing Programs That Meet the Needs of This Vulnerable and Growing Population. J Nutr, 137, pp. 243S-248S. Verweij, LM, Coffeng, J., van Mechelen, W., Proper, KI (2011). Meta-analyses of workplace physical activity and dietary behaviour interventions on weight outcomes. Obesity Reviews, 12, 406-429. 203 Simons, Bodegom, Dumaij, Jonker Ware Jr, J.E. and Gandek, B. (1998). "Overview of the SF-36 health survey and the international quality of life assessment (IQOLA) project." Journal of clinical epidemiology 51.11 (1998): 903-912. Wiedemann, A. U., Lippke, S., Reuter, T., Ziegelmann, J. P. and Schwarzer, R. (2011). How planning facilitates behaviour change: Additive and interactive effects of a randomized controlled trial. European Journal of Social Psychology, 41, 42–51. Willett WC. (2004). The Search for Optimal Diets: A Progress Report. Commemorative Lecture at the 25th Honda Prize Awarding Ceremony. Willett WC and Ludwig DS. (2011). The 2010 Dietary Guidelines — The Best Recipe for Health? N Engl J Med; 365:1563-1565. 204 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Standardisation of risk screening processes in healthcare through business rules management Joris Mens; Sander Luiten; Yannick Driel; Kobus Smit; Pascal Ravesteyn HU University of Applied Sciences Utrecht, The Netherlands joris.mens@hu.nl; sander.luiten@student.hu.nl; yannick.driel@student.hu.nl; kobus.smit@hu.nl; pascal.ravesteijn@hu.nl Abstract In 2012, an audit held by the Netherlands Institute for Accreditation in Healthcare (NIAZ) at the ‘Rivierenland’ hospital in The Netherlands, concluded that their processes were not sufficiently standardised. One of the suggested improvements was to develop and implement a hospital-wide method for analysing and standardising care processes. This paper focuses on the standardisation of the risk screening process, which is used to assess a number of patient risk factors prior to treatments or hospital admissions. By separating the decision logic of the risk screening processes into a set of business rules, the screening process was standardised to be identical for each risk factor. This allows for the decision logic and the process to be changed independently of each other. Additional business rules were introduced to serve as constraints, thereby limiting the number of performed screening processes depending on the age of the patient and the duration of the treatment or admission. Based on historical data from the year 2013, a retrospective analysis demonstrated potential time savings of around 1600 hours on a yearly basis thanks to the introduction of the new standardised process incorporating business rules. Similar standardisation methods may be useful to other hospitals facing increasingly stringent demands for quality, safety and efficiency. Keywords: Healthcare, business process management, standardisation, risk screening, business rules 1 Introduction In The Netherlands, reforms in the healthcare sector are increasing pressure on healthcare providers to provide high quality care in a decentralized and competitive market (Øvretveit, 2000). The variety of specializations and therapies is on the rise, while patients demand higher quality services and shorter waiting times. In response to requirements imposed by the government and accreditation bodies, hospitals must be able demonstrate transparency in the safety and quality of their healthcare processes (Government of The Netherlands, 2012). Adequate process management is included in current accreditation frameworks for the Dutch hospital sector (Netherlands 205 Mens, Luiten, Driel, Smit & Ravesteyn Institute for Accreditation in Healthcare, 2013). International accreditation bodies such as the Joint Commission International (JCI) take an even more rigorous approach by demanding continuous process improvement for ensuring patient safety and efficient, standardised healthcare. To transform into process-driven organisations, hospitals must continuously adapt and improve processes according to market demands. Information systems needed to support these processes are found to be relatively underdeveloped when compared to other sectors (Helfert, 2009), particularly in terms of low technological sophistication and integration sophistication (Paré & Sicotte, 2001). However, technology itself cannot provide a solution without taking the process into account (Jaana, Tamim, Paré, & Teitelbaum, 2011). The Rivierenland hospital studied in this paper was struggling with a similar situation. In 2012, the hospital’s accreditation by the NIAZ (The Dutch institute for accreditation in healthcare) was extended, but a critical note in the accreditation report was that the hospital’s processes were not sufficiently standardised. Some of the necessary technology to support the processes, such as a business rules engine, were already available but not utilised due to a lack of a process-driven approach. While an accreditation by NIAZ is not legally required to be able to provide care in The Netherlands, it serves as a mark of quality for healthcare providers and may be demanded by insurers. Accreditations are granted for a period of four years, after which a new accreditation is performed. At the Rivierenland, processes and their related activities were described in different formats and there was a lack of coherence between processes. One of the improvements suggested by the accreditation body was the analysis and standardisation of these processes. The hospital’s primary process is the examination and treatment of patients. One of the first activities performed when a patient is admitted is risk screening. Patients may be exposed to a number of risks, both during admission and treatment. For example, A patient lying still in a bed for too long may develop decubitus (pressure ulcers). If a patient is found to be at high risk for developing decubitus, measures are taken such as frequent repositioning of the patient or the installation of a special mattress. All activities related to the identification of risks, as well as the introduction of measurements to prevent these risks are labelled as the ‘risk screening and prevention process’. In this study, literature, documentation, interviews and observations are used to assess the current state of the risk screening and prevention process and to introduce a new and standardised process, which adheres to the quality requirements of the accreditation body. Potential time savings are expected, as a standardised process will lead to a more efficient execution of activities related to risk screening and prevention. The next section describes literature studied to gather insight into process standardisation in healthcare. In section three the research approach is described followed by an overview of the standardized process with the use of business rules in section four. The possible efficiency gain is shown in section five. In the final section a conclusion and discussion are provided. 206 Standardisation of risk screening processes in healthcare ... 2 Literature Review In order to identify which requirements and benefits are related to standardisation of processes in healthcare, a number of previous studies are reviewed. Standardisation has been applied with positive results in many different specialisations of healthcare. A study performed by Rozich et al. (2004) showed that the introduction of a standardised protocol for insulin administration in diabetes patients lead to a reduction in hypoglemic episodes from 2,95% to 1.1% over a period of 30 months, as well as a decrease in medication errors from 213 errors per 100 admissions to fewer than 50 per 100 admissions. The protocol was developed as a joint effort by various medical specialists, and includes a number of measurements such as the patient’s weight and the number of insulin units the patient takes in one day. Based on this patient data, the amount of medication needed can be determined on a sliding scale. In essence, the protocol ensures that patients are treated according to an agreed-upon set of business rules. Rozich et al. (2004) posit that standardisation of this process lead to reduced complexity, increased safety and possible cost savings. They recommend similar efforts to be taken in other clinical areas. A study by Arora & Johnson (2006) identified and standardised the hand-off process, which is concerned with care transitions such as patients going from one department of a hospital to another or shift changes of nurses. The hand-off process is critical to patient safety, as inadequate communication of patient information in care transitions may lead to the unintentional discontinuation of essential medication (Bell et al., 2011). Arora & Johnson (2006) show that the first step in standardising the process is identifying the process and its possible variations. By creating awareness, possible vulnerabilities can be detected and corrected. Building a standardised checklist was found to be instrumental in improving patient care. In the aforementioned studies, the importance of an agreed-upon protocol is established. These protocols usually consist of a certain process or procedure, prescribing the order of activities to be performed. Additionally, checklists or measurements provide information needed to support decisions. This knowledge can also be described as a set of ‘business rules’. A business rule is defined by Ross (2003) as “An atomic piece of re-usable business logic, specified declaratively”. As per the Business Rules Group (2015), a business rule is “a statement that defines or constrains some aspect of the business. It is intended to assert business structure, or to control or influence the behaviour of the business.” In the case of healthcare organizations, business rules are found to be present in deciding the type of medication given to a patient, for example. Another motive for the use of Business Rules is flexibility. By separating the order of activities (the process sequence) from the knowledge needed to support decisions in the process, these can be changed independently to respond to internal or external demands (Spreeuwenberg, 2004). The process models are often modelled using UML activity diagrams or the Business Process Modelling Notation (BPMN) (Goedertier & Vanthienen, 2006). BPMN is a standard for modelling business processes in a graphical manner using a business process diagram. This is done to clarify the management of business processes and in such a way that it is both understandable for technical users and non-technical users (Weske, Hofstede, & van der Aalst, 2003; White, 2004). Both 207 Mens, Luiten, Driel, Smit & Ravesteyn BPMN and Business Rules will be used in this study to aid the standardisation of the risk screening process. 3 Approach To assess the current situation concerning the execution and documentation of the risk screening process, different methods were used. The current documentation regarding the risk screening process was studied and a number of interviews and observations were conducted to assess how the process is executed in practice. While interviews provide insight into the experiences of the staff, observations will enhance our understanding by looking at what actually happens in the clinical setting (Fox, 1998) The Rivierenland hospital stores its documentation on an intranet portal accessible to staff within the hospital. This portal hosts four types of documents that relate to the risk screening process, namely (1) process models, (2) standards of care, (3) decision trees and (4) care protocols. The standards of care are imposed by external in regard to certain quality standards to which the process must adhere. Care protocols are developed internally and provide a more detailed step-by-step description of procedures that must be taken in providing care. The risk screening process is subdivided into the risk factors decubitus, delirium, falling, malnutrition and physical disability. The researchers were granted access to this internal portal for the duration of this study. To gather more information about the current (as is) situation within the hospital as well as the desired (to be) situation, interviews were held with staff from the quality management department. This provided further information on the boundaries within which the risk screening process must be executed as well as contacts with people in the workplace for our observations. The information provided by the quality management department serves as the guidelines to which the process must adhere. In addition, the quality management department provided historical data for the previous year, which were subsequently used for benchmarking and estimating the potential efficiency gain in utilising a standardised process. In the workplace, observations were made to assess the execution of the process in practice. In this process a nurse normally conducts anamneses during the intake of a patient prior to treatment or admission. During the observation, the time taken to screen the patient for each risk was recorded so that an estimate can be made for the total time spent screening all patients. The observation also provided information about the questions that are asked to the patient during their intake and revealed if there are any deviations from the documented protocols. The abovementioned information was be combined to create (1) a standardised process model for the risk screening process that includes all five risk factors and (2) a set of business rules that serve as directives on the decisions taken during the process. 4 Results Through the use of a BPMN diagram, this section demonstrates the differences between the as-is situation and the to-be situation regarding a standardised risk screening process. This is followed by the presentation of a set of business rules to constrain the risk screening process depending on patient characteristics. Following the 208 Standardisation of risk screening processes in healthcare ... demonstration of the process model and the business rule set, the potential timesavings resulting from an implementation of the standardised process are estimated. In previous research conducted at the Rivierenland hospital (Hau and Ilbey, 2014), a first step was made towards documenting a standardised process model. This process model is shown in Figure 1. Figure 1: The as-is process model for the risk screening process (Hau and Ilbey, 2014) The process model shown in Figure 1 consists of one high-level process containing two sub processes. The high level process encompasses the activities conduct anamnesis, risk screening, conduct preventive interventions and observe patients. The ‘risk screening’ activity constitutes a sub process for specific risks. The ‘observe patient’ activity is a repeating process (indicated by the circular arrow) in which changes in risk factors are observed for a patient who is undergoing care. The process model demonstrates that preventive interventions are applied when a patient is found to be at risk for developing complications. Patients who are at risk are then continuously monitored for changes in their risk factors. Based on the interviews with staff from the quality management department, it was found that this process could be further simplified. The sub process ‘risk screening’ was found to be redundant, as the risks are already screened for during the ‘conduct anamnesis’ activity. It therefore not necessary to explicitly mention these activities in a sub process and it was removed. The second sub process, ‘observe patient’ was also simplified by merging the activities ‘change nursing plan’ and ‘change/add preventive 209 Mens, Luiten, Driel, Smit & Ravesteyn interventions’. This was done because preventive interventions are described within the nursing plan, and therefore a change in interventions already implies a change in the nursing plan. Based on these changes, the simplified process model as shown in Figure 2 was created. Figure 2: The to-be process model for the risk screening process To achieve a standardised process, the process must incorporate the five risk factors decubitus (pressure ulcers), delirium, falling, malnutrition and physical disability. While the activities for each of the risk factors remain the same, the variations in measurements that need to be performed for each risk factor are different (based on care protocols) and can therefore be supported by business rules. These business rules are captured in a decision tree specific to each risk factor. The decisions trees incorporate industry-standard rating scales for determining the severity of the risk. In the case of decubitus, this is done according to the Braden scale (Bergstrom, Braden, Laguzza, & Holman, 1987). Based on the severity of the risk, the decision tree prescribes the use of a specialised mattress or frequent movement of the patient. Apart from the business rules related specifically to the risk factors, a new set of business rules was introduced to constrain when certain risk factors should or should not be screened for. According to current protocols each patient needs to be screened for all risk factors, despite some risk factors not being relevant to the patient, depending on their age, the duration of their treatment or admission and other characteristics. Patients have a higher risk to develop complications if they are present for a longer time in the hospital. In the case of an admission with a maximum duration of one day (day treatment) or treatment in the policlinic, the duration of the admission is too short to develop pressure ulcers, for example. Based on interviews with the quality management department and observations in the workplace, it was determined that 210 Standardisation of risk screening processes in healthcare ... only clinical admissions lasting longer than one day should incorporate risk screening. These rules are represented in Table 1. 211 Mens, Luiten, Driel, Smit & Ravesteyn Conditions Conclusion Rule pattern Admission type Conduct risk screening? 1 = Policlinic Is No 2 = Day treatment Is No 3 = Clinical Is Yes Table 1: Business Rules constraining the risk screening process based on admission type Based on the patient’s age, the risk screening process is further constrained. Younger patients are deemed to be of low risk for developing certain risks factors. The business rule set represented in Table 2 shows which risk factors are screened for depending on the age group of the patient. Conditions Conclusion Patients age Conduct risk screening? Rule pattern ion us al ity um g nutrit ecubit eliri lin Mal D Physic disabil D Fal 1 [ ] 0-18 Is X 2 [ ] 18-70 Is X X 3 ≥ 70 Is X X X X X Table 2: Business Rules constraining the risk screening process based on patient age Based on these business rules, a nineteen-year-old patient coming in for clinical treatment must be screened for the risk factors malnutrition and decubitus. This is then done according to the decision trees specific to each risk factor. 5 Efficiency through standardisation In this section an analysis based on historical admission data of the Rivierenland hospital over the year 2013 is presented. Based on this data, the potential efficiency gain when implementing the proposed standardised process was calculated. The admission data in Table 3 shows a total of 27,290 admissions over all age groups and admission types. For each admission, it is assumed that in the current situation (and according to protocol) patients are screened for the five risk factors. This amounts to a total of 126,450 risk screenings. 212 Standardisation of risk screening processes in healthcare ... Age Day Clinical Total category admissions admissions admissions 0-18 y 1,512 1,772 3,284 18-70 y 8,320 8,159 16,479 >70 y 3,759 3,768 7,527 Total 13,591 13,699 27,290 Table 3: Ziekenhuis Rivierenland admission data 2013 By applying the business rules proposed in the previous section it may be possible to reduce the number of redundant risk screenings that are performed, thereby improving efficiency. First off, the risk screening process can be eliminated for day admissions, thereby reducing the total number of admissions by 13,591. The clinical admissions will include risk screening for specific factors based on the patient’s age. Table 4 presents a summary of the number of risk screenings with and without the proposed business rules. The number of risk factor screenings is calculated by multiplying the number of admissions times the number of risk factors. In the as-is situation, this includes risk screenings for all admission types. In the to-be situation, this includes only risk screenings for clinical admissions. By reducing the number of risk factors screened for according to age category and by only performing risk factor screenings for clinical admissions, a total reduction of risk factor screenings of 72.94% is achievable. As-is To-be Total Risk Total Risk Age Risk Risk Reduction Factor Factor category factors Factors percentage Screenings Screenings 5 16,420 1 (clinical 1,772 89.21% 0-18 y only) 5 82,395 2 (clinical 16,318 80.20% 18-70 y only) 5 37,635 5 (clinical 18,840 49.94% >70 y only) Total 136,450 36,930 72.94% Table 4: Summary of conducted RSP’s To calculate the potential timesaving’s associated with the reduction of risk factor screenings, a calculation is presented in Table 5. Based on the observations conducted in this study, the assumption is made that each risk factor screening takes approximately one minute of time and that all risk factors screenings are conducted according to protocol. This implies that an anamnesis for one patient including all five risk factors takes approximately five minutes. Table 5 summarizes the amount of time taken to execute al risk screenings in the as-is situation compared to the to-be situation. It is concluded that this leads to potential time savings of more than 1600 213 Mens, Luiten, Driel, Smit & Ravesteyn hours on a yearly basis. Admissions with Time taken (hours) risk screenings As-is situation 27,290 2,274.14 (all age categories) To-be situation 0-18 years 1,772 29.53 18-70 years 8,159 271.97 > 70 years 3,768 314 To-be situation (total) 13,699 615.5 Potential time saving 1,658.67 Table 5: Time reduction by using Business Rules 6 Conclusion & Discussion The standardised process model proposed in this study has been shown to successfully include all five risk factors by separating the business logic from the process model using sets of business rules. This has improved the transparency in the hospitals business processes and also made them more manageable. Business rules used to further constrain the risk screening process based on type of admission and patient age category help to improve efficiency by eliminating redundant risk screenings. Currently, the protocols used in the workplace are contained in an intranet portal used by hospital staff. The documentation hosted on this portal will need to be updated to reflect the proposed standardised process and to be able to determine the practical efficacy. At the time of writing, this change has not yet been achieved. The actual implementation of the new standardised process is expected to be a challenge. Firstly, IT systems have to be configured to support and enforce the prescribed business rules. Secondly, it remains to be seen to which extent the prescribed process will align with the activities in practice. As was seen in the observation, not all risk factor screenings are performed for all patients, despite this being required according to protocol. Nursing staff do also use their own insights to determine which risk factors are unnecessary to be screened for, depending on the characteristics of the patient and the admission or treatment. In this regard, the paper provides a very ‘black and white’ comparison between a very inefficient ‘as-is’ situation and a potentially very efficient ‘to-be’ situation. In reality, the differences may be much smaller. Despite these facts, the hospital will still need to consider the application of IT systems to gain better control of and insight into processes into the organization. Without these efforts, a true process-driven organization cannot be achieved. This study provides a starting point for the transformation into a standardised, process-driven organization. 214 Standardisation of risk screening processes in healthcare ... References Arora, V., & Johnson, J. (2006). A model for building a standardized hand-off protocol. Joint Commission Journal on Quality and Patient Safety, 32(11), 646–655. doi:10.1007/s11606-009-1170-y Bell, C. M., Brener, S. S., Gunraj, N., Huo, C., Scales, D. C., Bajcar, J., … Urbach, D. R. (2011). Association of ICU or Hospital Admission of Medications for Chronic Diseases. JAMA, 306, 840–847. doi:10.1001/jama.2011.1206 Bergstrom, N., Braden, B. J., Laguzza, A., & Holman, V. (1987). The Braden Scale for Predicting Pressure Sore Risk. Nursing Research, 36(July/August), 205–210. doi:10.1097/01.ASW.0000411403.11392.10 Business Rules Group. (2015). Defining Business Rules. Retrieved February 14, 2015, from http://www.businessrulesgroup.org/defnbrg.shtml Fox, N. (1998). How to Use Observations in a Research Project. Retrieved from http://web.simmons.edu/~tang2/courses/CUAcourses/lsc745/sp05/observation.pd f Goedertier, S., & Vanthienen, J. (2006). Bedrijfsregels voor conforme en flexibele bedrijfsprocessen. Business In-Zicht, 2–3. Retrieved from https://lirias.kuleuven.be/bitstream/123456789/120493/1/bi.nr21-2.pdf Government of The Netherlands. (2012). Coalition agreement cabinet Rutte-Asscher. Retrieved from http://www.government.nl/files/documents-and- publications/reports/2012/10/29/coalition-agreement/coalition-agreement.pdf Øvretveit, J. (2000). Total quality management in European healthcare. International Journal of Health Care Quality Assurance, 13(2), 74–80. doi:10.1108/09526860010319523 Ross, R. G. (2003). Principles of the Business Rule Approach. Boston: Addison-Wesley. Rozich, J. D., Howard, R. J., Justeson, J. M., Macken, P. D., Lindsay, M. E., & Resar, R. K. (2004). Standardization as a mechanism to improve safety in health care. Joint Commission Journal on Quality and Safety, 30(1), 5–14. Spreeuwenberg, S. (2004, June). Beheer bedrijfsregels vergroot flexibiliteit. Informatie, 12–18. Retrieved from http://librt.home.xs4all.nl/downloads/0405-12Spr.pdf Weske, M., Hofstede, a. H. M., & van der Aalst, W. M. P. (2003). Business Process Management: A Survey. Lecture Notes in Computer Science, 2678, 1–12. doi:10.1007/3-540-44895-0_1 White, S. A. (2004). Introduction to BPMN, (c), 1–11. 215 Mens, Luiten, Driel, Smit & Ravesteyn 216 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Application of Lifetime Electronic Health Records Are we ready yet? Kai Gand Technische Universität Dresden, Germany kai.gand@tu-dresden.de Peggy Richter Technische Universität Dresden, Germany peggy.richter2@tu-dresden.de Werner Esswein Technische Universität Dresden, Germany werner.esswein@tu-dresden.de Abstract Integrated care concepts can help to diminish demographic challenges. Therefore, the use of eHealth solutions is recognised as an efficient approach. Lifetime electronic health records (LEHRs) are expected to increase continuity, effectiveness, efficiency and thus quality of the care process. With respect to these benefits, an overarching implementation of LEHRs is desirable but non-existent. Hence, the aim of the article is to analyse the current LEHR implementation readiness of EU member states to derive implications for further LEHR research and development. Therefore, a case study on Denmark, Germany and Italy was conducted. The analysis shows that all countries fulfil the technical requirements but Denmark has great experiences and willingness to implement advanced eHealth measures like LEHRs. First Italian pilot projects are quite promising as well. The article paves the way for LEHR implementation and therewith for integrated care. Keywords: Electronic Health Record, Personal Health Record, Implementation, Requirements, Integrated Care 1 Introduction The health care systems in Europe are faced with challenging transition processes due to demographic change, which leads to skilled worker shortage and an increasing number of multi-morbid patients (Harper, 2010). Given these challenging circumstances, integrated care can improve and ensure the quality of care (Lerum & Frich, 2012). An IT-based solution is a feasible approach to reduce the bottleneck of human resources and budgets (Iakovidis, 1998). Since the sketched changes are a Europe-wide challenge, the question raises whether there will be Europe-wide solutions. Anyway, EU’s policy aims at spreading IT-based systems in health care by means of electronic health records (EHRs) all across Europe by the end of the decade (Kierkegaard, 2011). The progress in information and communication technologies (ICT) and eHealth solutions offer various options to meet the change and enable integration and networking (Dixon, 2007). A much higher level of integrated care all 217 Kai Gand, Peggy Richter, Werner Esswein along an individual’s lifetime is possible. Lifetime EHRs (LEHRs) are seen as a tool to support this vision. An LEHR is a “lifelong electronic collection, storage and provision of all health related information about its owner, allowing integrated care and functioning as a data basis to improve the quality of health care on the individual and societal level” (Gand, Richter, & Esswein, 2015). They shall be used across institutions and sectors in health care and include not only medical information, but also information on alternative treatments, lifestyle etc. LEHRs differ from EHRs, because these typically do not cover the whole lifespan and represent a provider-based view. LEHRs also differ from personal health records, since the information is primarily collected, managed and used by the owning individual alone (Caligtan & Dykes, 2011; Tang et al., 2006; Waegemann, 2002). The vision of LEHRs is expressed by a number of synonyms such as EHRs for integrated care (International Organization for Standardization, 2005), lifelong personal health records (Barbarito et al., 2015), lifelong integrated EHRs (Katehakis et al., 2007) or lifelong virtual EHRs (van der Linden et al., 2009). LEHRs are expected to reduce information asymmetries by empowering the individual, increase continuity, quality, patient safety, effectiveness and efficiency along the care process and thus reduce costs as well as redundant work (Chaudhry et al., 2006; Katehakis et al., 2007; Tang et al., 2006). However, an area-wide implementation of (L)EHRs in Europe is not known so far. There are regional initiatives but no overarching approach. The interoperability of different solutions is not given. Nonetheless, the need for large-scale sharing of medical data has been expressed (Stroetmann et al., 2011). Research on LEHRs is rare so far, whereas EHRs are studied in more detail (Caligtan & Dykes, 2011; Häyrinen, Saranto, & Nykänen, 2008; Kukafka et al., 2007). The main LEHR characteristic (lifetime validity) raises specific research questions, such as on sustainability and data retention. Various organisational approaches for LEHR delivery have been analysed with respect to ethical and legal issues. The establishment of Independent Health Record Banks has been proposed for sustaining LEHRs, because neither the consumers nor the care providers seem to be capable of providing the compilation and a sustainable storage (Shabo, 2006, 2010). With regard to technological infrastructure, a federated architecture was argued to be best fitting (Tsiknakis, Katehakis, & Orphanoudakis, 2004) and a service-oriented architecture for LEHR delivery was demonstrated being appropriate as well (Katehakis et al., 2007). The divergence between recognised benefits and yet non-application of LEHRs makes advancement of LEHR implementation worth pursuing. Since a “one-size-fits-all” approach is not recommended in eHealth (Currie & Seddon, 2014), the analysis of country specifics is necessary. Therefore, the article aims at qualitatively assessing the current state of implementation readiness of EU member states by a case study deriving implications for further LEHR research and development. The method and country selection are described in section 2. In section 3, the criteria used for the assessment of LEHR implementation readiness are presented and demonstrated. The results are summarised in section 4. The paper closes with a discussion on open issues in section 5. 2 Method The case study is conducted for a sample of EU members demonstrating LEHR implementation readiness criteria with real countries’ data and to handle the multitude of aspects by reducing real world’s complexity (Yin, 2014). The countries were selected based on an even geographical distribution and the results of the cross- national eHealth analysis by Currie & Seddon (2014). This quantitative study assessed EU countries on two dimensions (ICT penetration and availability; eHealth access and 218 Application of Lifetime Electronic Health Records usage) and identified four distinct groupings with declining dimension values: frontrunners, followers, leapfroggers and laggards (Currie & Seddon, 2014). Denmark (frontrunner), Germany (follower) and Italy (leapfrogger) were chosen as they are geographically close centrally in Europe but also represent north, middle and south European countries. At first view, the categorisation supports the assumption of Italy being not ready for the introduction of LEHRs yet. However, the vision of LEHRs has already been regionally implemented (Barbarito et al., 2015). This makes an analysis even more interesting, because it allows the assessment of the chances for countrywide dissemination. The information for the assessment was collected through desk research. An argumentative-deductive approach in terms of a literature analysis (Palvia et al., 2003) has been conducted. 3 Case study 3.1 Assessment criteria for LEHR readiness Numerous articles already analysed non-functional requirements for (L)EHR systems, such as data security and integrity, authenticity, availability, portability, performance and efficiency, maintainability, reliability, and usability (Fernández-Alemán et al., 2013; Hoerbst & Ammenwerth, 2010; Iakovidis, 1998; van der Linden et al., 2009). As the present article focuses on implementation preconditions on country-level, such system-specifics are not of primary interest. The case study uses the partly adapted assessment criteria for LEHR readiness as proposed by Gand et al. (2015). They consider not only IT-based indicators (such as Currie & Seddon, 2014) but also socio- economic ones. The criteria are described in Table 1. Criterion Description Culture Culture is a question of overarching societal willingness and awareness. A faster and more extensive communication across the borders of distinct health care providers and the collection of data from every necessary or available source should be considered as reasonable and thus be practiced whilst sensitising for potential risks. Regulation & Regulations regarding the functioning of an LEHR system ensure the use of common Govern- standards and long-term interoperability. The documentation systems should be mental compliant with data protection laws without hampering necessary data exchange. Commitment Documentations should stand up in court (auditability) not only having informative character. Only authorised and auditable data accesses with health care related intensions are acceptable. Avoiding misuse by alert regulatory is recommended. Privacy aspects and informational self-determination are of high importance due to the intimacy of the collected data. Focussing on the rights (informed consent) and needs of the citizens is highly important. Positive regulation should accompany with governmental commitment promoting the advantageousness of LEHRs. Incentives Measurable incentives and benefits are important to change the long-term behaviour and therefore the acceptance of new approaches like an LEHR. Health care providers and consumers have to use ICT in their daily routines. This might be incentivised by the provision of explicit (monetary) benefits (or fines) or by laying down usage rules. Compliance Care process related regulations like continuous quality assurances and system evaluations are necessary to enable compliant integrated health care. LEHRs should be integrated in clinical practice and protocols (and in individuals’ daily routines). Decision support systems are a way to ensure consistent care by accessing an overarching database storing process and consumer related information. ICT Infra- A basic ICT infrastructure (i.e. ICT penetration and availability) is necessary to 219 Kai Gand, Peggy Richter, Werner Esswein structure implement LEHRs on the technical level. The focus lays on the existence of connectivity options between different health care providers and citizens and complementing infrastructure for authentication etc. Standards Open and universal standards, common data models/formats, terminology and the consistent use of those are necessary prerequisites for interoperability, authentication, data security, long-term retention and hence to exchange health data across several health care institutions and sectors. These also ensure the trust in the system’s capabilities and therefore acceptance. Table 1: Assessment criteria for LEHR readiness (Gand et al., 2015) The criteria show, that the government can establish the basic preconditions for successful LEHR implementation. 3.2 Country assessment The application of the criteria is demonstrated by assessing the present situation in Germany, Denmark and Italy. Possible reasons for differences regarding eHealth implementation will be examined. 3.2.1 Culture In Denmark there is the aspiration to become a highly IT-based society putting health care on a digital basis. This is based on deep-rooted openness for new technologies all across the society which in turn leads to a fruitful basis for the implementation and adoption of new eHealth solutions (Kierkegaard, 2013; Protti & Johansen, 2010; Currie & Seddon, 2014). In contrast, a definite separation between German health care providers leads to inflexible, not very permeable borders between different sectors. Huge differences between the professional cultures of different occupation groups further hinder interdisciplinary and cross-sectoral cooperation. A rather low willingness to change structures and realise innovative care concepts is present (Amelung & Janus, 2005; Degeling, Maxwell, Kennedy, & Coyle, 2003). Moreover, Germany has a large population and a highly developed well performing health care system. But this comes along with a higher level of conservatism and difficulties managing large-scale eHealth systems on the national level (Currie & Seddon, 2014; Stroetmann et al., 2011). In contrast, initial studies show, that there might be a majority of citizens having a positive attitude regarding the implementation of overarching EHRs. However, concerns regarding privacy and data protection, uncertainties about the concrete features and accompanying risks are also highly relevant issues (Hoerbst et al., 2010). Despite the plan of the Italian government to push forward innovative eHealth concepts (Barbarito et al., 2015), the information exchange between different health care providers in the regions is not fully satisfying (Bonacina, Marceglia, & Pinciroli, 2011). A way to raise quite narrow adoption and user interests might be the integration of features allowing interpersonal cooperation and exchange focussing on user empowerment (Cabitza, Simone, & De Michelis, 2015; Comandé, Nocco, & Peigné, 2015). 3.2.2 Regulation & Governmental Commitment The Danish government has the goal to spread integrated care concepts and mutually useful patient-carer interactions by means of eHealth solutions. Therefore, MedCom, an institution to coordinate eHealth actions, was founded in 1994 (Deutsch, Duftschmid, & Dorda, 2010) showing a strong commitment to eHealth advantageousness. Denmark also takes an exemplary role in designing an overall legal framework for sustainable eHealth solutions considering important aspects like privacy, confidentiality, liability and data protection (Stroetmann et al., 2011). The 220 Application of Lifetime Electronic Health Records government aspires a high degree of patient independency and empowerment (Kierkegaard, 2013). The use of EHRs is mandatory and they are well accepted by the physicians (Protti & Johansen, 2010). The German government promotes integrated care projects (see § 140a ff. German Social Act Five), but the main focus lays on integration on the indication level until now (Schreyögg, Weinbrenner, & Busse, 2006). This leaves room for higher levels of integration. Because informational self-determination is ranked as a constitutional fundamental right, Germany has strong data protection laws. These can hamper cross- sectoral data exchange and integrated care approaches (Amelung & Janus, 2005; Menzel, 2006). A project to implement an electronic health card ought to promote integrated solutions, but demonstrated eHealth to be a tough act to follow in Germany: the project was controversially discussed and delayed for many years; the range of functions is still limited (Engemann, 2013). The Italian law on privacy and security is also quite a hurdle for innovative overarching eHealth solutions. Here, special effort for clarification and introduction of regional implementation guidelines are a precondition for the application of those. Another issue is the split of responsibility in the Italian legislative structure: initiatives of the central level pushing innovative eHealth concepts (the broad implementation of the “Fascicolo Sanitario Elettronico” as an overarching EHR) are highly desirable. But the regional responsibilities for the real implementation of those laws in health care hamper an overarching implementation (Barbarito et al., 2015). 3.2.3 Incentives Considering the high Danish governmental (and societal) commitment, there seems to be low need for incentives to use eHealth. Nonetheless, there were some helps to boost these especially at the beginning of MedCom’s work (e.g. data consultants, peer influence and collegial pressure, funding by the ministry and physicians’ education seminars). This led to a high rate of eHealth use by practitioners even before the mandatory phase begun (Protti & Johansen, 2010). The German social security laws primarily offer incentives based on cost reductions for patients (e.g. for taking preventive actions) and health care providers (e.g. no-name drug prescriptions allowing the cheapest medication). Sustainable behavioural change is not a major goal (Schmidt, Gerber, & Stock, 2009). Latest legislative measures include incentives (and sanctions) for a dissemination of eHealth solutions and the implementation of a uniform infrastructure for telematics and interconnections in health care ( Draft of a law for secure digital communications and applications in health care, 2015). In Italy, incentives for the practitioners (presumably no experts for IT or documentation) to get them use new eHealth solutions are also a quite new issue. Incentive payment schemes are regionally implemented to overcome this. Other prospective ideas are special educations and training programmes. Furthermore, mandatory goals for the specific use of eHealth solution shall play an important role, too (Barbarito et al., 2015; Comandé et al., 2015). 3.2.4 Compliance The use of EHRs in Denmark is mandatory and these are well accepted in general, which in turn leads to a highly compliant use. Nevertheless, there are still frictions in the physicians’ work due to eHealth usage. Acceptance problems and technical shortcomings might have been undervalued. But the authorities permanently observe these considerations, so improvements are planned (Grosen, 2009; Kierkegaard, 2013; Protti & Johansen, 2010). 221 Kai Gand, Peggy Richter, Werner Esswein The increasing publication of highly evidence-based medical guidelines is a way to obtain compliance within the health care process. But the voluntary usage, and insufficient linkage to practical implementation in clinical practise guidelines are open issues. This enforces the argument for mandatory regulations establishing a higher rate of integrated care (Kopp, 2011; Perleth, Jakubowski, & Busse, 2000). In Italy, the necessity to develop or use established interoperability guidelines has also been recognised as relevant. Where introduced, they (together with mandatory goals) had a positive impact on the adoption of newly designed eHealth solutions (Barbarito et al., 2015). In contrary, the absence of guidelines and process management is the suboptimal standard. Educational efforts are also necessary to get systems adopted. The effort to design new workflows that integrate and adapt existing standards, clinical and administrative processes and practitioners’ work was inadequately considered so far and resulted in a lacking diffusion (Barbarito et al., 2015; Bonacina et al., 2011). 3.2.5 ICT Infrastructure On the technical level, Denmark is highly competitive. The disciplines of health care and informatics work very closely together, there is broad internet access and usage almost across the whole country (Currie & Seddon, 2014). The use of EHRs is mandatory since 2004 (Kierkegaard, 2013; Protti & Johansen, 2010). There are also several national and regional strategies to reach the IT-related goals. These result in a multitude of distinct health information exchange and storage initiatives, platforms and portals covering the majority of clinical relevant processes (Kierkegaard, 2013). Although Germany performs less well on eHealth indicators (usage and access), it has a mature health infrastructure, an excellent medical technology sector and performs well on ICT indicators (Currie & Seddon, 2014; European Commission, 2010).. So, it is assumed that technical preconditions are fulfilled. The individuals’ usage of the Internet seeking health-related information is much more common in Germany and Denmark (usage rate: about 50 %) then in Italy (ca. 30 %). The rate of households with internet access is also much lower: about 90 % in Denmark and Germany compared to ca. 70 % in Italy (Eurostat, 2014). So, the technological preconditions need to be improved for a broad implementation of LEHRs. Again, heterogeneous healthcare information systems are an observable obstacle for overarching eHealth solutions in different hospitals and organisations. Missing interoperability as well as inadequate ways to represent the complex and uncertain clinical processes can be observed (Bonacina et al., 2011). But there are regional initiatives (e.g. in Lombardy region) to expand the capabilities of the infrastructure for accessing, storing and managing health data. The vast amount of unstructured and noisy (partly irrelevant) data may also lead to further ICT needs (e.g. for Big data analyses) that were not fully operationalized yet (Barbarito et al., 2015). 3.2.6 Standards In Denmark, commonly used frameworks and communication standards for eHealth solutions were developed or made compulsory (Kierkegaard, 2013). Nonetheless, due to fragmented responsibilities for health care providers within the Danish administration there is a multitude of partly incompatible health record systems leading to frictions in data exchange. There are commonly used and compulsory procedures, but not a single one all over the country. Partly, these circumstances diminish the advantages of the eHealth solutions and lead to suboptimal situations when it comes to inter-institutional or unscheduled treatments. Structural reforms shall lead to improvements (Grosen, 2009; Kierkegaard, 2013). In Germany, overarching electronic exchange and integration of health-related data are not common. Standards are mainly used for data exchange within a single 222 Application of Lifetime Electronic Health Records institution. There is no common usage of standards for inner-sectoral or cross-sectoral exchange. This strongly limits a quick implementation of LEHRs (Klar & Pelikan, 2009). For Italian hospitals, a special strategy is necessary allowing integration of different systems by adopting HL7 standard. The idea was to design a special middle-layer infrastructure that builds an interface for all connected health service providers and the definition of interoperability specifications. Shortcomings in the HL7 standard, hampering the regional implementation of new solutions were also recognised. Suggested changes for this standard shall help to overcome this (Barbarito et al., 2012, 2015). In general, there are no commonly accepted terms or ontologies supporting or even allowing clinical communication on a conceptually integrated level (Bonacina et al., 2011). 4 Summary The results of the case study are summarised in Figure 1. Overall, Denmark shows a lead regarding readiness for LEHR implementation. The broad experience with EHR usage and the problems that came along with it have to be considered for LEHR implementation and support the proposed assessment criteria. In contrary, Germany is only partly ready to implement an LEHR yet. The country’s considerable capabilities have only lead to first important steps. Especially the regulation criterion might currently be excessively fulfilled, what makes it an obstacle rather than an enabler for higher integration. Italy overall is partly ready. Progressive initiatives like in the Lombardy region (Barbarito et al., 2015) may be predestined for further analyses and functioning as a starting point for dissemination activities to other regions. Figure 1: Summarised country assessment of the case study One limiting factor of the present case study is the non-existence of real, broad LEHR implementations by now. However, the great experiences in Denmark with eHealth solutions, the LEHR-like pilot projects in Italy and the highly developed health care systems in the analysed countries provide a good basis for conjecturing positive future developments towards LEHR realisations. Still, smaller populations accompanying with greater willingness to change (like in Denmark or Lombardy) seem to be an advantage to successfully implement new eHealth solutions (Currie & Seddon, 2014; Kierkegaard, 2013; Stroetmann et al., 2011). 5 Discussion and Conclusion In summary, the article contributes to the research field of eHealth by showing prerequisites for the implementation of advanced concepts like LEHRs and by exemplarily demonstrating whether those are given or not. Hurdles causing the disuse are touched on but shall be further analysed. Together with the findings of the present study, implications and measures for a successful implementation of LEHRs can be derived. The case study revealed great potentials for Denmark whereas Germany showed hindering conditions, which need to be adjusted. Italy takes a medium position with first piloting regions. To the authors’ mind, a uniform, Europe-wide LEHR solution is not desirable. Country-specific (considering the respective health care systems) but 223 Kai Gand, Peggy Richter, Werner Esswein interoperable solutions should be aimed at. Furthermore, it is essential to intensify research on possible LEHR implementation solutions as well as on the societal acceptance of such. Already proposed solutions need to be evaluated with respect to country-individual conditions. Furthermore, the indicators of the adapted assessment criteria for LEHR implementation readiness should be further developed, to improve their applicability. For the practice, the present study can function as starting point for the assessment of own solutions and comparison through the criteria with country readiness and therefore helping to successfully design and spread concrete LEHR implementations. References Amelung, V. E., & Janus, K. (2005). Modelle der integrierten Versorgung im Spannungsfeld zwischen Management und Politik. In J. Klauber, B.-P. Robra, & H. Schellschmidt (Eds.), Krankenhaus-Report 2005 (pp. 13–26). Bonn, Magdeburg: Schattauer. Barbarito, F., Pinciroli, F., Barone, A., Pizzo, F., Ranza, R., Mason, J., Mazzola, L., Bonacina, S., Marceglia, S. (2015). Implementing the lifelong personal health record in a regionalised health information system: The case of Lombardy, Italy. Computers in Biology and Medicine, 164–174. http://doi.org/10.1016/j.compbiomed.2013.10.021 Barbarito, F., Pinciroli, F., Mason, J., Marceglia, S., Mazzola, L., & Bonacina, S. (2012). Implementing standards for the interoperability among healthcare providers in the public regionalized Healthcare Information System of the Lombardy Region. Journal of Biomedical Informatics, 45(4), 736–745. http://doi.org/10.1016/j.jbi.2012.01.006 Bonacina, S., Marceglia, S., & Pinciroli, F. (2011). Barriers Against Adoption of Electronic Health Record in Italy. Journal of Healthcare Engineering, 2(4), 509– 526. http://doi.org/10.1260/2040-2295.2.4.509 Cabitza, F., Simone, C., & De Michelis, G. (2015). User-driven prioritization of features for a prospective InterPersonal Health Record: Perceptions from the Italian context. Computers in Biology and Medicine, 59, 202–210. http://doi.org/10.1016/j.compbiomed.2014.03.009 Caligtan, C. A., & Dykes, P. C. (2011). Electronic Health Records and Personal Health Records. Seminars in Oncology Nursing, 27(3), 218–228. http://doi.org/10.1016/j.soncn.2011.04.007 Chaudhry, B., Wang, J., Wu, S., Maglione, M., Mojica, W., Roth, E., Morton, S. C., Shekelle, P. G. (2006). Systematic Review: Impact of Health Information Technology on Quality, Efficiency, and Costs of Medical Care. Annals of Internal Medicine, 144(10), 742–752. http://doi.org/10.7326/0003-4819-144-10- 200605160-00125 Comandé, G., Nocco, L., & Peigné, V. (2015). An empirical study of healthcare providers and patients׳ perceptions of electronic health records. Computers in Biology and Medicine, 59, 194–201. http://doi.org/10.1016/j.compbiomed.2014.01.011 Currie, W. L., & Seddon, J. J. M. (2014). A cross-national analysis of eHealth in the European Union: Some policy and research directions. Information & Management, 51(6), 783–797. http://doi.org/10.1016/j.im.2014.04.004 Degeling, P., Maxwell, S., Kennedy, J., & Coyle, B. (2003). Medicine, management, and modernisation: a “danse macabre”? BMJ, 326(7390), 649–652. http://doi.org/10.1136/bmj.326.7390.649 224 Application of Lifetime Electronic Health Records Deutsch, E., Duftschmid, G., & Dorda, W. (2010). Critical areas of national electronic health record programs—Is our focus correct? International Journal of Medical Informatics, 79(3), 211–222. http://doi.org/10.1016/j.ijmedinf.2009.12.002 Dixon, B. E. (2007). A Roadmap for the Adoption of e-Health. E-Service Journal, 5(3), 3– 13. Draft of a law for secure digital communications and applications in health care (2015). Engemann, C. (2013). Elektronische Gesundheitsakte oder Fallakten. In Qualität in der Medizin dynamisch denken (pp. 149–175). Wiesbaden: Springer Fachmedien. European Commission. (2010). Country Brief: Germany. Retrieved from http://www.ehealth- strategies.eu/database/documents/Germany_CountryBrief_eHS_12.pdf Eurostat. (2014). Information society statisitics. Retrieved from http://ec.europa.eu/eurostat/web/information-society/data/main-tables Fernández-Alemán, J. L., Señor, I. C., Lozoya, P. Á. O., & Toval, A. (2013). Security and privacy in electronic health records: A systematic literature review. Journal of Biomedical Informatics, 46(3), 541–562. http://doi.org/10.1016/j.jbi.2012.12.003 Gand, K., Richter, P., & Esswein, W. (2015). Towards Lifetime Electronic Health Record Implementation. Presented at the eHealth 2015 (accepted), Wien. Grosen, L. (2009). Electronic Health Record in Denmark (No. 14). Retrieved from http://www.hpm.org/survey/dk/a14/5 Harper, S. (2010). The capacity of social security and health care institutions to adapt to an ageing world. International Social Security Review, 63(3/4), 177–196. http://doi.org/10.1111/j.1468-246X.2010.01374.x Häyrinen, K., Saranto, K., & Nykänen, P. (2008). Definition, structure, content, use and impacts of electronic health records: A review of the research literature. International Journal of Medical Informatics, 77(5), 291–304. http://doi.org/10.1016/j.ijmedinf.2007.09.001 Hoerbst, A., & Ammenwerth, E. (2010). Electronic health records. A systematic review on quality requirements. Methods of Information in Medicine, 49(4), 320–336. http://doi.org/10.3414/ME10-01-0038 Hoerbst, A., Kohl, C. D., Knaup, P., & Ammenwerth, E. (2010). Attitudes and behaviors related to the introduction of electronic health records among Austrian and German citizens. International Journal of Medical Informatics, 79(2), 81–89. http://doi.org/10.1016/j.ijmedinf.2009.11.002 Iakovidis, I. (1998). Towards personal health record: current situation, obstacles and trends in implementation of electronic healthcare record in Europe. International Journal of Medical Informatics, 52(1–3), 105–115. http://doi.org/10.1016/S1386- 5056(98)00129-4 International Organization for Standardization. (2005). ISO/TR 20514:2005. Health informatics - Electronic health record - Definition, scope and context. Retrieved from https://www.iso.org/obp/ui/#iso:std:iso:tr:20514:ed-1:v1:en Katehakis, D. G., Sfakianakis, S. G., Kavlentakis, G., Anthoulakis, D. N., & Tsiknakis, M. (2007). Delivering a Lifelong Integrated Electronic Health Record Based on a Service Oriented Architecture. IEEE Transactions on Information Technology in Biomedicine, 11(6), 639–650. http://doi.org/10.1109/TITB.2006.889711 225 Kai Gand, Peggy Richter, Werner Esswein Kierkegaard, P. (2011). Electronic health record: Wiring Europe’s healthcare. Computer Law & Security Review, 27(5), 503–515. http://doi.org/10.1016/j.clsr.2011.07.013 Kierkegaard, P. (2013). eHealth in Denmark: A Case Study. Journal of Medical Systems, 37(6), 1–10. Klar, R., & Pelikan, E. (2009). Stand, Möglichkeiten und Grenzen der Telemedizin in Deutschland. Bundesgesundheitsblatt - Gesundheitsforschung - Gesundheitsschutz, 52(3), 263–269. http://doi.org/10.1007/s00103-009-0787-7 Kopp, I. (2011). Wie eine Leitlinie entsteht. Zeitschrift Für Herz-, Thorax-Und Gefäßchirurgie, 25(4), 217–223. Kukafka, R., Ancker, J. S., Chan, C., Chelico, J., Khan, S., Mortoti, S., … Stephens, K. (2007). Redesigning electronic health record systems to support public health. Journal of Biomedical Informatics, 40(4), 398–409. http://doi.org/10.1016/j.jbi.2007.07.001 Lerum, S. V., & Frich, J. C. (2012). Normative assumptions in integrated care: A conceptual discussion. International Journal of Healthcare Management, 5(1), 32–39. http://doi.org/10.1179/2047971911Y.0000000004 Menzel, D. H.-J. (2006). Informationelle Selbstbestimmung in Projekten der Gesundheits-Telematik. Datenschutz und Datensicherheit - DuD, 30(3), 148–152. http://doi.org/10.1007/s02045-006-0043-4 Palvia, P., En Mao, P., Salam, A. F., & Soliman, K. S. (2003). Management Information Systems Research: What’s There in a Methodology? Communications of the Association for Information Systems, 11, 289–308. Perleth, M., Jakubowski, E., & Busse, R. (2000). “Best Practice” im Gesundheitswesen - oder warum wir evidenzbasierte Medizin, Leitlinien und Health Technology Assessment brauchen. /. Zeitschrift Fuer Aerztliche Fortbildung Und Qualitaetssicherung, 94(9), 741–744. Protti, D., & Johansen, I. (2010). Widespread Adoption of Information Technology in Primary Care Physician Offices in Denmark: A Case Study. Issues in International Health Policy (Commonwealth Fund), (80), 1–14. Schmidt, H., Gerber, A., & Stock, S. (2009). What can we learn from German health incentive schemes? BMJ, 339(sep24 2), b3504–b3504. http://doi.org/10.1136/bmj.b3504 Schreyögg, J., Weinbrenner, S., & Busse, R. (2006). Leistungsmanagement in der Integrierten Versorgung. In P. D. R. Busse, D. J. Schreyögg, & P. D. C. Gericke (Eds.), Management im Gesundheitswesen (pp. 106–122). Springer Berlin Heidelberg. Retrieved from http://link.springer.com/chapter/10.1007/978-3- 540-29465-8_6 Shabo, A. (2006). A Global Socio-economic-medico-legal Model for the Sustainability of Longitudinal Electronic Health Records - Part 1. Methods Inf Med, 45(3), 240– 245. Shabo, A. (2010). Independent health record banks for older people--the ultimate integration of dispersed and disparate medical records. Informatics for Health & Social Care, 35(3-4), 188–199. http://doi.org/10.3109/17538157.2010.528635 Stroetmann, K. A., Artmann, J., Stroetmann, V. N., Protti, D., Dumortier, J., Giest, S., … Whitehouse, D. (2011). European countries on their journey towards national eHealth infrastructures. ICT for Health, Directorate ICT addressing Societal 226 Application of Lifetime Electronic Health Records Challenges, Information Society & Media Directorate-General, European Commission. Tang, P. C., Ash, J. S., Bates, D. W., Overhage, J. M., & Sands, D. Z. (2006). Personal Health Records: Definitions, Benefits, and Strategies for Overcoming Barriers to Adoption. Journal of the American Medical Informatics Association : JAMIA, 13(2), 121–126. http://doi.org/10.1197/jamia.M2025 Tsiknakis, M., Katehakis, D., & Orphanoudakis, S. C. (2004). A health information infrastructure enabling secure access to the life-long multimedia electronic health record. International Congress Series, 1268, 289–294. http://doi.org/10.1016/j.ics.2004.03.275 Van der Linden, H., Kalra, D., Hasman, A., & Talmon, J. (2009). Inter-organizational future proof EHR systems: A review of the security and privacy related issues. International Journal of Medical Informatics, 78(3), 141–160. http://doi.org/10.1016/j.ijmedinf.2008.06.013 Waegemann, P. (2002). Status Report 2002: Electronic Health Records. Retrieved January 8, 2015, from http://www.nasbhc.org/atf/cf/%7BCD9949F2-2761-42FB- BC7A-CEE165C701D9%7D/TA_HIT_what%20is%20an%20emr.pdf Yin, R. K. (2014). Case Study Research – Design and Methods. Sage Publications. 227 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Buying-off privacy concerns for mobility services in the Internet-of-things era A discrete choice experiment on the case of mobile insurance Sebastian Derikx Delft University of Technology, The Netherlands sajpderikx@gmail.com Mark de Reuver Delft University of Technology, The Netherlands g.a.dereuver@tudelft.nl Maarten Kroesen Delft University of Technology, The Netherlands m.kroesen@tudelft.nl Harry Bouwman Delft University of Technology, The Netherlands IAMSR, Abo Akademi University, Finland w.a.g.a.bouwman@tudelft.nl Abstract Internet-of-things technologies will enable collecting vast amounts of mobility data from car owners. Such connected car services can be value-adding but also create new privacy hazards. This paper studies whether and how privacy concerns of car owners can be compensated by offering monetary benefits. We study the case of usage based car insurance services for which the insurance fee is adapted to measured mileage and driving behaviour. A conjoint experiment shows that consumers prefer their current insurance products to usage based car insurance. However, when offered a minor financial compensation, they are willing to give up their privacy to car insurers. Consumers find privacy of behaviour and action more valuable than privacy of location and space. The study is a first to compare different forms of privacy in the acceptance of connected car services. Hereby, we contribute to more fine-grained understanding of privacy concerns in the acceptance of digital services, which will become more important than ever in the upcoming Internet-of-things era. Keywords: Internet-of-things, Privacy, Mobile value services, Conjoint analysis 1 Introduction Internet-of-things is transforming the mobility industries as cars are increasingly becoming connected through dedicated SIM cards or smartphones. Connected cars will generate large amounts of data about mileage and driving behaviour that can be used for 228 Derikx, de Reuver, Kroesen, Bouwman a large variety of value-added services in many areas, like traffic safety, vehicle diagnostics and preventive maintenance and advanced real time navigation. However, there are also many opportunities for customer relationship management, (proximity) marketing and after-sales service. Services can be offered by the car industry (e.g., large and small car dealers, equipment producers), financial industry (e.g., insurers) or other service providers (e.g., leasers, rental providers). However, vast amounts of data collected in connected cars can create privacy and ethical hazards. In general, privacy concerns negatively affect the intention to use digital services (Malhotra et al 2004; Miyazaki & Fernandez 2001). Service providers can compensate privacy concerns by offering convenience or monetary rewards as has been shown for e-commerce services (Hann et al 2007; Li et al 2010; Laudon 1996). However, sensitivity of disclosed personal data will be substantially higher for connected car services than traditional electronic services as highly detailed habits and mobility patterns can be inferred. Since sensitivity of disclosed personal data has a significant positive effect on related privacy concerns (Bansal, Zahedi & Gefen 2010), the question arises whether and how such elevated privacy concerns can still be compensated by service providers. This paper studies if and how privacy concerns for connected car services can be compensated financially. We study this issue through a discrete choice experiment in which the buy-off value of different types of privacy risks is evaluated. We define privacy as “an interest that individuals have in sustaining a ‘personal space’ free from interference by other people and organizations” (Clarke 1999). As a case to study this issue, we focus on usage based insurance services (cf., Handel et al 2013). Insurance services are especially relevant as privacy concerns regarding the insurance industry and its online platforms are already high. Specifically, we consider usage based insurance services for which the insurance fee is based on actual car-use. Differentiating insurance fees based on car use is relevant since damage risks are correlated to the amount of driven kilometres (Vonk, Janse, van Essen, & Dings, 2003) as well as driving behavior (Lajunen, Karola, & Summala, 1997). Usage based insurance services are starting to emerge on the market that utilize not only GPS-data but also motion sensors to measure car acceleration/deceleration and driving behavior. Section 2 provides a theoretical background on privacy. Section 3 provides the method, followed by results in Section 4. Section 5 discusses the findings and concludes the paper. 2 Background 2.1 Defining privacy in the Internet-of-things paradigm Many definitions of privacy exist in literature. Traditionally, privacy has been conceptualized as a right to control over information about oneself. Westin (1967) defines privacy as the ability of individuals to determine for ourselves when, how, and to what extent information about us is communicated to others. Altman (1976) regards privacy as a dialectic and dynamic boundary regulation process which allows a selective control of access to the self or to one’s group. Alternatively, privacy is defined as a condition of not having undocumented personal information known or possessed by others (Parent 1983). 229 Buying-off privacy concerns for mobility services in the ... More recently, utilitarianists have conceptualized privacy as an interest rather than an absolute right. Clarke (1999) considers privacy as a thing that people like to have. Clarke (1999) defined privacy as “an interest that individuals have in sustaining a ‘personal space’ free from interference by other people and organizations”. This study will follow the utilitarian view of privacy as an interest, since this implies that privacy can be redeemed and traded long as the benefits of the service overrun related sacrifices, users will be persuaded to participate. In line with this conceptualization, several studies on e-commerce consider privacy as a tradeoff between the disclosure of personal information and service related benefits (Chorppath & Alpcan, 2013; Dinev & Hart, 2006; Hann et al., 2007; Laudon, 1996; Li et al., 2010). Assuming that privacy is an interest, Clarke (1999) suggests various types of privacy that may be relevant. Clarke defined four categories of privacy, including privacy of the person, privacy of personal data, privacy of personal behavior and privacy of personal communication. Privacy interests can be affected by various activities, i.e. (1) information collection, (2) information processing, (3), information dissemination, and (4) invasion (Solove 2006). Finn et al (2013) argue that these four types of privacy do not cover potential privacy issues of recent technological advances. Technologies such as whole body image scanners, RFID-enabled travel documents, unmanned aerial vehicles, advanced DNA enhancements, second-generation biometrics and connect mobile services raise additional privacy issues. Therefore, Finn et al (2013) expanded Clarke’s categorization to seven types of privacy: privacy of the person, privacy of behaviour and action, privacy of personal communication, privacy of data and image, privacy of thoughts and feelings, privacy of location and space, and privacy of association. Mobile insurance services especially affect privacy of behaviour and action, data and image, and location and space. Privacy of behaviour and action can be affected as data from mobile devices allow identifying travel activities. Especially when combining positioning data from mobile devices, GPS chips and social media, extensive information on one’s behaviour and action can be generated. Privacy of data and image is affected as mobile insurance will typically require personal data to be shared. Privacy of location and space is especially impacted by tracking technologies in mobile phones and cars. Usage based insurance products typically require sharing location information with insurers. Almost all connected devices, even without GPS-sensors, provide detailed information on their location IP addresses, WiFi hotspots and router information. 2.2 Privacy and monetary compensation Privacy is generally seen as a value that stimulates individual freedom and social development (Solove, 2006). Based on a review of existing studies, Paine et al. (2007) show that the general public is increasingly concerned about their online privacy and willing to take countermeasures. At the same time, studies show that most consumers consider disclosing personal information as an integral part of modern life, necessary to obtain products and services (Preibusch, 2013; TNS Opinion and Social, 2008). As such, individuals do consider a utilitarian trade-off between perceived benefits of online services and sacrifices of disclosing personal information. 230 Derikx, de Reuver, Kroesen, Bouwman Disclosure of personal information generally results in elevated privacy concerns (Bansal et al 2010). Various empirical studies show that elevated privacy concerns negatively affect the intention to use online and mobile services (Malhotra et al., 2004; Miyazaki & Fernandez, 2001). Laufer and Wolfe (1977) suggest that individuals perform a “calculus of behavior” to assess the consequences of providing personal information. On the basis of this theoretical construct, individuals consider a trade-off between perceived benefits and sacrifices of disclosing personal information. This implies that unavoidable privacy concerns, associated with the use of mobile insurance, have to be compensated in order to persuade consumers to adopt. Hann, Hui, Lee & Png (2007) state that providers can mitigate the negative effect of privacy concerns on intention to use in two ways: (1) by offering privacy policies regarding the handling and use of personal information and (2) by offering benefits such as monetary rewards or convenience. The latter type of compensating benefits have been further operationalized by Li, Sarathy, & Xu (2010) into expected monetary benefits and perceived usefulness. Laudon (1996) argues that personal information is a commodity that can be priced and exchanged for monetary benefits. Further research by Jen, Ingying, Wei-Ting & Chang showed that the expected monetary benefits have a positive influence on intention to use electronic services. Hereby monetary benefits could be achieved through a discount on existing services or direct pay-outs (Jen et al., 2013). 3 Method We conduct a discrete choice experiment to evaluate the interplay of privacy concerns, monetary compensation and the intention to use usage based insurance services. Conjoint analysis is a statistical approach, often used in market research to determine customer preferences (Green et al 2001; Henscher et al 2005; Louviere et al 2000). Based on implicit trade-offs, perceived utilities by the respondents can be estimated per profile characteristic. By involving financial dimensions in the composition of these profiles, the willingness to pay might also be an output of the conjoint analysis (Henscher et al 2005). We use stated-choice model (Louviere et al., 2000) rather than rating-based conjoint analysis since in reality consumers also make a discrete choice between multiple car insurance packages. 3.1 Sample The population of interest comprises all Dutch private car owners. The survey was carried out at a car ferry service in the Netherlands (Schoonhoven) in October 2014. To maximize the chance of finding private car owners, the survey was carried out on a Friday. After approval of the ferry service, car owners were approached to complete the questionnaire. Hardcopy questionnaire results were imputed into a spreadsheet. Sixty respondents completed the questionnaire, of which five were omitted due to missing data. The resulting sample is representative in terms of gender (48% male compared to 49% in the target population) and car use (55 kilometers per day on average compared to 37 kilometers in the target population). The sample is biased towards highly educated (51% higher education compared to 34% in the target population) and younger people (34% between 18 and 25 compared to 13% in the target population). 231 Buying-off privacy concerns for mobility services in the ... 3.2 Measurement instrument In order to value individuals’ privacy in monetary units, the three relevant forms of privacy identified in Section 2 are operationalized into attributes, see Table 1. Hereby, the attribute levels are composed in such a way that one level involves privacy harm and the other level involves no privacy harm. Privacy type Attribute Level 1 Level 2 (no privacy harm) (privacy harm) (construct) Privacy of location and Kilometer registration Manual Automatic space (web platform) (in-car GPS) Privacy of behavior and Registration road Yes action behavior No (in-car motion sensor) Additional insurance No Yes Privacy of data and offerings image Third party No Yes advertisement Table 1: Conjoint attributes and levels Operationalization of the privacy types is done by building upon examples of mobile insurance products that are emerging on the market currently. As such, operationalization is as close to reality of respondents as possible, which contributes to the external validity of the study. Privacy of location and space is operationalized into the attribute Kilometer registration, which is an important input for usage based car insurance. The insurer can measure the number of kilometers driven automatically through GPS tracking, which harms privacy of location and space. Alternatively, the consumer could register the number of kilometers driven manually through a website, which does not harm privacy of location and space. Privacy of behavior and action is operationalized into the attribute Registration road behavior. Driving behavior could be measured automatically through an in-car motion (G-force) sensor that registers acceleration, deceleration and abrupt steering movements. By doing so, insurers gain in-depth insights in the actual user behavior which harms privacy of behavior and action. Privacy of data and image is operationalized into the reuse of data generated by a usage based insurance service for secondary purposes. The attribute Additional insurance offerings refers to the insurer sending personalized offerings and promotions based on the data collected about the user. The sending of promotions by parties other than the insurer is referred to as Third party advertisement. As both options reuse data provided by the user for secondary purposes, they both negatively affect privacy of data and image. The results of the conjoint analysis will provide the utility that participants derive from every attribute level. By adding a fifth attribute, these utilities can be converted into monetary compensation level, thereby eliciting the buy-off value of privacy. This fifth attribute Relative consumer saving is defined as the discount consumers will receive when adopting the usage based insurance policy. To analyze potential non-linear effects, three attribute levels are included: 0, 10 or 20 euros discount. The level of discount is considered appropriate considering the average monthly fee of all-risk Dutch car insurance policies equals €34. 232 Derikx, de Reuver, Kroesen, Bouwman Based on the defined attributes, choice-sets are composed in which respondents compare two alternative usage based insurance options. In addition, respondents were asked whether they prefer the proposed insurance policy or their current policy. A balanced composition of twelve choice-sets and related attribute levels was generated using Ngene software. Based on the choice-sets and defined attributes, a questionnaire was designed and subsequently pretested with three participants. The consistency of the model results was verified randomly dividing all respondents’ choice-preferences in two equal parts and running the analysis individually for both parts. All estimated coefficients in the sub-groups have the same direction as in the full model, and deviations are generally acceptable. Finally, the uniqueness of each attribute was assessed by computing the correlations between coefficients. All correlations were lower than 0.80, which indicates that the model was able to unique identity the influences of the included attributes. (Hensher et al., 2005). 4 Results Biogeme software is used to analyze the choice behavior of the respondents (Bierlaire 2003). The dataset includes all predefined choice-sets and all respondents’ choices from the questionnaire. The model-file includes a syntax program language to provide instructions to the Biogeme engine. Table 2 provides the part worth utilities of the attributes. All attributes are statistically significant. Relative consumer saving has the highest importance: 65% of a choice for usage based insurance depends on the discount offered. The residual importance is almost equally distributed over the other attributes which implies a balance willingness to pay for all attributes. Attribute Attribute Part worth utility Range Importance Rank Level Kilometer Manual 0 .288 7.34% 5 registration † Automatic -.288 Registration No 0 .378 9.64% 2 road behavior Yes -.378* Additional No 0 .369 9.41% 3 insurance Yes .369* offerings Third party No 0 .351 8.95% 4 advertisement Yes -.351* Relative € 0 -1.42* 2.536 64.66% 1 consumer € 10 .304 saving € 20 1.116 Constant -1.21 233 Buying-off privacy concerns for mobility services in the ... Table 2: Part worth utilities † p < .10; * p < .05 Table 2 also shows a significant disutility of 1.21 compared to the current car insurance policy. In other words, respondents derive a structural disutility from usage based insurance services of 1.21. Next, we transform utility levels to buy-off values using the Relative consumer saving attribute. As 20€ savings corresponds to 2.536 utility points (see Table 4), 1 utility point equals 7.89€. Based on this valuation, the structural disutility of usage based insurance equals €9.54, i.e. a buy-off value of €9.54 per month should be offered for consumers to switch to usage based insurance services. Table 3 presents the buy-off values for each form of privacy harm. In the table, the utility is calculated in a buy-off value using the attribute Monetary compensation. Type of privacy Attribute Involved attribute Utility Buy-off value level per month † Privacy of location Kilometer registration Automatic -0,288 €2,27 and space (in-car GPS) Privacy of behavior Registration road Yes -0,378* €2,98 and action behavior (in-car motion sensor) Privacy of data and Additional insurance Yes 0,369* -€2,91 image (internal) offerings Privacy of data and Third party Yes -0,351* €2,77 image (external) advertisement † p < .10; * p < .05 Table 3: Conjoint utilities and buy-off value privacy Table 3 shows that all buy-off values are in a similar range. Privacy of behavior and actions has a slightly higher buy-off value, equaling €2.98 per month. Regarding the privacy of data and image two buy-off values are determined, relative to the internal and external reuse of personal data. Respondents are willing to sell their personal data for third party advertisements if they receive a financial compensation of €2.77 per month. Strikingly, to receive relevant personalized promotions from the insurance company itself, respondents are willing to pay a monthly fee equaling €2.91. Next, we explore moderating effects of demographics on the utilities, which is especially relevant considering the sampling bias towards younger and higher educated people. Table 4 shows that demographics have considerable effect on the utilities in the conjoint model. For instance, highly educated respondents only require €4.42 to adopt usage based insurance, while lower educated respondents demand €21.33. Moreover, respondents driving more than 30,000 kilometers per year require more compensation than those that drive less. Full Age group Education level Average number of model kilometres per year <41.5 >41.5 Low High <30,000 >30,000 (N=27) (N=26) (N=22) (N=31) (N=38) (N=17) Constant 9.46* 7.12* 11.29* 21.33* 4.42* 7.48* 15.67* 234 Derikx, de Reuver, Kroesen, Bouwman Kilometer registration 2.27 1.51 3.16 2.37 2.35 3.14* -1.27 Registration road behavior 2.98* 3.55* 1.65 .54 3.03* 2.48 1.32 Additional insurance offerings -2.91* -1.49 -4.21* -6.71 -1.66 -2.39 -2.19 Third party advertisement 2.77* 2.34 3.81 3.77 2.30 2.93* -.10 * p < .05 Table 4: Buy-off values for different demographic groups (in euros per month) Regarding the privacy attributes, demographic groups differ only slightly. For instance, younger respondents derive more disutility from registration of road behavior than older people (€3.55 and €1.65 respectively). Higher educated people appear to derive more disutility with the registration of road behavior. However, we should point out here that sample size for the sub-groups is low and thus results can only be used in a speculative manner. 5 Discussion and conclusions Our study shows that specific privacy concerns about usage based insurance services can be compensated by offering a marginal monthly fee. Consumers perceive privacy of behavior and action as more valuable than privacy of location and space. Regarding privacy of data and image, the buy-off value depends on who exploits privacy-sensitive data. While usage of personal data for personalized offerings from the insurer is positively evaluated, third party advertisements have a negative utility. Our findings do indicate that consumers prefer conventional car insurance policies considerably compared to usage based insurance, regardless of privacy concerns. As such, other considerations than privacy will likely play a role in the adoption decision of consumers. For instance, unwillingness to switch in general or normative considerations of fairness in insurance policies may play a role. We also found that people driving more kilometres are less likely to accept usage based insurance, which can be explained because this group would pay a higher fee due to the nature of the product. The main downside of this survey is its representativeness. Highly educated people and people in the age-interval of 18-35 are overrepresented, and the conjoint analysis suggests that younger and highly educated people are less concerned about privacy risks. Another limitation is that interaction effects between the different dimensions of privacy were not included, which could be added in future studies. In terms of operationalization, different dimensions of privacy could have been measured differently. For instance, privacy of data and image could also relate to the degree to which users have control over who uses their data for non-commercial purposes. Moreover, if the operationalization of privacy of data and image would have included calculation of risk profiles and raising of rates based on driving behaviour, higher disutility may have been found. 235 Buying-off privacy concerns for mobility services in the ... This paper takes a utilitarian view on privacy and assumes privacy concerns can be compensated financially. While this view fits the increasingly dominant utilitarian privacy literature, we are aware that there are other privacy schools that have differing conceptualizations and consider privacy as a right that cannot be bargained for (e.g., Westin, 1967). The study contributes to theories on privacy by distinguishing multiple dimensions of privacy rather than the typically one-dimensional operationalization in literature. The study shows that the buy-off value for privacy varies depending on the dimension of privacy concerned. This is especially relevant as Internet-of-things and connected cars concepts will involve ever more complex data to be released which may affect different dimensions of privacy in different ways. References Altman, I. (1976). Privacy: A Conceptual Analysis. Environment and Behavior, 8, 7– 30. Bansal, G., Zahedi, F. “Mariam”, & Gefen, D. (2010). The impact of personal dispositions on information sensitivity, privacy concern and trust in disclosing health information online. Decision Support Systems, 49(2), 138–150. Bansal, G., Zahedi, F. “Mariam”, & Gefen, D. (2010). The impact of personal dispositions on information sensitivity, privacy concern and trust in disclosing health information online. Decision Support Systems, 49(2), 138–150. Chorppath, A. K., & Alpcan, T. (2013). Trading privacy with incentives in mobile commerce: A game theoretic approach. Pervasive and Mobile Computing, 9, 598–612. Clarke, R. (1999). Introduction to dataveillance and information privacy, and definitions of terms, Roger Clarke’s Dataveillance Inf. Priv, 1–17. Dinev, T., & Hart, P. (2006). An Extended Privacy Calculus Model for E-Commerce Transactions. Information Systems Research. Finn, R.L., Wright, D., & Friedewald, M. Seven Types of Privacy, in In European data protection: coming of age, 2013, pp. 3–32. Green, P. E., Krieger, A. M., & Wind, Y. (2001). Thirty Years of Conjoint Analysis: Reflections and Prospects. Interfaces. Händel, P., Ohlsson, J., Ohlsson, M., Skog, I., & Nygren, E. (2013). Smartphone-based measurement systems for road vehicle traffic monitoring and usage-based insurance. IEEE Systems Journal. Hann, I.-H., Hui, K.-L., Lee, S.-Y. T., & Png, I. P. L. P. (2007). Overcoming Online Information Privacy Concerns: An Information-Processing Theory Approach. Journal of Management Information Systems, 24, 13–42. Henscher, Rose, & Greene. (2005). Applied Choice Analysis: A Primer. Journal of the American Statistical Association. 236 Derikx, de Reuver, Kroesen, Bouwman Jen, W., Ingying, L., Wei-Ting, W., & Chang, Y.-T. (2013). Effects of Perceived Benefits and Perceived Costs on Passenger’s Intention to Use Self-ticketing Kiosk of Taiwan High Speed Rail Corporation. Proceedings of the Eastern Asia Society for Transportation Studies, 9. Lajunen, T., Karola, J., & Summala, H. (1997). Speed and acceleration as measures of driving style in young male drivers. Perceptual and Motor Skills, 85, 3–16. Laudon, K. C. (1996). Markets and privacy. Communications of the ACM. Laufer, R. S., & Wolfe, M. (1977). Privacy as a concept and a social issue: A multidimensional developmental theory. Journal of Social Issues, 33, 22–42. Li, H., Sarathy, R., & Xu, H. (2010). Understanding situational online information disclosure as a privacy calculus Understanding situational online information disclosure as a privacy calculus. Journal of Computer Information Systems, 51, 1–29. Louviere, J. J., Hensher, D. A., & Swait, J. D. (2000). Stated choice methods: analysis and applications. Analysis and Applications (Vol. 12, p. 402). M. Bierlaire, “BIOGEME: A free package for the estimation of discrete choice models,” in Proceedings of the 3rd Swiss Transportation Research Conference, 2003. Malhotra, N. K., Kim, S. S., & Agarwal, J. (2004). Internet Users’ Information Privacy Concerns (IUIPC): The Construct, the Scale, and a Causal Model. Information Systems Research. Milne, G. R., & Boza, M.-E. (1999). Trust and concern in consumers’ perceptions of marketing information management practices. Journal of Interactive Marketing, 13(1), 5–24. Miyazaki, A. D., & Fernandez, A. (2001). Consumer Perceptions of Privacy and Security Risks for Online Shopping. The Journal of Consumer Affairs, 35, 27– 44. Paine, C., Reips, U. D., Stieger, S., Joinson, A., & Buchanan, T. (2007). Internet users’ perceptions of “privacy concerns” and “privacy actions.” International Journal of Human Computer Studies, 65, 526–536. Parent, W. (1983). Privacy, Morality and the Law. Philosophy and Public Affairs, 12, 269 – 288. Preibusch, S. (2013). Guide to measuring privacy concern: Review of survey and observational instruments. International Journal of Human-Computer Studies, 71(12), 1133–1143. Solove, D. (2006). A taxonomy of privacy. University of Pennsylvania Law Review, 154, 477–560. TNS Opinion and Social. (2008). Attitudes on Data Protection and Electronic Identity in the European Union. Special Eurobarometer. Vonk, T. (TNO), Janse, M. (TNO), van Essen, H. (CE), & Dings, J. (CE). (2003). Pay As You Drive Mogelijkheden voor een variabele autoverzekeringspremie in Nederland? Delft. 237 Buying-off privacy concerns for mobility services in the ... Westin, A. F. (1967). Privacy and Freedom. New York: Atheneum. Westin, A. F. (2003). Social and political dimensions of privacy. Journal of Social Issues, 59, 431–453. 238 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Digital Wellness for Young Elderly: Research Methodology and Technology Adaptation Christer Carlsson IAMSR/Abo Akademi University, Finland christer.carlsson@abo.fi Pirkko Walden IAMSR/Abo Akademi University, Finland pirkko.walden@abo.fi Abstract The age group 60-74 is labelled the “young elderly” and refers to people in transition from working life to retirement. Studies of mobile services have shown that young elderly customers are regarded as “not trainable” and “not interesting”. Digital wellness services for the “young elderly” with mobile technology represent a new approach to wellness. We compared wellness services on mobile smartphones and did a detailed study of one of them. We found out that standard methodology for developing digital services does not work out too well for the “young elderly” and implemented action design research. Keywords: Wellness, Mobile value services, Young elderly group 1 Introduction Mobile technology has quickly become a global phenomenon on an unprecedented scale; we now have estimates of 6.9 billion mobile phone connections in use this year (ITU Statistics 2014). This gives rise to a couple of proposals on the context and use of mobile technology. The high level of penetration means that mobile technology by necessity has an impact on the daily routines of a majority of the population and that the increasing presence of smartphones can be expected to leave footprints of more advanced use in their daily routines. The standard approach is that mobile service innovations are built and quickly laun- ched after which it is found out if they make/made sense to the potential users by tra- cing the demand. In the early years it was possible to identify distinct user groups and to work out what kind of mobile services would be useful for them (Carlsson and Walden 2012) and then it was possible to build some acceptance and demand models. Since 2010-12 this has become much more demanding as the general population is now the users of smartphones and mobile services and they appear to do whatever they like with the services, wherever they decide to use them and for whatever purpo- se they decide to focus on. When we could identify distinct user groups it made sense 239 Christer Carlsson, Pirkko Walden to work out explanatory models for the use of mobile services (Carlsson, Walden et al 2002, 2004, 2005, 2006, 2012; many of the original results were first presented at the eBled conferences) with the help of the Technology Acceptance Model (Davis 1989) and the Unified Theory of Acceptance and Use of Technology (Venkatesh et al, 2003, 2012). This does not appear to be the case anymore as mobile technology and digital services change how we build and carry out our daily routines; the routines may not offer good explanations for how we select mobile services. In this paper we will study how daily wellness routines of a specific consumer group – the “young elderly” – may be formed by mobile technology and digital services. We have called the age group 60-74 the “young elderly” and distinguish it from the age group 75-84, “senior citizens”, as we found out that the issues and needs for the two age groups are different. The “young elderly” is a sizeable customer group in the Nor- dic countries (about 4-5 million people) for building the same or similar mobile services as the countries have similar cultures, similar social care legislation and similar political value systems. The “young elderly” will be about 97 million in the EU by 2020 and the age group represents 18-23 % of the population in most EU countries (cf. UN statistics 2014); the age group is expected to be 1 billion globally by 2020. This “not attractive, not interesting” market could potentially be the basis for a global digital service industry (Turban et al 2009). This context is the one we are addressing. The age group 60-74 will have some age-related functional impairment and we propo- se that digital services on mobile platforms could be designed and implemented to counter the impairments. This will keep the “young elderly” active, autonomous and self-reliant, which will improve individual quality of life. It will also have a significant impact on society as improved health for the “young elderly” will reduce the cost for tax-funded health and social care; the impact is significant as we deal with a very large group of people. On a global scale the impact will be somewhat dramatic. We will sketch a few more proposals. Digital services for the “young elderly” will be delivered over mobile platforms but will be produced and supported through cloud architectures. This allows the core of the services to be designed and built according to a common standard – as the EU initiated and sponsored the GSM standard for mobile networks – but will require nationally, regionally and culturally tailored services. This will initiate and drive new business models for an agile development of digital services and ecosystems. The ecosystems will be growth environment for hundreds of SMEs that develop digital services in each country; the ecosystems will form a global industry over the next decade. The final proposal is a change of attitude – the technology for “young elderly” needs to be advanced, this age group will drive the next generation of technology development; why ? –1 billion users will have an impact on technology developers. In this paper we can work out only parts of this context and test only a few of the pro- posals; in section 2 we will introduce digital wellness services and explain experiments we carried out with groups testing the services; in section 3 we have worked out the adaptation of IS research methodology for the “young elderly” context; section 4 summarizes the results. 240 Digital Wellness for Young Elderly 2 Wellness Services 2.1 Wellness Dimensions It is taken as common wisdom that people get more concerned about their health and well-being with increasing age as impairments of various kinds start to appear. Counte- ring the effects of impairments will be the challenge for a quickly growing wellness technology. Wellness services could be digital services developed for and used over mobile smartphones. Adams (2003) identified four main principles of wellness: (i) wellness is multidimensio- nal; (ii) wellness research and practice should be oriented towards identifying causes of wellness rather than illness; (iii) wellness is about balance, and (iv) wellness is rela- tive, subjective or perceptual. Saracci (1997) - health and happiness are encompassed in the term “wellness”. Myers, Sweeney and Witmer (2005) – “wellness is a way of life … optimal health and well-being … body, mind and spirit are integrated by the indivi- dual to live more fully within the human and natural community”. The WHO defines wellness as “the complete mental, physical as well as social well-being of a person or groups of persons in achieving the best satisfying or fulfilling life and not merely the absence of disease or any form of infirmity (WHO 2014). There has been quite some debate over the years about the dimensions of wellness, one of the most complete lists includes: emotional, financial, occupational, environmental, intellectual, physical, social and spiritual wellness (UC Davis 2015). After a series of interactive, semi-structured workshops with groups of “young elderly” in the fall 2014 and spring 2015 we built wellness with four dimensions: (i) intellectual wellness, (ii) physical wellness, (iii) social wellness, and (iv) emotional wellness; one reason is that wellness now mirrors the functional impairment dimensions. Figure 1 shows that physical wellness will influence emotional, intellectual and social wellness; an improvement of physical wellness may have a positive (or sometimes negative) impact on the other wellness dimensions; a decline of physical wellness may have eit- her positive or negative impact on the other wellness dimensions. Figure 1: Wellness dimensions We found out during the workshops that the effects are individual and personal – the “young elderly” are set in their routines – and that we should avoid being too detailed in the way we work out the wellness effects. Systematic studies of larger groups of “young elderly” should produce wellness profiles, which then would support wellness services (fig.2). 241 Christer Carlsson, Pirkko Walden Figure 2: Wellness profiles A further study of the wellness dimensions indicated that we can identify “grades of wellness” with some min and max levels that would be relevant for a majority of the “young elderly” (but there will always be individuals that do not fit any pattern); this is indicated with the dotted lines in figure 2. Then we worked out the red profile for peo- ple active on social and emotional wellness, but not so active on intellectual and physical wellness (our easy-going friends). As a contrast we worked out the yellow profile for people active on physical wellness but not that active on emotional, intellec- tual and social dimensions (our marathon and triathlon active friends). The profiles will find support in analytics (with data from different countries, cultures, socio-economic groups, etc.) and form a strong basis for wellness services. In 2014 90% of the Finnish population aged 55-64 had used the Internet in the past 3 months; 56% used the Internet several times per day; 83% has used Internet banking in the past 3 months and 31% had followed some social network service; the share of Finns using smart phones in 2014 was 60% - the 55-64 age group showed 28% in 2012 but was around 50% in 2014 (the estimated increase per year was 10% points) (cf. Statistics Finland 2014); the use of digital services is becoming common among the “young elderly”. In Carlsson and Walden (2012) is described the results of a series of annual studies 2003-2011 of the adoption and use of mobile services among Finnish consumers; the studies were based on random samples of 1000-1300 consumers; the samples were representative for Finnish consumers. After the 2008 generation shift in mobile phones there was no significant change in the use of mobile services; most users continued with the same, basic services (voice, SMS, ringtones and icons, search functions) with more functionality on the new generation phones (Sell et al 2012). A new generation shift in 2011 came with the smart mobile phones. Smart phones drove the develop- ment of mobile applications, users with a smart phone downloaded applications and started using more and/or more advanced mobile services (Sell et al 2012). Application downloads from Apple’s App Store are counted in billions; in Europe 7.2% of the sub- scribers will use downloaded applications by 2015; worldwide this is estimated at 5.9%, but in North America at 26.9% (Portio 2011). The 2011 survey of Finnish con- sumers found that besides a small ‘power user’ group (15 %) and ‘interested but inac- tive users’ (47 %), 38 % of smart phone users do not have advanced services (Sell et al 2012). Similar results were found by the network operator DNA (AddValue 2012) where 29% of the respondents were ‘passive smart phone users’ with only voice and 242 Digital Wellness for Young Elderly short message services. The market studies are reality checks to remind us that the general population is not that keen on getting new and advanced mobile services. We should bear this in mind when we get enthusiastic about digital wellness services – it is not a given that the intended users will actually be excited about using them. 2.2 Wellness Service Applications In the spring of 2014 we explored the market for wellness services over smartphones with a group of 26 graduate students. We restricted the state-of-the-art to the three most important operating systems – Android, iOS and Windows Phone – and collected the applications that were most often mentioned on the Internet in surveys of wellness services (our study is not claimed to be an exhaustive search; the market is also most dynamic and fragmented, the borderline between health care and wellness services is not well-defined). There are already quite a few applications (fig.3); they are innovati- ve, overlapping and competing; some of them are already showing millions of down- loads, others are just finding their first customers. Android, iOS and Windows Phone o Sports Tracker: Shows running distance, maps; tracks movement by walking, cycling or run- ning. o MyFitnessPal: Set target weight and get meal suggestions; shows daily routine for calories and exercise. o Endomondo: GPS integration for running, cycling; Pep –Talk; integration with social networks. o Wellmo: Tracking weight, steps, sleep, exercise, alcohol, mood; personal targets, resolutions; integrates with external devices, databases; o Runtastic: Tracking fitness routines; keeps track of routines done with applications. Android and iOS o Lift: A virtual coach for reaching set goals with different pre-installed objectives. o Weight Watchers Mobile: Tracks food consumption, activity level, weight; healthy habits, exercise advice. o Noom Weight Loss Coach: Tracks calories, activity; coaching the user to achieve weight loss goals. o Fig: Holistic goals for 300 activities; wellness guide; sharing common goals with friends. o Fitbit: Tracking daily goals, progress; steps taken, distance travelled, calories burned, sleep; food plan. o Nike + Running: Only running; speed time, distance with GPS, smartphone accelerometer; routes, maps. o Moves: Activity monitoring; step, calorie counters; walking, running, bicycling; routes on map. o TactioHealth: Monitoring everyday life; steps taken, weight, exercise, body fat, heart rate, blood pressure, cholesterol and BMI; integrates with external devices. o Fitocracy: Role playing game on workouts; achievements, points, rewards, levels and competiti- ons. o ‘8700’: The ideal average daily consumption in kilojoules (kJ); healthy personalised eating plan. Android o S Health: Pedometer, exercise tracking, weight tracking, heart rate monitoring and food intake tracking. o Lose It: Weight loss goal set; trackers for weight, sleep, steps and exercise; meal and exercise planning. o My Tracks: Tracks movement for walk, run or bike; measures speed, distance and elevation. o DBSA Wellness Tracker: Tracking emotional, mental, physical health; at-a-glance summary of health trend. iOS 243 Christer Carlsson, Pirkko Walden o Datalove: Wellness values as input by the user; analysis and graphics; metrics through GPS. o Cardiio: Follows heart rate on smartphone camera; gives a fitness level rating, potential life expectancy. o WellnessFX: Health goals with data analysis, visualization and trends; integrates with laboratory tests. o Teemo: Fitness as a game (climbing, running, sports), communication in social media. Figure 3: Wellness applications for Android, iOS and Windows Phone The students searched for mention of “five winning wellness products for smart pho- nes” on a large number of websites and registered the wellness products that were nominated as “winners” by various experts; the MyFitnessPal got the most frequent nominations followed by Fooducate, Fitocracy and Runkeeper (of which only Fitocracy made the students’ list of promising applications). Many applications appear to have a narrow focus on a few wellness activities but there are exceptions (Endomondo, Fig, Fitbit, S Health, Sports Tracker, Wellmo); it appears that the ideal solution for the “young elderly” should integrate the functionality of many of the applications. Next the students were given access to the Wellmo application (actually a platform for scores of applications and close to the “ideal” solution) and asked to install it on their smartphones (free access by Wellmo Inc. which is a Nokia spin-off). Their tasks: (i) learn how to use Wellmo; (ii) build a profile of personal targets and resolutions; (iii) define weekly targets and find out how well you meet them over 4 weeks; (iv) find out how well personal targets fit your wellness objectives; (v) find out if there are features missing that should be included. The students completed diaries (in the form detailed by Brandt et al 2007) over their activities and wrote reports on their findings to give us a first summary of the digital wellness services; in total we collected 26 diaries. The students were rather critical of the Wellmo – in typical student fashion – but the reasons varied quite a bit. Some of the students were active users of wellness apps (typically Sports Tracker, Nike+ Run- ning, etc.) and found Wellmo not as advanced as these; some users found weight watching ridiculous as they were body builders and wanted to build body mass; several users missed GPS-based tracking and maps to support training activities, other users wanted integration with social media to challenge friends. There were specifics: techni- cal excellence is expected for interfaces (typical for many competing apps); wireless connection and integration with smart watches, sensor armlets, sleep monitors, diet support, digital scales for weight watching, etc. mostly because there are apps on the market featuring one or several of these functions. Overall the judgment was that Wellmo is a “pretty good” but yet “not just right” app that has the potential to become a useful wellness tool. Discussions over the integration possibilities brought the idea that an omnivore solution would be beneficial [omnivore, Latin omnes, omnia for “all, everything” and vorare for “devour”, cf. Wikipedia] – the user should be able to build a wellness package by allowing series of apps to interconnect and offer her/him the synergistic functionality of many specialised apps. This should be a worthwhile challenge for the smartphone developers and for Wellmo. Wellmo is, in fact, rather a good prototype of an omnivore platform and as the Wellmo developers have continued developing it with more omni- 244 Digital Wellness for Young Elderly vore features (it is now having interfaces to 100+ digital devices and services, cf. fig. 4). Figure 4: Wellmo as an omnivore platform A final observation was that Wellmo and most of the competing apps are planned and designed with some wellness features in mind; the software was built (using advanced, well-tried modules and agile methods) and composed as prototypes; the prototypes went through extensive testing and validation; then the process of persuading potenti- al users to become adopters started. It appears that this approach is not very good and many of the “not just right” features are the results of the development method. Thus the group of graduate students was given an additional task: to demonstrate Wellmo to “young elderly” in their family and among their friends and to find answers to several questions: (i) what wellness features and criteria are important for the young elderly? (ii) if Wellmo is to be adapted to the young elderly market, what featu- res are still relevant, what features should be changed, what features need to be added? The students carried out interviews with 52 people from 9 countries. It quickly became clear that the substance of “wellness” becomes different with advan- cing age; the focus of the requirements for improvements to Wellmo differed from what the students were looking for. There was not so much active use of athletic apps as wish lists including (i) BMI calculators and weight monitoring; (ii) diet advice and calorie counters; (iii) trackers of sleep patterns; (iv) activity trackers, “personal trai- ners”, pedometers; (v) reminder, appointment booker & tracker; (vi) medication moni- tor, reminder; (vii) blood pressure monitoring; (viii) health care adapted monitors, symptom checkers; and, (ix) smoking, drinking monitors. The people interviewed rep- resent, no doubt, a convenience sample that is not very representative. Nevertheless, it is possible to capture the notion that the “reminder-tracking-monitoring” functionality will probably be valuable as part of their everyday routines. 3 IS Methodology Adaptations In terms of IS research methodology we used a case-study approach to work out the digital wellness services that could fit the “young elderly”. This helped us to realize that we are on the wrong track. The “young elderly” group does not understand “wellness” in the same way as the students do – and not in the same way as the Wellmo desig- ners do (the same seems to apply to most of the wellness apps). The students happily spent time to figure out how to install, activate and use the Wellmo app and several of 245 Christer Carlsson, Pirkko Walden the competing apps that they found on Internet. The “young elderly” group did not have any patience with spending time to learn how to use an app; the Wellmo that was demonstrated to them and which they tried to use was found non-intuitive and too complicated; their reaction was that the app should have a clear and demonstrated value before they start investing time with it. We find ourselves in an interesting situation i.e. to find and/or work out a research methodology that will support the development of digital wellness services for “young elderly”, when these services do not yet exist and when we do not know if the inten- ded user group actually would like to have them. The methodology should also guide the development of business models and ecosystems. Design science is bringing out practical relevance in theory constructs and – if necessary - compromising on science precision to get constructs that are useful for planning, problem solving and decision making. The designs build on an understanding of what is needed to deal with real world problems. The design is both a process (a set of activities) and a product (an artefact) and both can be validated and verified to be logically consistent and technically free of errors. Design science has strong appeal as a conceptual framework for digital wellness services; they will be software constructs (artefacts) that we can (i) design jointly with potential or coming users; (ii) the arte- facts can be validated for design and construct errors, the usability of the artefacts can be tested and (iii) the functionality of the services can be worked out in the context and with the users for which they were designed. Most of the results of (i)-(iii) can be generalised in the positivistic sense; the insight can be reused for other contexts and for the development and implementation of other artefacts. Action design research (ADR) (Sein et al 2011) found that design science is too technologically oriented and is not paying enough attention to the organisational context of the artefacts. In our case the forming of ecosystems for wellness services is the organisational context that we should allow to influence the artefact design. This is, of course, to deal with the dynamics and the complexity of the ecosystem that will be a problem for engineering-inspired methods. ADR intervenes in problem situations and works out IS artefacts that help solve the problems, even if the setting is dynamic and complex. Gronroos (2008) proposed that service design should be a co-creative process between the service producer and the customer; if a firm adopts a service logic it will be possib- le to get involved with their customers' value-generating processes. As we found out that “young elderly” will not be keen on getting educated on digital wellness services we have proposed another approach: to work out the services in co-creation processes with groups of young elderly. We realize that we will have to find similarities in knowledge and skill backgrounds, in functional abilities, in social skills and needs, in wellness status, in the structure of daily routines, in wellness objectives, etc. and use that in order to develop design principles. We also realize that we will be having inte- ractive development processes with these groups of people that are somehow similar, and that we will be doing this first with tens of groups in a country, then with hundreds of groups in several countries, moving to thousands of groups in tens of countries. This will require effective IS support in order to be manageable. 246 Digital Wellness for Young Elderly Proposal for a methodology. The foundation should be in ADR which builds on four stages and seven principles (Sein et al 2011): 1. Problem formulation: (i) identify situations for which an ME is needed [practice-inspired research]; (ii) work out an overview of state-of-the-art technology for ME [theory-ingrained artefact] 2. Building, intervention and evaluation: (iii) design and animate ME functionality [reciprocal sha- ping], (iv) interactive, iterative design of functions with input from team [mutually influential roles], (v) interactive and iterative validation and verification of emerging ME artefact(s) [authentic and concurrent evaluation] 3. Reflection and learning: (vi) team evaluation and context adaptation of ME artefact (guided emer- gence) 4. Formalization of learning: (vii) specifications of ME application for software builders (generalized outcomes) ADR offers the flexibility and innovation processes of design science combined with the possibility to verify and validate the technical and logical correctness of artefacts thro- ugh strict testing methods. The ADR methodology was tested by a (new) group of 30 graduate students during the spring 2015. Their assignment was to take Wellmo as their initial platform and then to work out wellness services (based on the results of the previous student group) that would fit the daily routines of the “young elderly” and support them in developing rou- tines for wellness to be included among the daily routines. The students got 18 volun- teers from one of the associations for elderly in the city of Turku to work with them as experts and sparring partners on the daily routines of “young elderly”, on their percep- tions of wellness and on the type of wellness services that would benefit them. This turned out to be a rewarding process: the students did not have any deeper insight of the context and the world of the “young elderly” – their ideas of what digital wellness services may be useful and value-producing were rather far off the mark. The volunte- ering experts had a great time with the students, they learned about the new technology and about possibilities that this could offer them, they had ideas about what wellness services should be and how they should be used – ideas the students could identify from their work with the Wellmo – ideas they could work out to service and software designs in the ADR process. The ADR process required a number of iterations that were collected in design reports (we collected 15 design reports, the students worked pairwise) that were presented to the Wellmo service and software designers in a seminar. A new version of Wellmo with the wellness service designs worked out by the students is due in spring 2015. Our plans are now to run this set of wellness services in new ADR processes with several groups of 20-25 “young elderly” volunteers; these groups are set up to run during summer and fall of 2015. In our work with the associations for elderly we realized that the work with digital wellness for “young elderly” may serve as pro-active prevention, i.e. by getting the age group 60-74 years to adopt wellness services, and to build and sustain wellness routi- nes, we can contribute to keeping the 75+ age group healthy, active and independent. Besides other positive effects this will have a significant impact on tax-funded health and social care costs for the society. Advancing age carries with it risk for suffering functional impairment. This risk is understood through common sense, spreads through common wisdom and is verified through statistics and studies carried out by the UN (2014). It is possible to influence the risks with a systematic use of wellness routines, including changes and adaptations 247 Christer Carlsson, Pirkko Walden of these routines. The wellness routines are now understood in the four-dimensional sense we have introduced. Functional impairment can be loosely classified as cognitive, physical, social and emoti- onal impairment (cf. Saracci 1997) – there are a number of classifications, but we sho- uld keep the classification simple and useful for our context. In our definition of wellness – emotional, intellectual, physical and social – we matched the impairment classification in order to get a basis for the development of wellness services and wellness routines. Functional impairment is non-static, it is a dynamic process that changes in multiple dimensions; thus the wellness services should match the dynamics and the multi-dimensionality. A final lesson learned from the semi-structured, interactive workshops with the “young elderly” volunteers was that the “digital wellness” processes will produce results over extended periods of time. First, the adoption of wellness services on mobile platforms is its own process; second, the forming of wellness routines, and then maintaining and sustaining them, is also a process that will work out over time; third, the development of the digital platforms and services runs in parallel with the first two processes – and is actually much faster than these; fourth, there needs to be an ecosystem (or several eco-systems in several countries) of digital service developers and providers, platform developers and providers, back-end support (most probably cloud service-based) deve- lopers and providers, personal trainers and consultants, public service integrators, etc. – all the actors needed to make a modern society provide support to hundreds of thou- sands, probably millions of citizens. 4 Summary and Conclusion We identified the “young elderly” age group as a growing and now very large group of people for which systematic development of mobile services has not been foreseen nor planned. We noted that the key argument is proactive prevention, i.e. if the “young elderly” are healthy, active and independent then possibly/probably also the next, “senior”, age group will show better health, activity and independence. There is a growing concern that EU member countries will not be able to provide the health and social care they should give their ageing population because of the unfavourable age structure of their societies – the number of citizens in the productive population will not be enough to carry the cost for health and social care of the ageing population. We have worked out the idea of digital wellness services for the “young elderly”. First we carried out a state-of-the-art, and then we had a group of students to thoroughly test one promising wellness application (actually a platform) and compare its strengths and weaknesses with five competing applications. The tested platform – Wellmo – was then evaluated with a group of “young elderly” and we found out that their needs for wellness services differed from the perceptions and design assumptions of the Wellmo designers. This was then the starting point for another group of students to carry out an ADR-based design process of wellness services for the “young elderly”. Their results and design protocols were adopted by the Wellmo service and software designers and are now being implemented. In parallel with technical work with digital wellness services and the development of a platform for them – which we defined as an omnivore platform – we also carried out 248 Digital Wellness for Young Elderly work on the key elements of wellness and the relationships between adoption of wellness routines and the risk for functional impairment. It was formulated as common sense by our “young elderly” volunteers when they were working with the students – if wellness routines keep you from suffering functional impairment it is a small inves- tment, because “it is much more fun to become older if you are in good shape”. References Adams, T. B. (2003), The Power of Perceptions: Measuring Wellness in a Globally Acceptable, Philosophically Consistent Way, Wellness Management AddValue (2012), Report for DNA on Smart Phones, Helsinki 2012, 32 p Brandt, J. Noah, W. and Scott R. Klemmer. (2007), Lowering the burden for diary studies under mobile conditions. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 2303-2308, ACM Carlsson, C. and Walden, P. (2012), From MCOM Visions to Mobile Value Services; Roger Clarke, An-dreja Puchar and Joze Gricar (eds.), The First 25 Years of the Bled eConference, University of Maribor, Bled, pp 69-91 Carlsson, C., Carlsson, J., Puhakainen, J. and Walden, P. (2006), Nice Mobile Services do not Fly. Observa- tions of Mobile Services and the Finnish Consumers, in Pirkko Walden, M. Lynne Markus, Joze Gri- car, Andreja Pucihar and Gregor Lenart (eds.), Proceedings of the 19th Bled eCommerce Conferen- ce, eValues, Bled, Slovenia Carlsson, C., Carlsson, J., Denk, M. and Walden, P. (2005), Mobile Commerce: Insights from Expert Surveys in Austria and Finland, in Douglas R. Vogel, Pirkko Walden, Joze Gricar and Gregor Lenart (eds.), Proceedings of the 18th Bled eCommerce Conference, eIntegration in Action, Bled, Slovenia Carlsson, C., Hyvonen, K., Repo, P. and Walden, P. (2004), It’s all about my phone! Use of Mobile Services in Two Finnish Consumer Samples, in Yao-Hua Tan, Douglas R. Vogel, Joze Gricar and Gregor Lenart (eds.), Proceedings of the 17th Bled eCommerce Conference, eGlobal, Bled, Slovenia Carlsson, C. and Walden, P. (2002), Mobile Commerce: A Summary of Quests for Value-Added Products & Services, 15th Bled Conference Proceedings, Bled, pp 463-476 Carlsson, C. and Walden, P. (2002), Further Quests for Value-Added Products & Services in Mobile Com- merce, ECIS 2002 Proceedings, Gdansk, pp 715-724 Davis, F. D. (1989), Perceived Usefulness, Perceived Ease of Use and User Acceptance of Information Technology, MIS Quarterly, 13(3), pp 319-340 Gronroos, C., (2008), Service logic revisited: who creates value? And who co-creates? European Business Review, Vol. 20 Iss: 4, pp.298 - 314 ITU Statistics (http://www.itu.int/ict/statistics) Myers, J. E., Sweeney, T. J. and M. Witmer (2005), A Holistic Model of Wellness. http://www.mind- garden.com-/products/wells.htm Portio (2011), Disruptive Mobile Applications and Services 2011-2015, Portio Research Limited Saracci, R. (1997), The World Health Organization Needs to Reconsider its Definition of Health. http://bmj. bmjjournals.com Sein, M.K., Henfridsson, O., Sandeep, P., Rossi, M. and R. Lindgren (2011), Action Design Research, MIS Quarterly,, Vol.35, No.1, 37-56 Sell, A., Walden, P. and Carlsson, C. (2012), I am a Smart Phone User – Key Insights from the Finnish Market; Harry Bouwman, Virpi Tuunainen (eds.), Proceedings of the ICMB 2012, TU Delft, pp 265- 276 Statistics Finland 2014, http://www.stat.fi Student Health and Counselling Services, UC Davis (2015), https:// shcs.ucdavis.edu/wellness/, retrieved March 28, 2015 Turban, E., King, D. and Lang, J. (2009), Introduction to Electronic Commerce, Pearson, Harlow, UK United Nations Department of Economic and Social Affairs (2014), Population ageing and sustainable development, No. 2014/4; available at: http://www.un.org./en/development/desa/population/publications/pdf/pop-facts/Popfacts_2014-4. Pdf; accessed on March 21, 2015. Venkatesh, V., Morris, M.G., Davis, G.B. and Davis, F.D.(2003), User Acceptance of Information Technology: Toward a Unified View, MIS Quarterly,Vol. 27, No. 3, pp. 425-478 249 Christer Carlsson, Pirkko Walden Venkatesh, V., Thong, J. Y. L., and Xu, X. (2012), Consumer Acceptance and Use of Information Technology: Extending the Unified Theory of Acceptance and Use of Technology, MIS Quarterly, Vol. 36, No.1, pp 157-178 World Health Organization (2014), Preamble to the Constitution, www.who.int, retrieved March 28, 2015 250 BACK 28th Bled eConference #eWellbeing June 7 - 10, 2015; Bled, Slovenia Aggregating Community Resources of Care and Assistance Services for the Elderly Population Maria Manuela Cruz-Cunha Polytechnic Institute of Cávado and Ave, Portugal mcunha@ipca.pt Isabel Maria Ferreira Miranda Municipality of Guimarães, Portugal isabel.miranda@cm-guimaraes.pt Ricardo Simoes Polytechnic Institute of Cávado and Ave, Portugal rsimoes@ipca.pt João Varajão University of Minho, Portugal varajao@dsi.uminho.pt Abstract Changes in population age structure are a major concern and represent a priority in the agendas and policies of the developed world, which are demanding for renewed models of social and healthcare as well as assistance services to the elderly population. Studies indicate that as far as possible these types of services should desirably be provided at the user’s home, and that ICT-based solutions can have tremendous impact on the delivery of new services. This paper highlight and discusses some of the main results of a project undertaken in a Portuguese Municipality that demonstrates the potential contribution of an e-Marketplace of care and assistance services to the well- being of elderly people. Studies undertaken allowed identifying the main services that should be provided by such e-Marketplace (termed GuiMarket), the relevance that the population grant to this platform and, conversely, the fact that the Digital Divide phenomena influences the potential utilization of this project (and alike projects). The findings support that there is a strong relation between age and qualifications, and between access to ICT and the intended use of GuiMarket. Keywords: electronic marketplace, health care, social care, care services, assistance services, elderly. 251 M. M. Cruz-Cunha, I. M. Miranda,R. Simoes, J. Varajão 1 Introduction According to the “Ageing Report” (European_Commission, 2012a), by 2060 one in three Europeans will be over 65; the population aged 65 will almost double, and the number of people in the 80-and-over age group is projected to almost triple from 2010 to 2060. Changes in population age structure, consequence of declining fertility rates in recent decades, together with recent increases in life expectancy, may exert a significant influence on economic growth (Bloom, Canning, & Fink, 2008). This demographic trend has led to increased health care spending and a higher demand for care and assistance services, threatening existing public health and welfare systems (OECD 2008). Several studies indicate that as far as possible health care and social care services should desirably be provided at the user’s home (Kaye, LaPlante, & Harrington, 2009; Makai, Brouwer, Koopmanschap, Stolk, & Nieboer, 2014; Tang & Venables, 2000; United_Nations, 2009). It is argued that the existence of a network of health care, social care and professional services providers, working articulately with an underlying effective management and intermediation service can be a powerful tool and result in improving the quality of life for people with special needs (elderly and permanently or temporarily disabled people) and to the population in general (Tanner, 2005). Literature and official documents evidence that Internet, telecommunications technologies and infrastructures may contribute significantly to health care system performance (Babulak, 2006; European-Commission, 2007; Kerzman, Janssen, & Ruster, 2003; Séror, 2002; Smits & Janssen, 2008). Literature indicates that Assistive Technologies and Information and Communication Technologies (ICT) may improve quality of life, extend length of community residence, improve physical and mental health status, delay the onset of serious health problems and reduce family and care- giver burden (for example Blaschke, Freddolino, & Mullen, 2009; Doukas et al., 2011; Magnusson, Hanson, & Borg, 2004; Muncert et al., 2012). The Digital Agenda for Europe 2010 suggests ICT-based actions to support ‘independent living’ (European_Commission, 2010a). In this framework, the authors were invited in 2010 to study, in a Portuguese Municipality, the development of an ICT-based solution for improving well-being of elderly people and people with special needs staying at home,. The authors suggested an electronic marketplace (e-Marketplace) for care and assistance services, to support the above-mentioned network of health care, social care and professional services providers ( GuiMarket). This paper presents a compilation and discussion of several results achieved in several studies undertaken within the project during the last four years, and presents briefly the developed prototype. The paper is structured as follows. Section two introduces some challenges for ageing and new ICT-based proposals, including information services, electronic marketplaces and social networks. Section three introduces the authors' proposal and the prototype, section four presents results of several wide studies undertaken and section five discusses the impact of digital divide, also based on the results of the study. Limitations of the study and some concluding remarks are presented in sections six and seven. 2 Ageing: new challenges and new responses In the 21st Century economy sectors like health and social services have a tendency to grow, in GDP percentage as well as in creating employment (European_Commission, 2013). 252 Aggregating Community Resources of Care and Assistance Services for Elderly 2.1 New challenges The demografic effect of changes in population age structure will lead to greater demand for elderly care. Although this brings some challenging demands to the different systems of care, it is well known that enabling elderly to stay at home as long as possible can help to improve their quality of life and is an important mechanism in meeting rising demands (OECD, 2008, 2013; Palm, 2014). These growing demands on welfare services due to an ageing population is leading policy makers to suggest the use of ICT as a support to a cost-effective delivery of social and health care (European_Commission, 2010a). Industry for ageing well must invest and innovate at a European level and scale – in close cooperation with users and consumers (European_Commission, 2012a). And all of us must feel empowered to integrate ICT-products and services for ageing well in our private lives and professional practice. Over 50% of Europeans use the internet daily – but 30% have never used it at all (European_Commission, 2010b)! This disadvantaged social group, largely made up of people aged 65 or more, can hamper the digital society, and may contribute to the health and well-being divide across the EU region (European_Commission, 2013). Moreover, disabled persons face particular difficulties in benefiting fully from new electronic content and services. As ever more daily tasks are carried out online, everyone needs enhanced digital skills to participate fully in society (European_Commission, 2012b). 2.2 And new responses Xie et al. (2012) presented a survey made across several local health agencies in the UK, in order to determine the level of personalization in social care services for elders. Among the factors studied, the survey identified that people are making use of a range of community services beyond typical health care. Among the services that are receiving more interest from older people are shopping, housework, leisure activities support, and gardening. These services are growing in demand from local management centers but also from older people that are gaining independence from institutional social services and want to contract these services themselves (Xie et al., 2012). The rising demand for care services is stimulating a new offer of services and inniciatives of Ambient Assisted Living (Rashidi & Mihailidis, 2013; Venkatesh, Vaithyanathan, Kumar, & Raj, 2012). E-Marketplaces implement the concept of 'market' and were developed to bring together large numbers of buyers and sellers expanding the choices available to buyers, and giving sellers new opportunities and access to new customers (buyers), simultaneously reducing transaction costs for all participants (Cunha & Putnik, 2006; Kaplan & Sawhney, 2000). We are currently witnessing an attempt to use in the health and social care sectors some solutions already in use by the business sector, to optimize processes of product sourcing and supply chain improvement, such as the several well-succeeded “last generation” e-Marketplaces (e.g. www.broadlane.com, www.Med2med.com, www.labx.com, www.saniline.com), and many others referred by directories like eMarketServices, available online at http://www.emarketservices.com (Cunha, 2003; Cunha, Putnik, Gunasekaran, & Ávila, 2005; eMarketServices, 2007; Putnik, Gonçalves, Sluga, & Cunha, 2008; Zallah, 2005). A few examples exist of Internet-based services and markets between users of care and providers of care, such as “CareAuction.nl”, a new intermediary on the market for maternity care in the Netherlands (Smits & Janssen, 2008). 253 M. M. Cruz-Cunha, I. M. Miranda,R. Simoes, J. Varajão 3 GuiMarket - Aggregating Community Resources The authors' objective was to test if a network of health care, social care and professional services providers, working articulately with an underlying effective management and intermediation service, based on an e-Marketplace, could be a powerful tool and contribute to the well-being of people with special needs, namely elderly people, and simultaneously to support their caregivers. 3.1 The motivations Some of the motivations for the development of GuiMarket were:  The isolation of the elderly;  The distance of residence to centers/facilities that provide various forms of social support or personal services;  The lack of resources in families to assist their elderly;  The difficult access to several services that would make their life more comfortable and independent day-to-day (home care, primary health care, social care services, specific or not, hygiene, feeding, monitoring, housekeeping, etc.) For this purpose, the authors developed a prototype to demonstrate the potential of an e-Marketplace as an agregator of healthcare and assistance services providers. 3.2 The proposal The eMarketplace for healthcare and social care services is an environment to coordinate and manage the match between the offer of healthcare and social care services and the individuals (users or patients). The offer and demand side is represented as follows:  Offer or services providers can be (a) entities providers of services in the covered domains of health or social care or assistance services and (b) individuals and enterprises providers of special services, that use the e- Marketplace to make available information about their products and services.  Demand consist of individuals (elderly, but also individuals with special needs and their caregivers) that use the service to satisfact their needs; currently a large majority of these targeted users cannot access these technologies, but this task can be performed by their caregivers, relatives, neighbors or friends. They can require daily home assistance of hygiene, special care, health care, physiotherapy, care giving, nursing, etc. Figure 1 represents these classes of participants and their interaction with the e- Marketplace. 3.3 Main activities The main activities offered by GuiMarket include search and selection of service providers are the following:  Request: Request involves the specification of the required service. This can be done navigating through the market of resources providers (or more narrowed sets of providers), or for complex situations, using a chat facility with the “broker” of the market, when the specification requires “knowledge” about the required service; 254 Aggregating Community Resources of Care and Assistance Services for Elderly  Search and Selection: Search, negotiation and selection consist of several steps: the identification of potential providers, separation of eligible resources, negotiation among these to identify the candidate resources (according to availability, price, conditions to provide the service), and finally the selection of the most suitable. Negotiation is a facility that is possible for certain classes of professional services (request for quotations is the most usual). When it is not needed negotiation, selection is made from the services directory or catalogue. For complex situations, the final selection can be controlled by the broker or in interaction with him.  Contractualization: An automated contractualization by which the user and the provider agree on the conditions to be respected in the service to be provided. Figure 1: Interaction between participants in the GuiMarket (Cruz-Cunha, Tavares, Simoes, & Miranda, 2010) 3.4 The prototype Figure 2 represents the homepage of GuiMarket (the webpage is in Portuguese language). Figure 2: GuiMarket homepage (in Portuguese language) 255 M. M. Cruz-Cunha, I. M. Miranda,R. Simoes, J. Varajão 4 Study on the perceived interest of GuiMarket The implementation of this service requires the understanding of the needs, expectancies and importance granted by the inhabitants towards the GuiMarket platform. With this purpose, the authors have undertaken a study (Cruz-Cunha, Miranda, Lopes, & Simoes, 2013) which results allow to understand the viability of the solution and the requirements to the deployment of the pilot experiment, as well as to drive the selection of domains of activities or typology of services to be offered by the platform. 4.1 Methodology and sample The methodology consisted of gathering information from a stratified random sample of residents of a number of parishes on the perceived interest of the electronic marketplace, its expected use and the services deemed most relevant, together with the demographics of the sample considering age, education, internet access, possession of a computer and internet usage. The information collection was performed at different times of day and different places of each parish, to encompass a high diversity of people and also to fulfill the defined stratification by age. It was used a semi-structured interview based on a questionnaire with open questions and closed questions. The sample is layered beginning at the age of 18. Of the 333 interviews, 18 could not be considered. Some demographic data of respondents is summarized in Table 1. It should be noted that it was a prerequisite for being respondent, to be currently or have already been a caregiver, or cohabit with people with special needs. 4.2 Discussion of results From the set of research questions addressed in the study, this section presents the conclusions for these three questions: 1. How important is the existence of an e-Marketplace of social care and assistance services? 2. What is the expected utilization of this platform? 3. What services are more important to be offered via an e-Marketplace? Importance granted to GuiMarket Table 2 represents the importance that participants attribute to the existence of an electronic marketplace for social care and assistance services. To 49.2% of the citizens, the service is stated as very important. Expected frequency of use Table 3 presents the expected frequency of use of the services provided by GuiMarket. More than a half of the inquired expect to use them several times, and only a few intent to use them frequently (daily or several times per week), what may look contradictory to the importance granted to this platform. However, the authors were able to confirm a strong relation among the recognized importance of the GuiMarket and the intention of use (Cruz-Cunha et al. , 2013). 256 Aggregating Community Resources of Care and Assistance Services for Elderly Table 1: Respondent demographics Characteristics N % Age groups < 30 years old 51 16.2 30 - 39 74 23.5 40 - 49 59 18.7 50 - 59 68 21.6 60 - 69 44 14.0 70 or more 19 6.0 Total 315 100.0 Level of education Illiterate 11 3.5 Incomplete primary education 157 49.8 Complete primary education 59 18.7 Secondary education 49 15.6 Higher education 39 12.4 Total 315 100.0 Owning a computer Have a personal computer at home 243 77.1 and Internet access at home Has Internet access 223 70.8 Does not have Internet access but has someone to help to solve a problem if 31 9.8 there is the need of Internet access Frequency of Internet Never 131 41.6 utilization Rarely 16 5.1 Sometimes 49 15.6 Often 41 13.0 Everyday 78 24.8 Total 315 100.0 Table 2: Importance granted to GuiMarket (Cruz-Cunha et al. , 2013) frequency % Not important / Little importance / Indifferent 14 4.4 Important 146 46.3 Very important 155 49.2 Total 315 100.0 257 M. M. Cruz-Cunha, I. M. Miranda,R. Simoes, J. Varajão Table 3: Expected frequency of utilization of the e-Marketplace (Cruz-Cunha et al. , 2013) frequency % Never 15 4.8 Rarely 101 32.1 Sometimes 177 56.1 Frequently (daily or a few times per week) 22 7.0 Total 315 100.0 Relevant services to be provided by GuiMarket Services such as information about healthcare services, home monitoring/ accompanying services 24 hours per day, and personal hygiene services provided at home are the ones recognized by the inquired citizens as very important to be provided by GuiMarket (Figure 3), and this indicates that the potential users will be mostly people with special needs or their family or caregivers. Figure 3: Importance granted to the proposed services to be provided by the platform 5 Digital divide impact The digital divide refers to the gap between individuals, companies, regions and countries in accessing and using ICT (Bach et al., 2013). The authors analyzed also the extent to which the access and use of ICT affects the project, by finding the dependence of the perceived interest on the platform and the sample characteristics (such as age, educational level, owning a personal computer and Internet access, among other possibilities), as an attempt to identify the impact of digital divide on the potential use of services like GuiMarket. To evaluate the chalenge of digital divide on the recent ICT-based solutions, the authors undertook a study deeply discussed in (Cruz-Cunha, Simoes, Varajão, & 258 Aggregating Community Resources of Care and Assistance Services for Elderly Miranda, 2014; Miranda, Cruz-Cunha, Varajão, & Simoes, 2014), where several hypotheses were formulated and examined: H1: There is a relationship between demographic characteristics of the sample and the importance granted to GuiMarket; H2: There is a relationship between demographic characteristics of the sample and the expected use of GuiMarket; H3: There is a relationship between access to ICT (possession of personal computer, internet access, and be internet user) and the importance recognized to GuiMarket; H4: There is a relationship between access to ICT (possession of personal computer, internet access, and be internet user) and the expected use of GuiMarket. Spearman correlation tests were performed and validated the above hypotheses, allowing to conclude that the importance granted to GuiMarket and the intended use of this service increases when the inquired has a computer at home, or Internet access, has a higher rate of Internet utilization or has an Internet user nearby. It is noted also that the inquired people with higher level of education consider making a more frequent use of GuiMarket and those under 40 years old consider using it more frequently (Cruz-Cunha et al. , 2014). 6 Limitations and future research It is certain that at this moment many potential users will be excluded by the digital divide problem, global concern of today. New progresses are being made at this level, and new actions and efforts (such as active ageing and e-inclusion) at the world-wide level policies that will gradually allow going beyond these limitations. The next step consists on the implementation of the prototype for demonstration and validation in a pilot study with a selected set of citizens in order to perform an adoption study and usability testing. After the adjustments that the adoption study will indicate, the platform will be further developed aiming at the real utilization. 7 Conclusions The results of the study enabled us to understand the feasibility of the proposed solution and also identifying the types of services to offer. These are two fundamental aspects considering the development of a platform prototype for validating the use of this innovative solution in the field of social assistance in the form of an e-marketplace of social services and health and welfare services. This platform will foster a local economy of service providers, most who currently have little occupation, on self-employment. Results indicate that such an e-Marketplace is of recognized relevance and that we can expect high involvement of the target population. If in a Europe increasingly ageing, the importance and the potential of the service industry for ageing well based on ICT is recognized, this study demonstrates that the issue of the digital divide cannot be diverted from the priorities of the European agenda. The expected impact of such products and services on the well-being and quality of life of the population continues to depend of the access to ICT. We believe the current work provides a small but important contribution towards that goal. 259 M. M. Cruz-Cunha, I. M. Miranda,R. Simoes, J. Varajão References Babulak, E. (2006). Quality of service provision assessment in the healthcare information and telecommunications infrastructures. International Journal of Medical Informatics, 75(3-4), 246-252. Bach, M. P., Zoroja, J. & Vukšić, V. B. (2013). Review of corporate digital divide research: A decadal analysis (2003- 2012). International Journal of Information Systems and Project Management, 1(4), 41-55. Blaschke, C. M., Freddolino, P. P., & Mullen, E. E. (2009). Ageing and Technology: A Review of the Research Literature. The British Journal of Social Work, 39(4), 641- 656. Bloom, D. E., Canning, D., & Fink, G. (2008). Population aging and economic growth. Commission on growth and development working paper N.º 32. Washington, DC: World Bank (available online at http://documents.worldbank.org/curated/en/ 2008/01/13198371/population-aging-economic-growth). Cruz-Cunha, M. M., Miranda, I. M., Lopes, N., & Simoes, R. (2013). An e-Marketplace of Healthcare and Social Care Services. The Learning Organization, 20(6), 408-418. Cruz-Cunha, M. M., Simoes, R., Varajão, J., & Miranda, I. M. (2014). O impacto da exclusão digital na utilização potencial de um mercado eletrónico de serviços de cuidados de saúde e serviços sociais. Revista Ibérica de Sistemas e Tecnologias de Informação, 14, 33-50. Cruz-Cunha, M. M., Tavares, A. J., Simoes, R. J., & Miranda, I. M. (2010). GuiMarket: an e-Marketplace of Healthcare and Social Care Services for Individuals with Special Needs. In M. M. Cruz-Cunha, A. J. Tavares & R. J. Simões (Eds.), Handbook of Research on Developments in e-Health and Telemedicine: Technological and Social Perspectives (pp. 904-916). Hershey, PA: Medical Information Science Reference. Cunha, M. M. (2003). Organisation of a Market of Resources for Agile and Virtual Enterprises Integration. Doctoral thesis, University of Minho, Guimarães, Portugal. Cunha, M. M., & Putnik, G. D. (2006). On the Dynamics of Agile/Virtual Enterprise Reconfiguration. International Journal of Networking and Virtual Organisations, 3(1), 102-123. Cunha, M. M., Putnik, G. D., Gunasekaran, A., & Ávila, P. (2005). Market of Resources as a Virtual Enterprise Integration Enabler. In G. D. Putnik & M. M. Cunha (Eds.), Virtual Enterprise Integration: Technological and Organizational Perspectives (pp. 145-165). London: Idea Group Publishing. Doukas, C., Metsis, V., Becker, E., Le, Z., Makedon, F., & Maglogiannis, I. (2011). Digital cities of the future: Extending @home assistive technologies for the elderly and the disabled. Telematics and Informatics, 28(3), 176-190. eMarketServices. (2007). eMarket Directory. Retrieved March 2007, from http://www.emarketservices.com/start/eMarketDirectory/index.html European-Commission. (2007). eHealth - Priorities and Strategies in European Countries. Luxembourg: Office for Official Publications of the European Communities. European_Commission. (2010a). A Digital Agenda for Europe (Communication from Commission to the European Parliament, the Council, the European Economic ans Social Committe of the Regions. COM(2010) 245). Brussels. 260 Aggregating Community Resources of Care and Assistance Services for Elderly European_Commission. (2010b). Europe 2020: A European strategy for smart, sustainable and inclusive growth. Brussels: European Commission. Communication from the Commission COM(2010) 2020 final. Available online at http://ec.europa.eu/europe2020/index_en.htm European_Commission. (2012a). The 2012 Ageing Report: Economic and budgetary projections for the 27 EU Member States (2010-2060): European Commission, Directorate-General for Economic and Financial Affairs, available online at https://ec.europa.eu/digital-agenda/en/news/2012-ageing-report-economic- and-budgetary-projections-27-eu-member-states-2010-2060. European_Commission. (2012b). Digital Agenda Scoreboard 2012: Directorate-General for Communication Networks, Content and Technology (CONNECT). Available online at http://ec.europa.eu/digital-agenda. European_Commission. (2013). Guide to social innovation (Publications Office of the European Union). Luxembourg: Directorate-General for Regional Policy. Kaplan, S., & Sawhney, M. (2000, May-June). e-Hubs: The New B2B Marketplaces. Harvard Business Review, 78, 97-103. Kaye, H. S., LaPlante, M. P., & Harrington, C. (2009). Do Noninstitutional Long-Term Care Services Reduce Medicaid Spending? Health Affairs, 28(1), 262-272. Kerzman, E., Janssen, R., & Ruster, M. (2003). e-Business in Health Care: Does it Contribute to Strengthen Consumer Interest? Health Policy, 64, 63-73. Magnusson, L., Hanson, E., & Borg, M. (2004). A literature review study of Information and Communication Technology as a support for frail older people living at home and their family carers. Technology and Disability, 16(4), 223-235. Makai, P., Brouwer, W. B. F., Koopmanschap, M. A., Stolk, E. A., & Nieboer, A. P. (2014). Quality of life instruments for economic evaluations in health and social care for older people: A systematic review. Social Science & Medicine, 102, 83-93. Miranda, I. M., Cruz-Cunha, M. M., Varajão, J., & Simoes, R. (2014). The Impact of the Digital Divide on the Perceived Interest of an e-Marketplace to Support Healthcare and Social Care Services. Procedia Technology, 16, 1471-1476. Muncert, E. S., Bickford, S. A., Guzic, B. L., Demuth, B. R., Bapat, A. R., & Roberts., J. B. (2012). Enhancing the Quality of Life and Preserving Independence for Target Needs Populations Through Integration of Assistive Technology Devices. Telemedicine and e-Health, 17(6), 478-483. OECD. (2008). OECD Health Data 2008: OECD (available online at: http://ec.europa.eu/health/ph_information/dissemination/echi/echi_25_en.pdf) OECD. (2013). Health at a Glance 2013: OECD Indicators: OECD Publishing (available online at http://dx.doi.org/10.1787/health_glance-2013-en). Palm, E. (2014). A Declaration of Healthy Dependence: The Case of Home Care. Health Care Analysis, 22(4), 385-404. Putnik, G. D., Gonçalves, P., Sluga, A., & Cunha, M. M. (2008). Virtual environments for dynamically reconfigurable Concurrent/Collaborative Engineering "virtual" teams. CIRP Annals - Manufacturing Technology, 57(1), 171-174. Rashidi, P., & Mihailidis, A. (2013). A Survey on Ambient-Assisted Living Tools for Older Adults. IEEE Journal of Biomedical and Health Informatics, 17(3). Séror, A. C. (2002). Internet infrastructures and health care systems: a qualitative comparative analysis on networks and markets in the British National Health Service and Kaiser Permanente. Journal of Medical Internet Research, 4(3), e21. 261 M. M. Cruz-Cunha, I. M. Miranda,R. Simoes, J. Varajão Smits, M., & Janssen, R. (2008). Impact of Electronic Auctions on Health Care Markets. Electronic Markets, 18(1), 19-29. Tang, P., & Venables, T. (2000). 'Smart' homes and telecare for independent living. Journal of Telemedicine and Telecare, 6(1), 8-14. Tanner, D. (2005). Promoting the well-being of older people: Messages for social workers. Practice, 17(3), 191-205. United_Nations. (2009). World population prospects: The revision 2008. New York: United Nations. Venkatesh, V., Vaithyanathan, V., Kumar, M. P., & Raj, P. (2012, 10-12 Jan. 2012). A secure Ambient Assisted Living (AAL) environment: An implementation view. Paper presented at the Computer Communication and Informatics (ICCCI), 2012 International Conference on. Xie, C., Hughes, J., Sutcliffe, C., Chester, H., & Challis, D. (2012). Promoting Personalization in Social Care Services for Older People. Journal of Gerontological Social Work, 55(3), 218-232. Zallah, S. (2005). Significant e-Marketplaces Report. In eMarket_Services (Eds.) Available from http://www.emarketservices.com/clubs/ems/artic/Significante Markets.pdf. 262 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Fuzzy optimization to improve mobile wellness applications for young-elderly József Mezei Institute for Advanced Management Systems Research (IAMSR), Åbo Akademi University, Finland RiskLab Finland, Arcada University of Applied Sciences, Finland jmezei@abo.fi Shahrokh Nikou Institute for Advanced Management Systems Research (IAMSR), Åbo Akademi University, Finland snikou@abo.fi Abstract Mobile applications and specifically wellness applications are used increasingly by different age-segments of the general population. This is facilitated by the large amount of data collected through various built-in sensors in the smartphone or other mobile devises, e.g. smart watches. Young-elderly cohort (60-75 year old individual) is probably one of the most potential user groups that would benefit from using mobile health and wellness applications, if their needs and preferences are precisely addressed. General knowledge is limited on understanding to what extent mobile wellness applications can and should provide precise recommendations which improve the users’ health and physical conditions. To address this problem, the current study identifies the potential benefits of utilizing fuzzy optimization tools to design recommendation systems that can take into consideration the (i) imprecision in the data and (ii) the imprecision by which one can estimate the effect of a recommendation on the user of the system. The proposed approach, depending on the context of use, identifies a set of actions to be taken by the users in order to optimize the physical or mental condition from various perspectives. The model is illustrated through the example of walking speed optimization which is an important issue for the young- elderly. Keywords: Possibilistic Optimization, Mobile Wellness Application, Young-Elderly, Chance Constrained Programming 1 Introduction An important issue in modern society is to continuously improve health-related decision making. The needs and requirements of different age segments are not necessarily 263 Mezei, Nikou similar to each other. The use of mobile technology can greatly support users’ health status if the underlying decision support tool can properly take into consideration the specific (and individual) requirements. An important and increasing segment of the population is young-elderly: the group of individuals aged between 60 and 75 years old (Haddon & Silverstone, 1996). The number of senior people is constantly growing which has implications for many countries around the world (e.g., economic implications on the healthcare expenditures). Statistics Finland1 reports that over 14% of the total population of Finland was between 65-79 years old and it is estimated that the total number of Finnish citizens aged 65 or over will be approximately 23% in 2020 and it will reach 25.6% in 2030. The growth rate in other EU countries shows a similar trend, for instance, it is estimated2 that 25% of the Dutch population will be over 65 years old in 2038. Given the rapid growth in aging population and the latest advancement in mobile technology, one can argue that mobile health interventions and wellness applications could be solutions to expand the range of healthcare delivery options, reducing the healthcare expenditures and supporting health-related issues of aging adults. For instance, it has been pointed out that physical activities on a regular basis could increase longevity and assist older adults to maintain their independence (Singh, 2002; Paterson, Jones & Rice, 2007). This is agreed on by Weuve et al. (2004) who show that regular walking can reduce stiffness and aid cardiovascular health and it is known as one of the most common and practical activities practiced by older adults to stay active. To help older adults to maintain and modify their walking pace, Qian et al. (2011) develop a mobile (wellness) application prototype which sends tactile feedback and auditory information to users to maintain their walking pace. Fuzzy set theory has been introduced by Zadeh (1965) as an approach to deal with uncertainty different from randomness. In many real life situations, the available data cannot be specified either precisely or by relying on the tools of traditional probability theory. In these cases, the two most important sources of uncertainty present in a data relates to (a) the lack of available information to establish precise statement (imprecision) and (b) the uncertainty inherent in quantifying natural language and linguistic description of different phenomena (vagueness). In most of health-related decision making problems we have to deal with data that is imprecise (data collection methods with sensors, manual input by users) and user preferences that are unique for every individual. For example, a heart rate measurement for an individual: (i) provides an imprecise estimation of the actual heart rate and (ii) the same value for different individuals can indicate different health conditions. In this article, we show how these two aspects can be incorporated into optimization models to be used in mobile wellness and health applications. We specify a general class of models based on imprecise information that can be tailored to specific wellness and health decision problems as a recommendation system. In the optimization model, the parameters and the variables are represented by fuzzy numbers reflecting the imprecise nature of wellness/health recommendations in the sense that they usually apply to "average" people, not to specific individuals. We formulate chance constrained 1 Population Structure 2013. http://www.stat.fi/til/vaerak/2013/vaerak_2013-2014-03-21_en.pdf 2 Population grows to 17,5 million in 2038, http://www.cbs.nl/nl-NL/menu/themas/ bevolking/publicaties/artikelen/archief/2008/2008-085-pb1.htm 264 Fuzzy optimization to improve mobile wellness applications for young-elderly models using the specified fuzzy numbers. The use of the model is illustrated through the example of walking speed optimization by considering several health-related criteria. The rest of the paper is structured as follows. First, we briefly discuss the literature related to mobile health and fuzzy sets. Next, we present an approach to personalized health and wellness decision support with smart-phones, and representation of health data using fuzzy sets; this is followed by the conceptual description of the optimization model. Then, we present a numerical illustration of the general approach in the case of a recommendations system for running exercises and in the final section the conclusion and future work is discussed. 2 Literature Review and Preliminaries In this section, we summarize the most important contributions from the literature in health-related decision support systems focusing on two perspectives: (a) systems developed for mobile devices, specifically considering the young-elderly age segment, and (b) solutions utilizing fuzzy set theory. As we will see, one can find only a limited number of research articles in the intersection of these domains. Regarding fuzzy set theory applications, fuzzy optimization tools are practically absent from health decision support systems. 2.1 Mobile Health Intervention Mobile health applications are being increasingly designed and developed to improve healthcare service delivery. These types of mobile technology-based health interventions provide support and services to (i) healthcare providers (e.g., support in diagnosis or patient management) as well as (ii) communicate between healthcare services and patients (e.g., appointment reminders and test result notification) (Free et al., 2013). There is a plethora of mobile health and wellness applications in the market, each of which has a range of functionalities and benefits. For example, to help people suffering from different types of Diabetes, Ajay and Prabhakaran (2011) argue that while, patients need to educate themselves on self-care practices such as blood-sugar monitoring, adherence to recommendations on diet, exercise and regular foot inspection, the existing mobile applications in Diabetes care provide benefits in three domains: (i) to health systems (e.g., remote patient monitoring and promoting evidence-based management through decision-support software applications); (ii) to physicians (e.g., tool for continuing medical education and receive guidelines and advice); (iii) to the patients (e.g., tool for self-management and reminders for drug intake and follow-up visits). In an attempt to maintain and improve the older adults’ walking behaviour through tactile signals and multimodal feedback, Qian et al. (2011) develop a prototype using mobile technology to amplify vibrations which enables older adults to monitor and improve their walking habits. The authors conclude that, redundant auditory information together with the tactile feedback help older adults to maintain a desired level of pace more consistently and improved their walking habits after the experiment. Another important aspect is the usage of these types of services among this particular age group. Young-elderly are lagging behind in adopting and using mobile technology- based interventions such as wellness applications. Nikou (2015) has recently shown that the research in this domain is in its early stage and concluded that most of the current 265 Mezei, Nikou service and applications, if not all, have not been designed according to the specific needs of this particular age group. Unlike the younger generations who are the main users of mobile (smart) devices and technologies, young-elderly, due to specific physical and functional challenges has different needs and preferences with regard to using (advanced) mobile services. Although they are more likely to experience different health issues such as high blood pressure and hearing loss, followed by more natural but negative health issues like memory loss, Alzheimer’s, and Parkinson’s, recent advances in mobile technology-based interventions can be used to the advantage of this age group to support their health conditions by making, for instance, simple lifestyle changes in their daily routines and start using mobile wellness applications on a regular basis. Ahtinen, Huuskonen and Häkkilä (2010) claim that mobile wellness applications by providing tracking or sharing the personal data can motivate people to exercise and perform more physical activities. By leveraging the existing built-in sensors in smart phones, a mobile wellness application which records various personal data such as the level of activity, heartbeats of a user and body temperature can provide personalized recommendations to the users. 2.2 Fuzzy sets for health-related decision making One of the main application areas of fuzzy logic and fuzzy set theory concerns decision making problems in the presence of imprecise or vague information. A typical situation of this type is when the decision has to be performed by relying on the evaluations provided by an expert or a set of experts. This is typically the case in medical decision making: doctors rely on their extensive experience when making a diagnosis. A typical example of utilizing fuzzy logic in medical decision making is presented by Yao and Ya (2001), where the authors propose models that utilize different fuzzy relations (symptoms disease, patient-symptoms). In general, different mathematical approaches based on fuzzy sets have been extensively used in medical decision making problems (Abbod et al., 2001; Mahfouf, Abbod & Linkens, 2001). In the following we look at these applications mainly from the perspective of the employed methods. The most traditional approach in applying fuzzy logic in general and specifically in medical problems is through approximate reasoning. Originating mainly from control theory, this approach captures the general behaviour of a complex system (the human body) through a set of IF-THEN rules with the antecedent and consequent conditions represented as imprecise quantities. These so called rule-based systems have been extensively applied in different diagnostic problems. For example, Oad et al. (2014) proposed a rule-based system to estimate the risk level of heart diseases. A significant part of the applications making use of fuzzy rules considers the information gained from the system as a basis for classification. An important advantage of this method, as pointed out by Nauck and Kruse (1999), is that linguistic rules are easy to interpret by the user. User can be the doctor but also a patient who is using a medical or wellness recommendation system on a mobile device. A different type of application relies on different fuzzy clustering approaches. In contrast to traditional clustering methods, fuzzy clustering assigns every object to several clusters with different membership values. This approach is widely used in different pattern recognition problems, with image segmentation being the most important application in the medical context (c.f. Chuang et al., 2006). The most widely 266 Fuzzy optimization to improve mobile wellness applications for young-elderly used model is c-means clustering that is applied for example in tumour classification. Moon et al. (2011) found that fuzzy c-means clustering provides a reliable tool to detect breast tumours. Additionally to the variety of methods, we can also identify proposals utilizing different families of fuzzy sets (i.e., considering different types and levels of imprecision). Iakovidis and Papageorgio (2011) propose to use intuitionistic fuzzy cognitive maps to account for the hesitancy in the doctors’ evaluation on the relationship between symptoms and possible diseases. Innocent and John (2004) developed a similar system using type-2 fuzzy sets and they found that accounting for this higher level of imprecision can improve the accuracy of diagnosis. In the literature, we can identify very few contributions that offer insights on using fuzzy decision support systems in the context of health or wellness solutions for elderly. In one of the few examples, Zhang, Du and Sun (2010) have formulated a context-aware reminder system as a fuzzy decision making problem. To remind aging adults to perform certain activities in their daily routines, a reminder system which prompts an alert based on predefined (planned) activities can be used. However, planned activities may be interrupted by ’disruptive’ activities which cannot be predicted in advance. To overcome these types of problems, Zhang et al. (2010) have proposed and used fuzzy logic to quantify the degree of the ‘disruptive’ activity and the urgency of the planned activity. 3 Personalized Health and Wellness Decision Support with Smartphones 3.1 Representation of Health Data Using Fuzzy Sets In this section we will provide the necessary definitions and concepts from fuzzy set theory and fuzzy optimization in order to formulate mathematically the previously described problem. Our starting point is to consider a systems perspective on providing health and wellness tips and recommendations as well as a general motivation for using fuzzy logic in health-related decision making as it was outlined by Seising (2006). The most important philosophical underpinning of the use of fuzzy logic in the medical context is Sadegh-Zadeh’s (2000) characterization of health, illness and disease as fuzzy concepts. For example, health can be understood as a fuzzy set, more precisely the complement of the fuzzy set ‘patient-hood’ which defines the extent to which a person can be considered as a patient. In line with this approach, one can aim at providing suggestions to the user that can ensure the required results are obtained to an acceptable degree. For example, a mobile wellness application monitoring different activities of the user can recommend at a given time the necessity for taking a short walk but specifying the required distance, speed, location only in an imprecise manner, as these goals depend on several pieces of information that by themselves are imprecise. Following Hudson and Cohen (1994), in the context of health and wellness related decision making, the following different types (or levels) of imprecision can be identified: 267 Mezei, Nikou 1. Numerical measurements that are approximately stable if we consider only a short time period (e.g., age or weight of the person). 2. Numerical measurements that can change significantly in a very short time period (e.g., blood pressure or blood glucose level). 3. Subjective modifier of the previous measurements (e.g., low blood pressure). In addition to these types of imprecisions, Hudson and Cohen (1994) also identify symptoms that can be assessed only subjectively and difficult to measure numerically, for example the degree of sweating. The described three measurement types provide the motivation and interpretation of utilizing fuzzy logic for health-related decision problems. The first group of measurements can be described by using crisp values: age is 55 years or weight is 68 kilogram. Naturally, even during a short walk, the age of the person changes by few minutes and the weight by few grams, but the magnitude of these does not impact the outcome of a health or wellness recommendation significantly. Using set theoretical formalism, we can say that a person either belongs to the set of people who are aged 55 years (membership value 1) or does not belong (membership value 0). To model the second type of attributes, we can make use of fuzzy sets. As the blood glucose level of a person can change significantly, already shortly after a measurement, the data collected provides an approximate, imprecise value of the real blood glucose level. Accordingly, we assign non-zero membership values not only to the actual measurement, but also values close to the observation. The most commonly used fuzzy sets are triangular fuzzy numbers; in the discussed example, we assign membership value 1 to the measured value, and using linear functions, assign memberships to numbers that are in the neighbourhood of the observation. The membership function of a triangular fuzzy number can be written as:  x  a  if a  x  b b  a  c  A( X )  x  if b  x  c  c  b  0 otherwise  where a ≤ b ≤ c, [a, c] is the support of the fuzzy number and b is the centre. As for the third type of imprecision, linguistic modelling can be applied. We can make use of linguistic labels to model expressions such as high blood pressure or low blood glucose level. The linguistic labels are modelled as fuzzy sets with support in the unit interval and then applied on the interval of the feasible values of the specific health- related concept. In the following, we describe how to use fuzzy optimization models to suggest an optimal course of action for the users in order to achieve their goals (i.e., improve their health and wellness from different perspectives). The main motivation behind this approach is that as the data is imprecise, the recommendation formulated based on this data should not be calculated in a definite way. This means that we require that the “possibility” of achieving a goal is sufficiently high, or equivalently we require that the “necessity” that an event happens is higher than a predefined threshold. 268 Fuzzy optimization to improve mobile wellness applications for young-elderly We propose to use possibilistic chance constrained programming models to make use of imprecise information in developing health-related decision support systems. Chance constrained programming was originally proposed in probabilistic environment (Charnes & Cooper, 1959) recognizing that as the input data is not precise, one can only provide a solution to a problem with keeping some level of uncertainty in the solution process. The extension of this approach using possibility and necessity measures was introduced in (Liu & Iwamura, 1998). As we discussed above, the data encountered in health-related problems can be more classified as imprecise rather than random. According to this, in specifying the constraints of our optimization models, we only want to be confident enough (for example, to the degree of 0.9) that the heart rate will go down from 120 and stabilizes around 80. In this example the imprecisions that have to be taken into consideration are:  120 is measured imprecisely, we can just say it is approximately 120;  optimal heart rate depends on the person: for some 120 is already a very high value as their average is around 70, for some it is only a medium level as their average is around 90;  the threshold for the confidence level depends on the problem context; in some cases we require higher confidence (medical decision support), while in other cases a lower value is sufficient (wellness applications). In the following, we provide a general approach on how the discussed concepts can be put together in a model, but we do not go into specific details as they can be very different in various applications. Rather we chose the case of identifying the optimal walking pace as a case to illustrate the use of the model in the next section. 3.2 The Model As we discussed, the main goal of the model is to specify a set of actions to be taken by the user to optimize his/her physical or mental condition from different perspectives. One example is the case of optimizing walking habits. In the general case, the assumption is that there is a list of attributes describing the user of the wellness/health support system and also the context of use. This can include the height or weight of the user, blood pressure or heartbeat on the one hand, and temperature, type of physical activity performed or time of the day, on the other hand. The attributes can be of two main types:  possible to be adjusted by the user immediately, for example the number of steps taken in one minute (or equivalently the walking speed);  the user has no control (at least in a considered limited time interval), for example the temperature or the weight of the user. The attributes belonging to the first class are used as variables of a recommendation system (e.g., how much the number of steps should be adjusted and what activity could be beneficial to perform). In the most extreme case, the system can recommend that the user should visit the doctor as the measured attributes indicate the possibility of a potential health problem. The following general model can encompass both wellness and health-related recommendation systems (i.e., can be used as a personal trainer and a personal doctor). 269 Mezei, Nikou In any application, the first task is to define the main objective of the system. In general, we assume that this objective can be written as finding a specification of the variable attribute(s) that minimizes the distance from an ideal state. The ideal state in turn is specified as a configuration of the attributes present in the model. The parameters and the variables of the model are represented by triangular fuzzy numbers. As it can be observed from the definition of a triangular fuzzy number presented in the previous section, three values need to be identified: the centre (most likely value) and the upper and lower limit of the set of possible values. In practice, these parameters can be obtained as a combination of: (a) general recommendations based on some medical knowledge; and (b) user-specific values collected in the system. This way the system can provide individualized and personalized recommendations. To specify the model, we need several notations. First, the parameters of the model are denoted as: p ; these parameters put constraints on the choice of optimal value 1 , p 2 ,..., pm for the variables, x , and determine the ideal physical state for the user. 1 , x 2 ,..., xn Additionally, we denote by 0 0 0 x , x ,..., the present values of the variables. The 1 2 xn objective is then to identify new values for the variables that result in physical condition that is close to an ideal one, does not require too much effort from the user (in terms of change from the 0 x values to the x values) and satisfies constraints based on the i i parameter values. According to this, the objective function of the problem is the following weighted average: minimize w I p p A x x p p  w E x x x x x (1) I dist  ( 1 ,..., m ), ( 1,..., n, 1,..., m) E ( 1,..., n, 01, 02,..., 0 n) The weights in this expression are also user-specific. In practice, these weights can be predefined for new users as the average value of weights for similar existing users. In later stages, by continuous data collection, the system can deviate from this general average and learn the individual weights of the user. The new values for the variables should be chosen by considering specific constraints on to what extent they affect the different attributes (parameters of the model). For every attribute, three things need to be specified in order to formulate the associated constraint:  the acceptable range of the attribute;  how the change in the variable(s) affect the value of the attribute;  what is the required confidence level that is sufficient to achieve. The acceptable range specified in terms of a crisp interval,  l u p , p ( l stands for lower i i  and u for upper), and the effect of the change in the parameter is realized through a predefined transformation function f ( x x p . Based on this, we specify two j , o j , i ) constraints for every attribute: Nec o l f ( x , x , p )  p   j j i i  l i, j Nec o u f ( x , x , p )  p   (2) j j i i  l i, j where l  and u  can be considered as confidence values and are specified by the user i, j i, j and also based on some general medical knowledge; and Nec stands for the necessity 270 Fuzzy optimization to improve mobile wellness applications for young-elderly measure. As f ( x x p is a fuzzy number, if we denote its membership function j , o j , i ) by f , the left-side of the two constraints can be rewritten as i, j  x  Nec o l f ( x , x , p )  p  1  sup f x , | x  p j j i i   i j  l i  Nec o u f ( x , x , p )  p  1 sup f x (3) , | x  p j j i i   i j  u i  The same constraints need to be defined for all the parameters. 4 Numerical Illustration In this section, we work out the details of the proposed general model in the case of a wellness recommendation system that helps the user to keep the optimal pace of the walking that corresponds to some predefined goals. Without loss of generality, we can assume that the range of all the variables and parameters is in the [0, 1] interval, with 1 indicating the best value of that attribute for the user and 0 indicating the worst. Naturally, this representation is not necessarily obtained through a linear transformation of the original values, for some variables, from the original range low and high values will correspond to 0 and medium values to 1. This data transformation can be done for example by asking for the opinion of doctors or estimating from previously collected data sets. The parameters in this specific application can include the following:  Weather (temperature and the intensity of rain and wind),  Route (the slope and quality of the road),  Age, weight and height of the user,  Time elapsed from the last meal,  Number of calories consumed during the day, The variable of the decision model is walking speed: the system provides recommendation to the user on increasing, decreasing or maintaining the actual speed. The parameters of the model (the attributes that are considered to be important for the user in this walking speed example) are heart rate, number of calories burnt per minute and physical fitness. The optimal solution in this example is simply obtained as the weighted average of the attributes. In practice, the weights can be determined based on the physical condition and preferences of the user in advance. We use the value (0.3, 0.4, and 0.3) for the weights of the three attributes. In Table 1, the actual values measured for the user as the basis for the recommendation and the ideal values (for that specific attribute and the specific individual using the system) are listed (the triplets represent triangular fuzzy numbers as defined in the previous section). The ideal values can be determined based on the data previously collected from the user or as a population average from the set of all users in lack of specific data. The actual speed of the user is (0.5, 0.6, and 0.7). Attribute/value Actual Ideal Heart rate (0.6, 0.7, 0.8) (0.7, 0.8, 0.9) Calories (0.3, 0.4, 0.5) (0.6, 0.7, 0.8) Fitness (0.8, 0.9, 1.0) (0.5, 0.6, 0.7) Table 1: Parameters of the Optimization Model 271 Mezei, Nikou Naturally, increasing the speed would increase the heart rate (and keeping in mind that we have transformed values for the attributes, decreasing it on the [0, 1] scale), improves on the calories, and improves the physical fitness. For simplicity, we assume that change in speed affects all the attributes to the same extent: 1% change in speed results in 1 % change in the attribute value. The constraints are specified such that it is required (in any case with η = 0.9 confidence level), that the distance from the ideal attribute value should be not more than 25 %. In the objective function, the weight of the distance from the ideal point and the required effort to increase/decrease the speed are equal. Attribute/value Actual Heart rate (0.77, 0.89, 1.0) Calories (0.38, 0.51, 0.64) Fitness (1.0, 1.0, 1.0) Table 2: Parameters based on the Optimal Recommendation The optimization problem was implemented in Excel. The optimal solution was identified as 27% increase in the walking speed. This resulted in the attribute values listed Table 2. We can see that the heart rate increased but still close to the ideal, while the number of calories burnt is higher. In practice, the system informs the user regularly through the running exercises how he/she should adjust the running speed. This information also expressed in the form of linguistic terms (“Slightly increase your speed”, “Slow down a bit”). Naturally, a suggestion such as “Increase the speed by 27%” would be completely impractical, but the choice of the appropriate linguistic term that corresponds to a speed increase of approximately 27% for the specific user would be the appropriate way for the system to communicate the results of the analysis. 5 Summary and Conclusions In this paper we present a new way of looking at health-related data used in mobile health recommendation systems and apply it specifically in wellness applications as a basis for recommendation systems for young-elderly. The proportion of young-elderly in the population increases steadily which in the near future (or in some countries already presently) can cause significant increase in health-related cost. Using information and communication technology, most importantly mobile wellness and health applications, can help this age-segment to improve their health conditions resulting in lower healthcare expenditure. In this paper, we present an optimization model that can be utilized in wellness recommendation systems to identify the best course of action in a given situation that can help to improve (or simply maintain) the physical condition of the user. We reason why there is a need for models to incorporate imprecise information in health-related decision making problems and how fuzzy set theory provides a tool for decision support. The main difference compared to traditional fuzzy models is that we do not rely on IF- THEN rules when determining recommendations but rather define ideal solutions that are unique for the user and identify optimal actions to be performed and: (i) minimizes a distance (in a possibilistic sense) from the ideal solution, while (ii) minimizes the effort necessary to perform the action. The main contribution of the paper is that it is one of 272 Fuzzy optimization to improve mobile wellness applications for young-elderly the first approaches to (i) utilizing fuzzy optimization models in health-related decision making models and (ii) building wellness recommendation systems for young-elderly. The main limitation of the paper, and also an important future research direction, is that the proposed model needs to be validated. In the future, we will perform user studies to test the model in different contexts to evaluate the performance. Additionally, as it is pointed out in the paper, in many practical situations we have to deal with different levels of imprecision: in the future we can improve the model by incorporating different families of fuzzy sets in the models. Acknowledgement This research has been funded through the TEKES strategic research project Data to Intelligence [D2I], project number: 340/12. References Abbod, M. F., von Keyserlingk, D. G., Linkens, D. A. & Mahfouf, M. (2001). Survey of utilisation of fuzzy technology in medicine and healthcare. Fuzzy Sets and Systems, 120(2), 331–349. DOI: 10.1016/S0165-0114(99)00148-7. Ahtinen, A., Huuskonen, P., & Häkkilä, J. (2010). Let's all get up and walk to the North Pole: design and evaluation of a mobile wellness application. In Proceedings of the 6th Nordic Conference on Human-Computer Interaction: Extending Boundaries (pp. 3-12). ACM. Ajay, V. S. & Prabhakaran, D. (2011). The scope of cell phones in diabetes management in developing country health care settings. Journal of diabetes science and technology, 5(3), 778–783. DOI: 10.1177/193229681100500332. Charnes, A. & Cooper, W. W. (1959). Chance-constrained programming. Management science, 6(1), 73–79. DOI: 10.1287/mnsc.6.1.73. Chuang, K.-S., Tzeng, H.-L., Chen, S., Wu, J. & Chen, T.-J. (2006). Fuzzy c-means clustering with spatial information for image segmentation. Computerized Medical Imaging and Graphics, 30(1), 9–15. DOI: 10.1016/j.compmedimag.2005.10.001. Free, C., Phillips, G., Watson, L., Galli, L., Felix, L., Edwards, P., Patel, V. & Haines, A. (2013). The effectiveness of mobile-health technologies to improve health care service delivery processes: a systematic review and meta-analysis. PLoS medicine, 10(1):e1001363. DOI: 10.1371/journal.pmed.1001363. Haddon, L. & Silverstone, R. (1996). Information and communication technologies and the young elderly. SPRU CICT Report Series No 13. Hudson, D. L. & Cohen, M. E. (1994). Fuzzy logic in medical expert systems. IEEE Engineering in Medicine and Biology Magazine, 13(5), 693–698. DOI: 10.1109/CDC.1987.272817. Iakovidis, D. K. & Papageorgiou, E. (2011). Intuitionistic fuzzy cognitive maps for medical decision making. Information Technology in Biomedicine, IEEE Transactions on, 15(1), 100–107. DOI: 10.1109/TITB.2010.2093603. 273 Mezei, Nikou Innocent, P. R. & John, R. I. (2004). Computer aided fuzzy medical diagnosis. Information Sciences, 162(2), 81–104. DOI: 10.1016/j.ins.2004.03.003. Liu, B. & Iwamura, K. (1998). Chance constrained programming with fuzzy parameters. Fuzzy sets and systems, 94(2), 227–237. DOI: 10.1016/S0165- 0114(96)00236-9. Mahfouf, M., Abbod, M. F. & Linkens, D. A. (2001). A survey of fuzzy logic monitoring and control utilisation in medicine. Artificial intelligence in medicine, 21(1), 27–42. DOI: 10.1016/S0933-3657(00)00072-5. Moon, W. K., Chang, S.-C., Huang, C.-S. & Chang, R.-F. (2011). Breast tumor classification using fuzzy clustering for breast elastography. Ultrasound in medicine & biology, 37(5), 700–708. DOI: 10.1016/j.ultrasmedbio.2011.02.003. Nauck, D. & Kruse, R. (1999). Obtaining interpretable fuzzy classification rules from medical data. Artificial intelligence in medicine, 16(2), 149–169. DOI: 10.1016/S0933-3657(98)00070-0. Nikou, S. (2105). Mobile technology and forgotten consumers: The young-elderly. International Journal of Consumer Studies. DOI: 10.1111/ijcs.12187. Oad, K., DeZhi, X. & Pinial Khan Butt, K. (2014). A fuzzy rule based approach to predict risk level of heart disease. Global Journal of Computer Science and Technology, 14(3), 16-22. Paterson, D. H., Jones, G. R. & Rice, C. L. (2007). Ageing and physical activity: evidence to develop exercise recommendations for older adults. Applied Physiology, Nutrition, and Metabolism, 32(S2E), S69–S108. DOI: 10.1139/H07- 111. Qian, H., Kuber, R., Sears, A. & Murphy, E. (2011). Maintaining and modifying pace through tactile and multimodal feedback. Interacting with Computers, 23(3), 214– 225. DOI: 10.1016/j.intcom.2011.02.007. Sadegh-Zadeh, K. (2000). Fuzzy health, illness, and disease. Journal of Medicine and Philosophy, 25(5), 605–638. DOI: 10.1076/0360-5310(200010), 25(5), 605-638. Seising, R. (2006). From vagueness in medical thought to the foundations of fuzzy reasoning in medical diagnosis. Artificial Intelligence in Medicine, 38(3), 237– 256. DOI: 10.1016/j.artmed.2006.06.004. Singh, M. A. F. (2002). Exercise comes of age rationale and recommendations for a geriatric exercise prescription. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 57(5), M262–M282. DOI: 10.1093/gerona/57.5.M262. Weuve, J., Kang, J. H., Manson, J. E., Breteler, M. M., Ware, J. H. & Grodstein, F. (2004). Physical activity, including walking, and cognitive function in older women. The journal of the American Medical Association, 292(12), 1454–1461. DOI: 10.1001/jama.292.12.1454. Yao, J. F. F. & Yao, J. S. (2001). Fuzzy decision making for medical diagnosis based on fuzzy number and compositional rule of inference. Fuzzy sets and systems, 120(2), 351–366. DOI: 10.1016/S0165-0114(99)00071-8. 274 Fuzzy optimization to improve mobile wellness applications for young-elderly Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8(3), 338–353. DOI: 10.1016/S0019-9958(65)90241-X. Zhang, D., Du, K. & Sun, L. (2010). A fuzzy logic based context-aware reminder for elders with mild dementia. Gerontechnology, 9(2), 107–108. 275 BACK 28th Bled eConference #eWellBeing June 7 - 15, 2015; Bled, Slovenia Human oriented performance management: Is there a gap between executives and non-executives? Benny M.E. de Waal HU University of Applied Sciences Utrecht, the Netherlands benny.dewaal@hu.nl Pieter T. Hofste HU University of Applied Sciences Utrecht, the Netherlands pieter.hofste@gmail.com Jeffrey Benthem HU University of Applied Sciences Utrecht, the Netherlands jeffrey.benthem@student.hu.nl Jean-Luc J.N.T. Bonnier HU University of Applied Sciences Utrecht, the Netherlands jl.bonnier.aiesec@gmail.com Rob ter Hedde HU University of Applied Sciences Utrecht, the Netherlands rob.terheddel@gmail.com Abstract For organizations it is important to translate the strategy and goals of the organization in tangible targets for the employees. Often, this leads to many Key Performance Indicators (KPI) for the employees. However, the link between their personal KPI’s and the goals of the organization is not always recognised. Therefore, based on previous research into performance management and on theories of organizational behaviour, culture and performance a model was developed to measure human oriented performance management in organizations. Human oriented performance management is all about establishing a direct connection between the objectives and strategy of an organization (or part thereof) and the activities and tasks of the people in the different processes. The research question in this paper is to what extent the dimensions of human oriented performance management do occur within organizations and how these 276 De Waal et al. dimensions are perceived by executive and non-executive employees. Using a mixed method approach, survey data was collected among 64 employees of three organizations, and additional eight interviews with executives and eight interviews with non-executives were held to explore and understand the results of the survey. The results show that continues improvement and organizational learning have the highest scores within all of the organizations. The lowest score for all the organizations is Visualization. Executives score slightly higher on most dimensions of human oriented performance management. Especially, the difference between the dimensions Action orientation and Dialogue is notable. The main conclusion is that it's unclear to which extent management and employees really talk about the performance and how to improve it. For many non-executives it is unclear what the organization objectives are and how they perform on these objectives. Keywords: Human oriented performance management, strategy execution, executives and non-executives, quantitative research, qualitative research. 1 Introduction Once Michael Armstrong said: "Performance management is a process which is designed to improve organizational, team and individual performance and which is owned and driven by line managers" (Armstrong, 2009). Performance management is an HRM process (which has become increasingly popular since the 1980s) concerned with getting the best performance from individuals in an organization, as well as getting the best performance from teams, and the organisation as whole (Dransfield, 2000). Effective performance management therefore involves sharing an understanding of what needs to be achieved and then managing and developing people in a way that enables such shared objectives to be achieved. Human oriented performance management is added value to the well-known performance management theory. Human oriented performance management is all about establishing a direct connection between the objectives and strategy of an organization (or part thereof) and the activities and tasks of the people in the different processes. Human oriented performance management ensures that the objectives and strategy of an organization are anchored in the minds and hearts of people (De Waal and Ter Hedde, 2014). For organizations it is a big deal to translate the strategy and goals of the organization in tangible targets for the employees. Organizations often make use of Key Performance Indicators (KPI) for the employees. But it may sometimes be difficult for employees to see the direct link between their personal KPI’s and the goals of the organization. In this case human oriented performance management could be useful. Human oriented performance management is about establishing a connection between the organization objectives, the strategy of the organization and the activities and tasks of the people in the processes. Human oriented performance management ensures that the organizational objectives and the strategy of the organization is translated and understandable for the employees. The basic principle of human oriented performance management is that performance improvement only becomes significant within the direct and personal work relationships. We believe that there is significant room for improvement in the attention 277 Human oriented performance management to the human factor in performance improvement and strategy execution programs within organizations. This paper begins with the theoretical foundation of this research. Thereafter, the context of the study will be described. Then, the research methodology is presented to collect data and develop an answer to the research question, followed by the description and analysis of the results. In the closing section, conclusions and discussion will be presented. 2 Theoretical perspective 2.1 Human behaviour and performance within organizations In times of challenging economic times, human behaviour and performance within organizations becomes more important (Blahová, 2012; De Waal, 2012; Pudil et al., 2015). This is where the organizations distinguish themselves from each other. The highest performing organizations will survive during difficult economic times (De Waal, 2012). So the way employees behave becomes more and more important. Human behaviour becomes the key to success for a lot of organizations (Senge, 1990; Paul and Berry, 2013). There has been a lot of changes in regard to human behaviour within the organizations. According to Wierdsma and Swieringa (2011), there are two main changes in the way employees behave within organizations. The first one is self-direction: individuals focus more on self-developing which results into more loyalty towards themselves and the organization. And the second one is co-creation of knowledge: more interaction between people leads to a new light on the way people create knowledge and meaning. Thanks to those changes people act in a different way within organizations. The way they think, operate and execute all changed. While employees become more modern, their performances increases. These individuals are seeking to deliver an added value and in return the organizations need to connect their vision, mission and objectives to that of the employees. In three steps you can subscribe what’s necessary. Firstly they need to understand the strategy of the organization. Then they should be motivated by this strategy. And finally they have to start acting according this strategy. 2.2 Human Oriented Performance Management Human oriented performance management is all about establishing a direct link between the objectives and strategy of an organization and the activity of the people that execute the processes on the different workplaces. For an organization to achieve their objectives and strategy the following three cases are important (De Waal and Ter Hedde, 2014):  Organizations should translate their strategy in a way so every employee understands it.  Every employee should be motivated by the eventual translated strategy.  Organizations should perform the right actions to get the right things done. Only when an organization is successful on all of these three cases, she would be able to achieve her objectives. 278 De Waal et al. Figure 1: Human Oriented Performance Management-model The core of human oriented performance management (HOPM) is shown in Figure 1 (De Waal and Ter Hedde, 2014). The model shows that the link between strategy and action within an organization is established by means of initiating and keeping in motion of two continuous learning loops/improvement loops. "Are we doing the right things?" and "are we doing things right?" are the important questions within the model. The model shows that within organizations both a top-down (control) as a bottom-up (self-organization & feedback) motion must be visible forming a dynamic balance together. The 'moments of truth' of human oriented performance management are shown in the heart of the model above: The dialogue and feedback of results and relation between management and employees (How are we doing? And how are we working together?). The actual implementation of the strategy, the associated changes and the resulting performance improvements are only meaningful within the direct and personal relationships where everything comes together. Those are the catalysts of change and improvement. The HOPM model is built around four important dimensions:  Strategy translation: To what extend are the objectives and strategy of the organization translated into a focused, well-balanced set of Key Performance Indicators?  Dialogue and action orientation: To what extend are management and employees involved in dialogues and focused on actions to improve performance?  Continuous improvement and organizational learning: To what extend are management and employees focused on challenging themselves and the current performance of the organization?  Information, measurement tools and visualization: To what extend is the information within reports & dashboards easy to understand and can it easily be communicated? Furthermore does the information reflect current (KPI) performance? 279 Human oriented performance management In this paper we address the following three exploratory research questions:  To what extent do the dimensions of human oriented performance management occur within organizations;  How are these dimensions perceived by executive and non-executive employees;  What are the differences between executive and non-executive employees? Based on previous research into human oriented performance management and on theories of organizational behaviour, culture and performance an on-line questionnaire was developed to measure human oriented performance management in organizations. To understand the motives behind the findings of the on-line questionnaire, interviews were conducted under the participators of this research. 3 Three organizations operating in different markets For this paper research was conducted in three different organizations. The organizations were contacted as part of the study program Business Information Management from the Utrecht University of Applied Sciences. The organizations operate in different markets: media and communication, financial services and the construction sector. The departments of the organizations were the research was done mainly focus on the Dutch market. The case study organisation in the media and communication sector has 35 employees and it’s primarily goal is to provide media, brand and advertisement consultancy. The organisation in the financial services sector has taken place within one department of a Dutch Bank. The main goal of this department is to sell mortgages and has around 35 employees. The case study in the construction sector took also place within one department. In this department the primarily goal is railway maintenance and has around 75 employees. 4 Research methodology 4.1 Data collection of the quantitative research The empirical approach was to collect data from executive and non-executive employees in three organizations from different sectors. They were asked about their experiences in relation to the performance management function within their organization. The aim was to collect data on the four dimensions of human oriented performance management. The survey was conducted using a web-based tool and it was sent to the respondents’ corporate mail address. The respondents had a deadline of 14 days to fill in the survey. After a week a reminder was sent. In this survey the questions were based on the four dimensions of the human oriented performance management model: strategy translation, dialogue and action orientation, continues improvement and organizational learning, information/measurement tools and visualization. All data was collected in November 2014. From the people that were contacted, 78 were willing to participate, of which 64 fully completed the questionnaire. Of these respondents, 84% were male and 16% were female. Of the respondents, 38% were executives and 62% were non-executives. The 280 De Waal et al. average age of the respondents was 44 years old and their ages ranged from 22 to 60 years. Of the respondents 25% was employed less than 10 years by their organization, 41% was employed between 10 and 25 years and 34% longer than 25 years. Our sample of respondents had an average education: 46.9% of the respondents held a community college degree, 37.5% held a bachelor’s degree and 15.6% held a master’s degree. The respondents were employed in the fields of media and communication (14.0%), financial services (26.6%) and construction sector (59.4%). 4.2 Instrument validation 4.2.1 Strategy translation In order to validate the measurement of strategy translation, factor analysis was performed to analyse the construct validity of 6 items. Principal component analysis (PCA) with varimax rotation resulted in a two-factor solution with own values of 2.31 and 1.65, accounting for 38.5% and 27.6% of the explained variance. Table 1 shows the results The factor loadings were between 0.664 and 0.901, which can be considered as being significant (Hair et al, 1998).The reliability of the two scales – a four-item Goal- setting scale (ST01-ST04) and a two-item Participation scale (ST05-ST06) – was confirmed by Cronbach’s alpha value of 0.742 and 0.718 respectively (cf. Nunnally and Bernstien, 1994). No. Item Goal-setting Participation ST01 The organization has translated its objectives in clear .828 .121 measurable Key Performance Indicators (KPIs). ST02 The balance between financial and non -financial KPI’s is .760 .173 optimal. ST03 The non-financial KPI’s measure at least customer .664 -.206 satisfaction. ST04 The KPI’s are a good translation of the organization's .754 .268 objectives. ST05 A representative part of the employees were involved in -.012 .901 the strategy translation process. ST06 The strategy translation is widely accepted within the .195 .826 organization. Table 1: Factor loadings based on PCA analysis of items measuring Goal-setting and items measuring Participation (N=63). 4.2.2 Dialogue and action orientation Our measurement of dialogue and action orientation can be validated by factor analysis to analyse the construct validity of the group of 7 items. Principal component analysis (PCA) with varimax rotation resulted in a two-factor solution with own values of 2.37 and 1.81, respectively accounting for 33.9% and 25.9% of the explained variance. Table 2 shows the results. The factor loadings were between 0.578 and 0.811, which can be considered as being significant (Hair et al, 1998). The reliability of the two scales – a four-item Dialogue scale (DA01-DA04) and a three-item Action orientation scale (DA05-DA07) – was confirmed by Cronbach’s alpha values of 0.716 and 0.705 respectively (cf. Nunnally and Bernstien, 1994). 281 Human oriented performance management No. Item Dialogue Action orientation DA01 There is continuous dialogue between management and .756 .024 employees. DA02 Staff meetings between management and employees are .578 .217 held several times per month to discuss how to improve performance. DA03 Improving performance is always a separate item on the .672 .305 agenda of staff meetings. DA04 Mutual work relations and cooperation are always .811 .134 discussed during staff meetings. DA05 The reported information is consistently used for .304 .755 performance analysis of the results and the conversion into actions. DA06 Within our organization, it is completely clear which .508 .684 performance standards and targets need to be met. DA07 Within our organization, it is completely clear to which -.063 .784 extend the performance standards and targets are met. Table 2: Factor loadings based on PCA analysis of items measuring Dialogue and items measuring Action orientation (N=63). 4.2.3 Continuous improvement and organizational learning To validate the measurement of continuous improvement and organizational learning, factor analysis was performed which analysed the construct validity of 6 items. Principal component analysis (PCA) with varimax rotation resulted in a two-factor solution with an own values of 2.01 and 1.51, accounting for 28.7% and 21.5% of the explained variance. Table 3 shows the results. All factor loadings were between 0.581 and 0.855, which can be considered as being significant (Hair et al, 1998). The reliability of the two scales – a four-item Continuous improvement scale (CO01-CO04) and a two-item Organizational learning scale (CO05-CO06) – was partly confirmed by Cronbach’s alpha value of 0.683 and 0.524 respectively (cf. Nunnally and Bernstien, 1994). No. Item Continuous Organizational improvement learning CO01 Management is continuously coaching their employees .855 -.041 to improve results. CO02 The reported results are consistently used to evaluate .675 -.045 previously specified standards and targets. CO03 Management and employees provide performance .581 .335 feedback to each other. CO04 The organization is performance oriented. .631 .393 CO05 Employees are coaching each other to improve results. -.056 .837 CO06 Employees want to take responsibility for their results. .294 .731 Table 3: Factor loadings based on PCA analysis of items measuring Continuous improvement and items measuring Organizational learning (N=63). 282 De Waal et al. 4.2.4 Information, measurement tools and visualization The measurement of information, measurement tools and visualization can be validated by factor analysis to analyse the construct validity of the group of 7 items. Principal component analysis (PCA) with varimax rotation resulted in a two-factor solution with own values of 2.02 and 1.86, respectively accounting for 28.8% and 16.3% of the explained variance. Table 4 shows the results. The factor loadings were between 0.530 and 0.873, which can be considered as being significant (Hair et al, 1998). The reliability of the two scales – a four-item Information scale (IV01-IV04) and a two-item Visualization scale (IV05-IV07) – was partly confirmed by Cronbach’s alpha values of 0.666 and 0.651 respectively (cf. Nunnally and Bernstien, 1994). No. Item Information Visualization IV01 All necessary information is available to the management .822 -.075 to be able to evaluate performance. IV02 The information in the reports is designed in such a way .671 .155 that the message within the information can be read at a glance. IV05 Reports are easily accessible to everyone. .699 .227 IV06 The reports contain information from all relevant source .530 .293 systems within the organization. IV07 The capabilities of mobile devices (smartphones and pads) .105 .804 is taken into account when reports and dashboards are designed. IV08 All reports can also be accessed via mobile devices (Pads, .062 .873 smartphones). IV09 Within the (mobile) reporting environment it is possible to .327 .535 share information with each other via e-mail, discussion forums and/or chats. Table 4: Factor loadings based on PCA analysis of items measuring Information and items measuring Visualization (N=63). 4.3 Data collection of the qualitative research 4.3.1 Analyse procedure Eight interviews with executives and eight interviews with non-executives were held to explore and understand the results of the survey. For each case study the interviews were analysed using a cumulative editing approach (Runeson and Höst, 2009). Each interview report was read carefully by the researchers in order to determine the meaningful fragments of text. These fragments were coded using open coding. Fragments of text from within one interview and between interviews within the same case study were compared in order to determine whether or not they had the same code. If necessary, it was decided to merge codes or to change a fragment to another code following an axial coding procedure. This procedure was repeated for the other case studies. Thereafter, the fragments and codes of the three case studies were compared. In addition, when necessary, changes were made to codes, and fragments were replaced. The last step was to structure the codes at the level of main- and sub- variables/dimensions using selective coding. Thereafter the three cases were compared which resulted in a structured identification of fragments relating to the different 283 Human oriented performance management concepts of human oriented performance management (Miles and Huberman, 1994; Boeije, 2002). 4.3.2 Validity procedure of interview data In this investigation four aspects of validity were applicable: construct validity, internal validity, external validity, and reliability (Yin, 2009). Construct validity in this study was handled by using multiple sources of evidence and defining measurements by a protocol that was used to each case study. The internal validity was protected by conducting interviews with several actors in order to cross-check documentation, and to check statements made in different interviews. To govern external validity, multiple case studies were set up for comparison, in particular with regard to the different dimensions of human oriented performance management. Finally, to ensure reliability, interview reports were sent to interviewees for approval. To generally govern validity, the case study protocol and a case study database was created and communicated with all subjects. 5 Results 5.1 Results of survey In this section the results of the quantitative research are discussed. First, we describe the results of human oriented performance management in the different organizations. Second, the results between executives and non-executives are shown separately. In Table 5 the results of human oriented performance management on each sub variable are shown. The items of each variable had four answer categories (1 = fully disagree, 4 = fully agree). For all respondents together Continues improvement and Organizational learning have the highest scores (2.85 and 2.89 respectively). The lowest score (2.42) for all the organizations is Visualization. Divided by organization, we see that media and communication has the highest scores on Organizational learning and Continues improvement (3.22 and 2.92 respectively). The financial organization has the highest score on Goal setting (2.81). The construction organization has the highest scores on Participation and Action orientation (2.69 and 2.57 respectively). Comparing the means between the different organizations shows that only the difference between the media and communication organization and the construction organization on Organizational learning is significant (p < .034). Alle respondents Financial Media and Construction (N=64) services(N=17) communication(N=9) (N=38) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Goal-setting 2,66 0,48 2,81 0,51 2,56 0,37 2,62 0,48 Participation 2,58 0,54 2,38 0,45 2,50 0,43 2,69 0,58 Dialogue 2,70 0,50 2,63 0,52 2,78 0,52 2,71 0,50 Action orientation 2,53 0,47 2,51 0,50 2,41 0,40 2,57 0,48 Continues improvement 2,85 0,42 2,76 0,38 2,92 0,35 2,88 0,45 Organizational learning 2,89 0,50 2,82 0,50 3,22 0,44 2,84 0,50 284 De Waal et al. Information 2,50 0,42 2,59 0,52 2,42 0,25 2,49 0,40 Visualization 2,42 0,55 2,33 0,54 2,37 0,39 2,47 0,59 Table 5: Descriptive analysis of sub-scales of human oriented performance management for different organizations. Table 6 shows the results of the scores on human oriented performance management for executives and non-executives. Comparing the results of the executives and the non- executives it can be concluded that executives score slightly higher on the most variables. However, only the difference between Action orientation is significant (p < .012). Mean scores Two sided t-test of equality of means Non- degrees Sub-variable Executives p- (N=24) Executives difference t- of (N=40) value freedom value Goal-setting 2,71 2,62 0,09 0,719 62 ,475 Participation 2,60 2,58 0,02 0,208 62 ,836 Dialogue 2,83 2,61 0,22 1,749 62 ,085 Action orientation 2,71 2,41 0,30 2,575 62 ,012 Continues improvement 2,95 2,79 0,16 1,496 62 ,140 Organizational learning 2,88 2,88 0,00 0,000 62 1,000 Information 2,59 2,45 0,14 1,347 61 ,183 Visualization 2,42 2,42 0,00 -0,015 61 ,988 Table 6: Two sided t-test of sub-variables of human oriented performance management for executives and non-executives. 5.2 Results of interview data How can the difference between executives and non-executives explained? And what are the reasons of the differences between the organizations? The semi-structured interviews that were conducted after the survey provide some answers to these question. The findings will be explained in terms of the (sub-) dimensions of the HOPM-model. 5.2.1 Goal-setting and participation From the interviews it appears that the organizations have translated their organizational objectives into KPI’s. Most executives and non-executives indicated that the strategy of the organization is well known by the employees. Most of these organizational objectives are in line with the personal objectives of the employees. However, it seems that organizations struggle with the creation of a good balanced set of KPI’s. Often, there are too much KPI’s and they not always justify the word “critical”. Also, there’s not always a balance between the financial and non-financial KPI’s. From one organization it became clear that the link between what the employees do every day and what kind of impact this has on the results of the organizational objectives was not discernible. 5.2.2 Dialogue and action orientation All of the organizations seem to have frequent work meetings between management and employees. These work meetings are held in various forms. Two of the organizations make us of daily- and weekly starts. The other organization hold their work meetings 285 Human oriented performance management periodically. According to executives and non-executives, in all organizations employees have the opportunity to have their own contribution. However it is still clear that these meetings are guided by the executive. In two of the organizations performance improvement are often discussed, however it’s not a permanent item on the agenda. Most non-executives indicated that management information is rarely used during work meetings. Also for non-executives, targets and standards are not entirely clear for everyone and it is not clear to everyone how they satisfy the conditions at the moment. This can explain the difference between executives and non-executives on action orientation in the survey. 5.2.3 Continuous improvement and organizational learning Executives and non-executives seem to find coaching an important factor. Much time is spent on coaching within all three of the organizations. This confirms the findings in the survey. However most of the coaching takes place between executive and non- executives. Coaching between employees occurs, but seem to be difficult because of group dynamic issues. Non-executives indicated that it is difficult to talk with their colleagues about their performance. Further, in all organizations management allows employees to make mistakes although they do not allow to make the same mistake too many. To stimulate improvement, successes are shared but executives and non- executives find that they do not celebrate them exuberantly. 5.2.4 Information and visualization Regarding information provision, most non-executives stated that the management reports are not easy accessible for the employees. In some cases it’s difficult to access the reports at all, in other cases the accessibility of the reports depends on the function of the employee. The content of the management reports are mostly financial in nature and contain a lot of tables and graphs. For some non-executives the management reports are difficult to interpret. In two of the cases the information in reports have a direct link with the organization objectives. In the other case the reports were not used during work meetings and doesn’t had a direct link between the content and the organization objectives. 6 Discussion, conclusion and implications This paper presents a study on the application of human oriented performance management. Human oriented performance management consists of four dimensions: Strategy translation, Dialogue and action orientation, Continuous improvement and organizational learning, and Information and visualization. These dimensions we investigated as part of activities in the field of performance management within three different organizations in order to answer the three research questions: (1) To what extent do the dimensions of human oriented performance management occur within organizations; (2) How are these dimensions perceived by executive and non-executive employees; and (3) What are the differences between executive and non-executive employees? Data was collected with a survey from 64 employees of three organizations. Furthermore qualitative data was collected by means of interviews with eight executives and eight non-executives. The results show some interesting findings. Overall it can be concluded that the three organizations are all working with the following structure and 286 De Waal et al. process elements of performance management: KPI's, management reports and work meetings. The organizations score relatively high on these elements. This mean there is a lot of potential to use this structure and processes to increase the performance of the organization. The human factor is really important at this point. However, the results show that it is unclear to which extent executives and non-executives really talk about the performance and how to improve it. For many non-executives it is unclear what the organization objectives are and how they perform on these objectives. Besides that, the connection between the KPI's and what the people do in their daily work is moderate. Performance improvement is not always a permanent item on the agenda. At least, if the organizations already use management information in work meetings, the quality is low and the information is difficult to analyse. The story behind the figures often remain hidden. Looking at the different organizations we noticed some disagree between the executives and non-executives with regard to the structure of the KPI’s. Because of this not every non-executive is able to understand the goals of their organization. Based on these findings we conclude that sharing targets with employees is very important. During the various meetings executives should communicate more about improvement of their results. Using clear visual and relevant information during meetings will be helpful. Being clear about your targets and showing the scores of the KPI’s is something not every organization does. The executives of the organizations do not always have a clear KPI’s-structure. Attention for KPI’s-structure and dialogue is limited in different organizations. An analysis of the data reveals that asking questions and involving employees at the right places is the key to success. To improve the results of the organizations, executives should change their KPI’s-structure. When there will be a clear structure, employees will perform better. Based on the results it can be recommended that organizations improve the communication between executives and non-executives. Sharing ideas during meetings with regard to improving the results will ensure executives and non-executives have the same information. Finally it can be suggested that organizations use more relevant information and ensure that the information they are using is clear for everyone. Although this research was carefully designed, there are some limitations. Although different case study organizations were involved, the generalizability of the findings are limited. More (case) studies within the same branches and other branches are needed. Also the limited responses must be taken into account to generalize the findings. Furthermore, we noticed that the reliability check of the variables were partly confirmed. Not all variables reached the generally used threshold of 0.7. This mean that the items of the variables should be further examined. Despite these limitations, we believe this paper has demonstrated that the HOPM model is a useful basis for the empirical study of the practice of performance management. Acknowledgement The authors wish to acknowledge the different case study organizations for making it possible to investigate the practice of human oriented performance management. Without their corporation it would not have been possible to collect the data for this research. In that respect, many thanks also to the many respondents who were willing to fill in the questionnaires or willing to be interviewed. 287 Human oriented performance management References Armstrong, M. (2009). Armstrong’s handbook of performance management: An evidence-based guide to delivering high performance. London: Kogan Page. Blahová, M. (2012). Bata and amoeba: Successful management systems to maximize corporate performance. In: J. Politis (Ed.) Proceedings of the 8th European Conference on Management Leadership and Governance ECMLG 2012, Pafos, Cyprus, 8-9 November 2012 (39-49). Reading: Academic Publishing International Limited. Boeije, H. (2002). A purposeful approach to the constant comparative method in the analysis of qualitative interviews. Quality & Quantity. Vol. 36(4), 391-409. De Waal, A. (2012). What makes a high performance organization? Cranbrook: Global Professional Publishing Ltd. De Waal, B.M.E. and Ter Hedde, R. (2014). Is human oriented performance management the key to high performance organizations? In: P. Dover, S. Hariharan and M. Cummings (Eds.) Proceedings of the 2nd International Conference on Management Leadership and Governance ICMLG 2014, Wellesley Massachusetts, USA, 20-21 March 2014 (78-86). Reading: Academic Publishing International Limited. Dransfield, R. (2000). Human resource management. Oxford: Heinemann Educational Publishers. Hair, J.E., Anderson, R.E., Tatham, R.L. and Black, W.C. (1998). Multivariate Data Analysis. New Jersey: Prentice-Hall. Miles, M.B. and Huberman, A. (1994). Qualitative Data Analysis: An Expanded Sourcebook. Thousand Oaks: Sage Publications. Nunnally, J.C. and Bernstien, I.H. (1994). Psychometric Theory. New York: McGraw- Hill. Paul, G.W. and Berry, D.M. (2013). The importance of executive leadership in creating a post-merged organisational culture conducive to effective performance management. SA Journal of Human Resource Management/SA Tydskrif vir Menslikehulpbronbestuur. Vol. 11(1), 446-461. Pudil, P., Pirozek, P., Somol, P. and Komarkova, L. (2014). Identification of key organization components that influence enterprise performance by means of non- linear regression. In: V. Grozdanic (Ed.) Proceedings of the 10th European Conference on Management Leadership and Governance ECMLG 2014, Zagreb, Republic of Croatia, 13-14 November 2014 (278-284). Reading: Academic Publishing International Limited. Runeson, P. and Höst, M. (2009). Guidelines for conducting and reporting case study research in software engineering. Empirical Software Engineering. Vol. 14(2), 131-164. Schraeder, M. and Jordan, M. (2011). Managing performance. Journal for Quality & Participation. Vol. 34(2), 4-10. Wierdsma, A. and Swieringa, J. (2011). Lerend Organiseren en Veranderen. Groningen: Noordhoff Uitgevers. 288 De Waal et al. Yin, R.K. (2009). Case study research. Thousand Oaks: Sage Publications. 289 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs Michael H. Quade University of Applied Sciences and Arts Northwestern Switzerland FHNW, Switzerland michael.quade@fhnw.ch Uwe Leimstoll University of Applied Sciences and Arts Northwestern Switzerland FHNW, Switzerland uwe.leimstoll@fhnw.ch Abstract Today, mobile devices like smartphones and tablets are omnipresent in many parts of the world. They are used for private and business activities. The effects of mobile business are discussed more and more in micro-enterprises as well as in small and medium-sized enterprises (SMEs). The question is: Do these devices have an impact on the productivity, flexibility and business processes of companies? The goal of this paper is to develop an explorative model that helps to identify and explain these effects. The investigation is based on a quantitative empirical study conducted among 900 Swiss SMEs. The model is estimated and evaluated using Partial Least Square (PLS) structural equation modelling. The results show that the number of mobile devices used and the portion of work carried out offsite have only a low impact on the perceived value of smartphones and tablets. On the other hand, the impact on value is high if mobile devices support business processes and if the variety of information used is high. Keywords: Empirical Study, SME, E-Business, Mobile Business, Perceived Value of ICT 1 Introduction Companies in Switzerland show an increasing interest in mobile business. Mobile data connec- tions are increasingly affordable and data transmission is getting faster. Many consumers and employees use powerful smartphones and tablets. The term "mobile business" is often defined as e-business via mobile networks using mobile devices (Lehner, 2003; Meier & Stormer, 2009). The main difference between mobile business and e-business is therefore the mobile information and communication technology (ICT), e.g. the type of devices and network connections. Mobile ICT offers additional functionality, e.g. continually identifying the location of a device or person (Schiller & Voisard, 2004). Other im- portant characteristics of mobile ICT are portability and ubiquity (Junglas & Watson, 2003). Smartphones and tablets are always switched on and thanks to mobile network connections they can be used nearly everywhere. The mentioned features support mobile work, for example, by 290 Michael H. Quade, Uwe Leimstoll providing access to information and information systems (Figure 1). In this paper, mobile work is defined as work processes that take place offsite, meaning outside the company’s physical locations (see chapter 3.2). The usage of mobile devices during mobile work depends further on the information needed, the business processes and the amount of mobile devices. Finally, companies use smartphones and tablets because they expect positive effects such as flexibility, productivity, and the ability to redesign business processes (Basole, 2004; Scherz, 2008). Based on these considerations, a basic chain of effects can be set up (Figure 1). Scherz (2008) empirically tested a similar model. Nevertheless, the dependencies between the variables have not been evaluated and explained so far. The goal of this paper is to find out if mobile business has a positive impact on company performance and to identify constructs and indicators that explain the impact. Proportion Accessibility of mobile work Productivity Reachability Ubiquity Information used when mobile mobile ICT Business process Flexibility Localization Portability support by mobile devices Connectivity Redesign Number of mobile of business processes devices used Characteristics Usage Value Figure 1: Chain of effects model of mobile ICT (adapted from Basole, 2004; Scherz, 2008) Two main research questions guide the investigation described in this paper: RQ1: What impact do the characteristics of mobile ICT have on the usage of smartphones and tablets in SMEs? RQ2: What effect does the usage of smartphones and tablets have on the perceived value of mobile ICT in SMEs? The paper is structured as follows: In chapter two, related work is discussed and the gap in the research is described. Chapter three describes the model built, its constructs and the hypo- thetical relationships. Chapter four shows how data is collected and explains the sample. Chap- ter five presents the estimation and evaluation of the model with PLS and the interpretation of the results. The paper ends with a discussion and conclusions. 2 Related Work Several studies on mobile business and the usage and value of mobile ICT have already been conducted. Most of the studies are consumer oriented, focusing on mobile commerce. Some studies focus on the satisfaction and value the consumers will gain with mobile ICT (Anckar & D’Incau, 2003; Chong, Chan, & Ooi, 2012). Other studies examine which factors influence the behaviour and use of smartphones and tablets in mobile commerce (Venkatesh, Thong, & Xu, 2012; Yang, 2010). The literature includes studies on companies’ mobile business. They examine the use of mobile ICT by employees and the effects of the potential and value of mobile ICT such as higher flexi- bility, efficiency and effectiveness. They show the same user centric approach as the studies on mobile commerce mentioned above. Gebauer (2008) evaluated the potential and value on the basis of the technology acceptance model (TAM) (Venkatesh & Davis, 2000) and the technolo- 291 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs gy-to-performance chain (TTF) (Benbasat & Barki, 2007; Goodhue & Thompson, 1995). Gebauer (2008) also examines the effects of mobile work and the portability of ICT based on these user-centric approaches. Picoto, Bélanger, & Palma-dos-Reis (2014) take a broader view on business effects from an organizational perspective. They analyse the value of mobile ICT in the main three (e-)business areas: procurement, internal organization, and commerce and mar- keting. The research in this paper does not focus on the single employee (e.g. the user) but ra- ther on the organization and structures of the companies themselves. A theoretical basis of this paper can be seen mainly in the interdisciplinary approach of coordi- nation theory (Malone & Crowston, 1994). Coordination theory is widely used in the area of information systems because the value of these systems can be explained with a reduction of communication and coordination costs. As mobile devices are also information systems, and, in particular, as they directly support communication processes, coordination theory suggests itself as an appropriate theoretical basis. Other theories are partly relevant, e.g. the innovation diffu- sion theory (Rogers, 1983), the resource-based view (Wernerfelt, 1984) and the market-based view of the company (Porter, 1985). 3 The Research Model Along with the preliminary model in Figure 1 and the work of Basole (2004) and Scherz (2008), this section derives five sets of research hypotheses about the cause and effect chains of mobile ICT. The sub-sections focus on the respective exogenous variables and explain how and why they affect different endogenous variables. The individual hypotheses will be combined to cre- ate a comprehensive research model to show further interdependencies between the individual variables. 3.1 Characteristics of Mobile ICT The particular characteristics of mobile ICT – depicted in the preliminary model (Figure 1) – distinguish it from other forms of information technology. The significance of these characteris- tics for the mobile work of SMEs can be used as an indicator to measure the importance of mo- bile ICT (Tarasewich, Nickerson, & Warkentin, 2002). The following characteristics are taken into account in this study: accessibility (having access to information resources anywhere and anytime), reachability (being reachable anywhere and anytime), portability (being able to take devices with you) and localization (pinpointing the localization of the user) (Basole, 2004; Jun- glas & Watson, 2003). Ubiquity is not taken into account separately because it is a combination of the other characteristics mentioned. Connectivity is also not taken into account because it is a prerequisite for accessibility, reachability and localization. If these typical characteristics of mobile ICT are significant for a company, it can be assumed – following the TTF framework (H1a and H1b) and caused by a reduction of communication costs (H1c) – that they have a positive impact on the following three aspects: (a) the support of operational business processes with smartphones and tablets, (b) the number of smartphones and tablets used per FTE (full-time equivalent), and (c) the diversity of information used in mobile work scenarios. This leads to the following hypotheses: H1a: A higher significance of the characteristics of mobile ICT has a positive impact on the support of operational processes with smartphones and tablets. H1b: A higher significance of the characteristics of mobile ICT has a positive impact on the number of smartphones and tablets used per FTE. H1c: A higher significance of the characteristics of mobile ICT has a positive influence on the extent of the types of information used in mobile work. 292 Michael H. Quade, Uwe Leimstoll 3.2 Proportion of Mobile Work As already mentioned, this study is limited to mobile work processes taking place offsite. These mobile processes can be very diverse (Buser & Poschet, 2002; Gareis, 2003). Regardless of this diversity, it makes sense to use more mobile ICT to reduce communication and coordination costs the more working time employees have to spend offsite (outside the company). Therefore, it can be assumed that the ratio of mobile work time positively influences (a) the support of operational business processes with smartphones and tablets, (b) the number of smartphones and tablets user per FTE, and (c) the types of information used in mobile work scenarios: H2a: The proportion of time spent on mobile work offsite, has a positive impact on the sup- port of operational processes with smartphones and tablets. H2b: The proportion of time spent on mobile work offsite has a positive impact on the num- ber of smartphones and tablets used per FTE. H2c: The proportion of time spent on mobile work offsite has a positive influence on the ex- tent of the types of information used in mobile work. 3.3 Process Support by Smartphones and Tablets Mobile business can be divided into three areas: mobile procurement (mobile support of pro- curement processes), mobile organization (mobile support of internal processes) and mobile commerce (mobile support of sales and distribution processes) (Möhlenbruch & Schmieder, 2001). These main areas can be further divided using the process areas of Porter’s value chain (Porter, 1985). It is obvious that companies will not support all processes with mobile ICT at once. The diffusion takes place step-by-step, depending on the needs, the knowledge and the financial resources of the company as well as on the available applications. Due to the available resources, large companies more often develop individual software to support mobile work than SMEs (Walter & Sammer, 2012). The latter prefer to buy standard ERP software (Leimstoll & Quade, 2011). As only a few major business software providers in Switzerland offer mobile apps for their ERP systems to date, the use of mobile ICT in direct connection with business software depends on the availability of the necessary applications. Furthermore, it indicates a higher maturity level than just the support of e-mail and calendar functions (Basole, 2007). As a consequence, the mobile access to business software (known as process support) can cause a higher potential of mobile ICT. Therefore, one can assume that the availability of mobile access to business software has a positive impact on the number of smartphones and tablets used per FTE as well as on the perceived value of mobile ICT. Useful indicators for the perceived value can be seen in increased productivity and flexibility as well as in the option of reorganizing business processes (Scherz, 2008). H3a: The support of mobile processes with smartphones or tablets has a positive impact on the number of smartphones and tablets used per FTE. H3b: The support of mobile processes with smartphones or tablets has a positive impact on the perceived value of smartphones and tablets. 3.4 Information Used when Mobile The information needs in mobile work processes depend on the business sector and on the tasks that have to be fulfilled (Varian, 2010). A construction worker needs different information than an architect or a real-estate agent. Thus, the amount and types of information are very individual and can be very varied. A wide range of information needs means that the employee has to physically transport many documents or other sources of information as long as no access is available to the electronic versions of these documents or data (Scherz, 2008). Thus, it can be 293 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs assumed that the amount and variation of information needed has an influence on the number of smartphones and tablets used per FTE as well as on the perceived value of mobile ICT: H4a: The extent of the types of information used in mobile work has a positive impact on the number of smartphones and tablets used per FTE. H4b: The extent of the types of information used in mobile work has a positive impact on the perceived value of smartphones and tablets. 3.5 Number of Smartphones and Tablets Used Finally, the number of smart devices used in a company might have a direct positive influence on the perceived value of these devices. In some way, this seems to be tautological: The more employees use smartphones and tablets, the greater is the likelihood of realizing productivity or flexibility improvements. However, it has not yet been shown that the use of smart devices ac- tually leads to an increase in productivity and flexibility and that processes can be redesigned. To shed some light on the effects that can be achieved through the use of smart devices the fol- lowing hypothesis is made: H5: The number of smartphones and tablets used per FTE has a positive impact on the per- ceived value of smartphones and tablets. 3.6 Structural Equation Modelling Based on the variables and hypotheses described, a structural equation model can be developed (Figure 2). The model is explorative in the sense that indicators were evaluated whether they fit into the model or not (Hair et al., 2013). The model is evaluated in chapter five. Process support H1 a-c by smartphones and tablets H3 a+b Significance of mobile ICT Number of Value of H5 mobile devices smartphones used and tablets Proportion of mobile work H4 a+b Information H2 a-c used when mobile Figure 2: Structural equation model (own diagram) 4 Data Collection and Sample The study focuses on micro-enterprises and small and medium-sized Swiss companies with 1 to 250 full time equivalents (FTEs) in selected areas of the economic sectors two (manufacturing industry) and three (service industries) (Bundesamt für Statistik, 2008). The population of the selected economic sectors covers 266'715 companies. A random sample of 6'000 companies was chosen from this universal set, based on sector and company size (see Figure 3). 294 Michael H. Quade, Uwe Leimstoll Industries Health and social Manufacturing of services, 94, 10% goods, 129, 13% Freelance, technical and other economic services, 93, 9% Construction, 93, 10% Real estate and housing, 82, 8% Trade, Maintenance and repair of motor Provision of vehicles, 130, 13% financial and insurance services, 93, 9% Transportation and Information and storage, 89, 9% communications, IT, Accommodation, 93, 10% catering, 88, 9% Figure 3: Distribution of the industries in the sample (own diagram) Computer-aided telephone interviews (CATI) were used to collect the data. The survey was aimed at members of senior management. In total, 984 companies were interviewed from March to May 2013. In the industries "Manufacturing of goods" and "Trade, Maintenance and repair of motor vehicles", more companies were interviewed than in other industries. These two indus- tries are the largest in the universal set. In addition, more small companies were interviewed than larger ones. Of the 984 companies 40.55% belong to the category 1-9 FTEs. The data col- lection was sponsored by four Swiss companies: ABACUS Research, BusPro, Sunrise and Swisscom. 5 Research Methodology and Evaluation The Partial Least Squares (PLS) approach is chosen in order to evaluate and estimate the struc- tural equation model (SEM) (see Chapter 3.6). PLS is a widely-used approach for research sit- uations where the theory behind an SEM is still evolving (Wold, 1980). Considering the novelty of the present topic and the indicator scales used for the constructs, the choice of the PLS-SEM approach is justified. SmartPLS 2.0 M3 (Ringle, Wende, & Will, 2005) was used for the calcu- lations. The first step is to specify the indicators and scales for the measurement model based on the given conceptualization. The second and third steps are the evaluation of the model and the interpretation of the results. 5.1 Specification of the Measurement Model To measure each construct in the model, appropriate indicators and scales have to be specified. To ensure the content validity of the measured constructs, widely accepted scales are used (Chin, 2010; Hair et al., 2013). The construct measurement mode is based on decision criteria found in the literature (Diamantopoulos & Siguaw, 2006). Therefore, two constructs are meas- ured reflectively and four formatively. One construct is measured as a single-item (Diaman- topoulos et al., 2012). Table 1 shows the indicators and the scales used to collect data and to evaluate the constructs in the model. 295 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs Construct Measurement Indicators used in the study Source Significance of formative How significant are the following aspects in your company? (Adapted from characteristics 1. Permanent online access to information and communication channels. Basole, 2005; of mobile ICT 2. Permanent carrying of information or communication devices. Watson et al., (SIG) 3. Constant reachability of persons (e.g. employees, customers or suppliers). 2002) 4. Localization of the current location of persons (e.g. employees, customers or suppliers) 5. Availability of continuously updated information (e.g. prices, rates, stocks) (1= insignificant, 2=rather insignificant, 3=rather significant, 4= significant) Portion of reflective Please estimate the portion of time worked by your employees not on company’s site, (Adapted from mobile work (single-item) which means mobile work (estimate as a percentage of time spent in mobile work). Zhu et al., (WOR) 2006) Information formative Please tell us if the following information is used during mobile work. (Adapted from used during 1. Contact information (e.g. addresses, phone numbers, locations) Keller, mobile work 2. Data on plants, buildings, infrastructure or equipment (e.g. plans, schematics, Nüttgens, & (INF) maintenance history) Scheer, 1992) 3. Information on the service billing (e.g. reports, time sheets) 4. Catalogs or manuals (e.g. for reference) 5. Checklists (e.g. for documentation of the work) 6. Information about employees (e.g. operational plans, personnel files) 7. Other information in the form of text files, spreadsheets or presentations 8. Information about customers or suppliers (e.g., contract, order data, files, reservations, invoices) (0=not used, 1=used) Process formative Is the following field of activity supported with smartphones or tablets in your company? (Adapted from support by 1. Financial accounting Zhu, Kraemer, smartphones 2. Human resource / payroll & Xu, 2006) and tablets 3. Controlling, reporting, business intelligence (SUP) 4. Purchasing, supplier relationship management 5. Logistics, warehousing 6. Production of goods and services, production data acquisition 7. Order processing, project management 8. Marketing and sales 9. Customer Service, maintenance 10. Data management, file storage, archiving (0=not supported by smartphones and tablets, 1= supported by smartphones and tablets) Number of formative What types of smartphones and tablets are used in the company? Please give us the (Adapted from smartphones approximate number of devices. An estimate is sufficient. Zhu & and tablets 1. iPhone with Apple iOS, smartphone with Google Android, smartphone with Microsoft Kraemer, used (SAT) Windows 2005; Wang, 2. iPad with Apple iOS, tablet with Google Android, tablet with Microsoft Windows Wang, & Yang, (numbers are divided by the collected exact number of FTEs, percent of FTEs who use 2010) a smartphone or tablet) Value of reflective Do you agree with the following statements? (Adapted from smartphones 1. (In our work), we cannot work well without smartphones or tablets. Gattiker & and tablets 2. Smartphones or tablets reduce the mobile data capture on paper. Goodhue, (VAL) 3. Smartphones or tablets increase the productivity of our employees. 2005) 4. Smartphones or tablets increase the flexibility and responsiveness of our staff. 5. Smartphones or tablets allow us to design new business processes. (1=do not agree, 2=tend not to agree, 3=tend to agree, 4=fully agree) Table 1: Constructs and indicators 5.2 Evaluating the Model After the specification of the model and the collection of data, the data is examined and the con- structs and path of the model are evaluated. The SEM is evaluated by the recommended steps described by Hair et al. (2013). The data examination revealed that 81 cases must be removed from the sample (8.2%) because of missing values. To ensure that cases of a particular category were not removed systematical- ly, the cases were analysed: Those removed from the sample were in proportion and spread evenly across all industries and company sizes. Next, the convergent and discriminant validity was assessed in order to ensure the validity of the reflective constructs. Convergent validity is ensured by the calculated results (for details see Table A1). To ensure that the measured constructs are sufficiently different from each other, appropriate levels of discriminant validity are needed. This requirement is fulfilled for the re- flective constructs (for details see Table A2). 296 Michael H. Quade, Uwe Leimstoll To evaluate a formative construct, each set of indicators assigned to a formative construct must be assessed for collinearity issues. Issues are present if any indicator has a variance inflation factor (VIF) of 5 or higher. All indicators are below 5. If there are no issues with the VIF values, each single indicator is assessed regarding its significance and the relevance it has to the as- signed construct. The relevance is assessed with t values and outer weights (OW) values of each construct (for details see Table A3). According to the outer weights, few indicators are not sig- nificant or relevant to the assigned formative construct. Removing these indicators from the model has to be considered. However, if the theory-based conceptualization of the construct supports retaining the indicator, it should be kept in the model. Therefore, no indicator was re- moved. As all measured constructs have satisfactory quality levels, the structural model can be evaluat- ed. With the bootstrapping algorithm, the significance of the hypothetical relationships (path coefficients) is estimated. The path coefficients and significance levels are shown in Figure 4. All relationships are significant and the hypotheses are all supported. SUP 0.254*** R2=0.085 0.293*** 0.128*** SIG 0.136*** 0.114*** VAL SAT 0.229*** R2=0.355 R2=0.147 0.373*** Q2=0.197 WOR 0.177*** 0.144*** 0.302*** INF 0.147*** R2=0.176 (ns = not significant, * = 10%, ** = 5%, *** = 1%) Figure 4: Significance and relevance of relationships The model is also analysed on mediating effects of indirect paths between constructs. Within the six mediation paths in the model only one path has a partial mediating effect. The path SIG  INF  SAT has a variance accounted for of 22% (for details see Table A4). The effect can be considered as weak and can be explained as follows: The significance of the characteristics of mobile ICT is valued higher by the respondents if they have to use many types of information when working mobile. This leads to a higher amount of used mobile devices. The last step is to assess the predictive power of the model with the coefficient of determination (R2) and the predictive relevance with the Stone-Geisser’s Q2 value. The endogenous latent variable VAL (the only variable in the model which is only endogenous) has a R2 of 35.5%, a moderate level. The other endogenous constructs have a rather weak or very weak level (Figure 4). The predictive relevance Q2 is different from zero; therefore the model has a predictive rele- vance. 5.3 Interpretation and Refinement of the Evaluation From the evaluation it can be interpreted, that SIG affects INF and SUP relatively more than the construct WOR. On the other hand, WOR affects SAT slightly more than the other constructs. INF and SUP affect VAL to a similar degree. SAT affects VAL less. Based on the evaluation all paths are significant, therefore all hypotheses can be confirmed. The predictive power of the model is – according to the coefficients of determination – rather weak than medium. The reason for this weakness is the heterogeneity in the sample. In the de- 297 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs scriptive analysis of the sample it can be seen that there are significant differences between in- dustries and company sizes and that different types of information are used in mobile work in individual industries. Findings per company size: There are only slight shifts compared to the full sample. The most distinctive shift is shown in the path coefficients regarding VAL. In micro-enterprises (1-9 FTEs) the construct SAT affects VAL as much as the construct INF in SMEs (10-250 FTE). Each category shows a small increase in the R2 and Q2 values. This means that a small part of the heterogeneity is explained by the company size. But there are no major differences regard- ing the indicator outer weights on the formative constructs. Findings per industry sector: The evaluation of individual industries shows greater differences in the model. Depending on the sector, the structural equation modelling leads to significantly higher or lower values compared to the full sample. Some industries show higher R2 values in most endogenous constructs, particularly in the construct VAL. Some industries show lower values of R2 in VAL but much higher values in the other endogenous constructs. Table 5 sum- marizes the findings. R2 Q2 Industry INF SUP SAT VAL Manufacturing of goods 0.255 0.146 0.150 0.457 0.259 Construction 0.325 0.098 0.209 0.512 0.262 Trade, maintenance and repair of motor 0.331 0.096 0.212 0.356 0.214 vehicles Transportation and storage 0.300 0.102 0.321 0.268 0.161 Accommodation, catering 0.253 0.175 0.188 0.515 0.307 Information and communications, IT 0.203 0.212 0.103 0.487 0.222 Provision of financial and insurance services 0.191 0.264 0.298 0.304 0.144 Real estate and housing 0.175 0.285 0.431 0.390 0.130 Freelance, technical and other economic 0.122 0.178 0.221 0.323 0.175 services Health and social services 0.270 0.250 0.306 0.232 0.138 Table 5: Coefficient of determination and predictive relevance for each industry Additionally, the evaluation of the industries revealed that different indicators have more or fewer outer weights on the formative constructs. There are different characteristics of mobile ICT, types of information or process support in some industries with more weights than those in the full sample. Table 6 shows the indicators that have an outer weight >0.4 on a formative con- struct. If a cell is empty, there are no differences compared to the full sample. 298 Michael H. Quade, Uwe Leimstoll Constructs and Indicator No. Industry SIG INF SUP SAT Full sample 2 1 7, 9 1, 2 Manufacturing of goods 1, 2 1, 3 1 Construction 5 2 Trade, maintenance and repair of motor vehicles 3, 9 Transportation and storage 1, 2, 3 5 4, 10 Accommodation and catering 1, 2 1, 6 6, 9 Information and communications, information technology 5 10 2 Provision of financial and insurance services 1, 4, 7 8 1 Real estate and housing 5 7, 10 Freelance, technical and other economic services 1 Health and social services 10 Table 6: Indicators with outer weights >0.4 for each industry 6 Discussion Other studies also show that smartphones and tablets are most commonly used for e-mail and calendar functions (Causse, 2012; Pelino, 2012). Therefore, it is not surprising that information such as "Contact information (e.g. addresses, phone numbers, locations)" explains a large part of the latent variable information used when mobile (INF). Regarding process support, the mod- el makes sense in the respect that process areas such as "Order processing, project management" and "Customer service, maintenance" affect the process support (SUP).This result is in line with the descriptive results from Scherz (2008). Mobile work on-site at the customers’ locations of- ten is related to these two process areas (Buser & Poschet, 2002). Major software providers have not yet responded to this fact. The core modules of standard software packages often sup- port fields of activity such as financial accounting, controlling and human resources. These are fields of activities which have less influence or negative influence on SUP in the model present- ed. Therefore, software providers will only have success with solutions for smartphones and tablets when these solutions satisfy the need for information or process support in areas such as project management or customer service. Differences between the industries are shown through a separate analysis of the data. In some industries, the model calculates higher values. Within some other industries, the separate eval- uation of the model produces lower values. It seems that there is still an unobserved heterogene- ity in the sample and subsamples. Therefore, the sample heterogeneity should be checked using the FIMIX function of SmartPLS (Hahn, 2002). The calculation of the segmentation is based only on the given structural equation model. Some evaluations have already been carried out with this "a posteriori" approach (Sarstedt & Ringle, 2010). 7 Conclusions Based on the evaluation with the full dataset, a clear difference is shown in the impact of "Sig- nificance of characteristics of mobile ICT" (SIG) and the "Proportion of mobile work" (WOR) on “Information used during mobile work” (INF), “Process support by smartphones and tablets” (SUP) and “Number of smartphones and tablets used” (SAT): SIG has a medium effect on INF and a weak effect on SUP. WOR has only a weak effect or even no effect on SUP, SAT and 299 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs INF. This means that the share of mobile work processes has a smaller impact than the charac- teristics of mobile ICT or qualitative aspects of the information used in mobile work processes. This result corresponds to the results of Gebauer (2008). The answer to the first research ques- tion (RQ1) is therefore: The significance of the characteristics of mobile ICT has an impact on the usage of smartphones and tablets in SMEs, particularly when mobile work involves the us- age of information. The model identifies three main effects on the perceived value of smartphones and tablets (VAL): the different types of information used when mobile (INF), the type and degree of pro- cess support by smartphones and tablets (SUP), and the number of smartphones and tablets used in the company (SAT). These constructs have more or less the same effects on the perceived value of smartphones and tablets (VAL) in SMEs. This means that there is no single construct that determines the value SMEs see in the use of smartphones and tablets. The answer to the second research question (RQ2) is therefore: The effect on the perceived value of mobile ICT is not only a result of the amount of mobile devices used by a SME. It depends more on the inte- gration of mobile computing into business processes and on the variety of information used. The descriptive analysis of the sample revealed that in most cases companies support only few of their business processes with smartphones and tablets. Nevertheless, most companies use unstructured information for their mobile work. Storing and accessing document-based infor- mation on a smartphone or tablet is easy and quickly done. In contrast, implementing process support on smartphones and tablets is much more challenging, because it needs specialist appli- cations and an integration of the smart devices into the work processes. To fully explore the improvements in productivity and flexibility, processes often have to be redesigned according to the potential advantages provided by those devices. Nevertheless, the model is limited and only accounts for a small part of the reality, and it is in- complete. There could also be other exogenous constructs that affect the value of smartphones and tablets, e.g. the information systems integration of mobile apps, or the characteristics of tasks that smartphones and tablets are used for, or the specific types of mobile work. Future research could examine additional constructs or moderating factors that explain the dif- ferences between the industries. References Anckar, B., & D’Incau, D. (2003). Value creation in mobile commerce: Findings from a con- sumer survey. Journal of Information Technology Theory and Application (JITTA), 4, 43–64. Basole, R. C. (2004). The value and impact of mobile information and communication technol- ogies. In Proceedings of the IFAC Symposium on Analysis, Modeling & Evaluation of Human-Machine Systems (pp. 1–7). Atlanta, GA. Basole, R. C. (2005). Mobilizing the enterprise: A conceptual model of transformational value and enterprise readiness. In 26th ASEM National Conference Proceedings (pp. 364– 371). Basole, R. C. (2007). The emergence of the mobile enterprise: A value-driven perspective. In ICMB 2007. International Conference on the Management of Mobile Business (pp. 41– 48). Benbasat, I., & Barki, H. (2007). Quo vadis, TAM? Journal of the Association for Information Systems, 8, 211–218. 300 Michael H. Quade, Uwe Leimstoll Bundesamt für Statistik. (2008). NOGA 2008 - Allgemeine Systematik der Wirtschaftszweige. Neuchâtel: Bundesamt für Statistik BfS. Buser, M., & Poschet, L. (2002). Mobile Arbeit in vielfältiger Ausprägung. In L. Rey (Ed.), Mobile Arbeit in der Schweiz. Reihe“ Mensch-Technik-Organisation” (Vol. 28, pp. 77– 123). Causse, A. (2012). Smartphones in Business. Retrieved June 2, 2014, from http://blog.idate.fr/smartphones-in-business/ Chin, W. W. (2010). How to write up and report PLS analyses. In V. E. Vinzi, W. W. Chin, J. Henseler, & H. Wang (Eds.), Handbook of partial least squares (pp. 655–690). Heidel- berg Dordrecht London New York: Springer. Chong, A. Y.-L., Chan, F. T. S., & Ooi, K.-B. (2012). Predicting consumer decisions to adopt mobile commerce: Cross country empirical examination between China and Malaysia. Decision Support Systems, 53, 34–43. Diamantopoulos, A., Sarstedt, M., Fuchs, C., Wilczynski, P., & Kaiser, S. (2012). Guidelines for choosing between multi-item and single-item scales for construct measurement: a predictive validity perspective. Journal of the Academy of Marketing Science, 40, 434– 449. Diamantopoulos, A., & Siguaw, J. A. (2006). Formative versus reflective indicators in organiza- tional measure development: a comparison and empirical illustration. British Journal of Management, 17, 263–282. Gareis, K. (2003). Home-Based vs. Mobile Telework: The Interrelationship between Different Types of Telework. In P. B. Rapp & P. Jackson (Eds.), Organisation and Work Beyond 2000 (pp. 171–185). Physica-Verlag HD. Gattiker, T. F., & Goodhue, D. L. (2005). What Happens after ERP Implementation: Under- standing the Impact of Interdependence and Differentiation on Plant-Level Outcomes. MIS Quarterly, 29, 559–585. Gebauer, J. (2008). User requirements of mobile technology: A summary of research results. Information, Knowledge, Systems Management, 7, 101–119. Goodhue, D. L., & Thompson, R. L. (1995). Task-Technology Fit and Individual Performance. MIS Quarterly, 19, 213–236. Hahn, C. H. (2002). Segmentspezifische Kundenzufriedenheitsanalyse. Wiesbaden: Deutscher Universitätsverlag. Hair, J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2013). A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). Los Angeles: SAGE. Junglas, I. A., & Watson, R. T. (2003). U-Commerce: A conceptual extension of E-and M- commerce. In International Conference on Information Systems, Seattle, WA. Keller, G., Nüttgens, M., & Scheer, A.-W. (1992). Semantische Prozessmodellierung auf der Grundlage “Ereignisgesteuerter Prozessketten (EPK).” (A.-W. Scheer, Ed.). Saar- brücken: Instituts für Wirtschaftsinformatik. Lehner, F. (2003). Mobile und drahtlose Informationssysteme: Technologien, Anwendungen, Märkte. Berlin, Heidelberg: Springer-Verlag. Leimstoll, U., & Quade, M. H. (2011). Special Business-Software-Studie 2011. Netzwoche, 17,1–8. 301 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs Malone, T. W., & Crowston, K. (1994). The interdisciplinary study of coordination. ACM Com- puting Surveys (CSUR), 26, 87–119. Meier, A., & Stormer, H. (2009). eBusiness & eCommerce managing the digital value chain. Berlin Heidelberg: Springer-Verlag. Möhlenbruch, D., & Schmieder, U.-M. (2001). Gestaltungsmöglichkeiten und Entwicklung- spotenziale des Mobile Marketing. HMD-Praxis Der Wirtschaftsinformatik, 15–26. Pelino, M. (2012). The Expanding Role of Mobility in the Workplace - IT Pro White Paper Li- brary. Cambridge: Forrester Consulting. Picoto, W. N., Bélanger, F., & Palma-dos-Reis, A. (2014). An organizational perspective on m- business: usage factors and value determination. European Journal of Information Sys- tems, 23, 571–592. Porter, M. E. (1985). Competitive Advantage: Creating and Sustaining Superior Performance. New York, London: Free Press ; Collier Macmillan. Ringle, C., Wende, S., & Will, A. (2005). SmartPLS 2.0M3. Hamburg: SmartPLS. Rogers, E. M. (1983). Diffusion of Innovations (Third). New York, London: The Free Press. Sarstedt, M., & Ringle, C. M. (2010). Treating unobserved heterogeneity in PLS path modeling: a comparison of FIMIX-PLS with different data analysis strategies. Journal of Applied Statistics, 37, 1299–1318. Scherz, M. (2008). Mobile Business - Schaffung eines Bewusstseins für mobile Potenziale im Geschäftsprozesskontext. Technische Universität Berlin, Berlin. Schiller, J., & Voisard, A. (2004). Location-based Services. San Francisco: Morgan Kaufmann Publishers. Tarasewich, P., Nickerson, R. C., & Warkentin, M. (2002). Issues in Mobile E-Commerce. Communications of the Association for Information Systems, 8, 41–65. Varian, H. R. (2010). Intermediate Microeconomics: A Modern Approach (8th edition. Interna- tional Student edition.). New York: Norton & Company. Venkatesh, V., & Davis, F. D. (2000). A theoretical extension of the technology acceptance model: four longitudinal field studies. Management Science, 46, 186–204. Venkatesh, V., Thong, J. Y. L., & Xu, X. (2012). Consumer Acceptance and Use of Information Technology: Extending the Unified Theory of Acceptance and Use of Technology (SSRN Scholarly Paper No. ID 2002388). Rochester, NY: Social Science Research Network. Walter, T., & Sammer, T. (2012). Mobile Business Solutions Studie 2012: Best Practices, Bar- rieren und Chancen von Enterprise Mobility im deutschsprachigen Raum. Dresden: T- Systems Multimedia Solutions GmbH. Wang, Y.-M., Wang, Y.-S., & Yang, Y.-F. (2010). Understanding the determinants of RFID adoption in the manufacturing industry. Technological Forecasting and Social Change, 77, 803–815. Watson, R. T., Pitt, L. F., Berthon, P., & Zinkhan, G. M. (2002). U-commerce: expanding the universe of marketing. Journal of the Academy of Marketing Science, 30, 333–347. Wernerfelt, B. (1984). A Resource-based View of the Firm. Strategic Management Journal, 5, 171–180. 302 Michael H. Quade, Uwe Leimstoll Wold, H. (1980). Model construction and evaluation when theoretical knowledge is scarce. In Evaluation of econometric models (pp. 47–74). Academic Press. Yang, K. (2010). Determinants of US consumer mobile shopping services adoption: implica- tions for designing mobile shopping services. Journal of Consumer Marketing, 27, 262– 270. Zhu, K., Dong, S., Xu, S. X., & Kraemer, K. L. (2006). Innovation diffusion in global contexts: determinants of post-adoption digital transformation of European companies. European Journal of Information Systems, 15, 601–616. Zhu, K., & Kraemer, K. L. (2005). Post-Adoption Variations in Usage and Value of E-Business by Organizations: Cross-Country Evidence from the Retail Industry. Information Sys- tems Research, 16, 61–84. Zhu, K., Kraemer, K. L., & Xu, S. (2006). The Process of Innovation Assimilation by Firms in Different Countries: A Technology Diffusion Perspective on E-Business. Management Science, 52, 1557–1576. 303 Mobile Business with Smartphones and Tablets: Effects of Mobile Devices in SMEs Appendix AVE Composite Cronbach's Reliability Alpha WOR 1 1 1 VAL 0.562 0.864 0.802 Table A1: Convergent validity Construct SIG WOR INF SUP SAT VAL SIG formative WOR 0.137 single-item INF 0.393 0.198 formative SUP 0.270 0.148 0.315 formative SAT 0.252 0.243 0.273 0.236 formative VAL 0.460 0.194 0.457 0.442 0.381 0.750 Table A2: Discriminant validity 304 Michael H. Quade, Uwe Leimstoll Construct Ind. No. OW t Value Sig. Level OL t Value Sig. Level 1 0.192 2.363 ** 0.577 9.038 *** 2 0.752 10.696 *** 0.906 26.424 *** SIG 3 0.101 1.206 ns 0.471 6.348 *** 4 -0.243 2.856 *** 0.018 0.199 ns 5 0.285 3.442 *** 0.580 8.667 *** 1 0.570 8.632 *** 0.850 23.744 *** 2 0.168 2.483 ** 0.614 12.717 *** 3 0.213 2.993 *** 0.609 12.514 *** 4 0.184 2.675 *** 0.608 11.823 *** INF 5 0.127 1.760 * 0.600 11.788 *** 6 -0.013 0.187 ns 0.422 7.279 *** 7 0.071 0.886 ns 0.621 11.950 *** 8 0.091 1.186 ns 0.633 12.883 *** 1 -0.389 3.979 *** 0.090 1.110 ns 2 -0.308 2.977 *** 0.123 1.536 ns 3 0.098 1.219 ns 0.387 5.580 *** 4 0.182 2.332 ** 0.490 7.757 *** 5 0.025 0.318 ns 0.419 6.582 *** SUP 6 0.246 3.294 *** 0.511 8.003 *** 7 0.469 5.405 *** 0.757 16.225 *** 8 0.212 2.786 *** 0.596 10.021 *** 9 0.334 4.347 *** 0.650 11.584 *** 10 0.216 2.576 *** 0.515 7.798 *** 1 0.704 9.382 *** 0.927 27.657 *** SAT 2 0.437 4.965 *** 0.796 16.032 *** (ns = not significant, * = 10%, ** = 5%, *** = 1%) Table A3: Significance of outer weights / loadings indirect indirect indirect path direct path path 1 path 2 VAF SIG  SUP  SAT 0.159 0.254 0.128 17% SIG  INF  SAT 0.195 0.373 0.144 22% WOR  SUP  SAT 0.191 0.114 0.128 7% WOR  INF  SAT 0.193 0.147 0.144 10% SUP  SAT  VAL 0.329 0.128 0.229 8% INF  SAT  VAL 0.352 0.144 0.229 9% Table A4: Mediator analysis 305 BACK 28th Bled eConference #eWellBeing June 7 - 105, 2015; Bled, Slovenia Women and ICT: exploring obstacles and enablers of a possible career Ruxanda Berghi Bocconi University, Italy ruxanda.berghi@studbocconi.it Paola Bielli Bocconi University, Italy paola.bielli@unibocconi.it Abstract The ICT industry is a key contributor to the EU’s economy. Unfortunately, women’s presence is low overall and it decreases as they climb the corporate ladder. Underrepresentation of women in ICT is a research area that has received attention mostly in U.S.A., UK and some European countries. This phenomenon, termed “IT gender gap”, has not received much attention in Italy, yet. Therefore, the purpose of this study is doing an “initial” research and understanding the characteristics of the IT workplace culture in Italy. Based on the international research, a framework and a questionnaire have been developed. To test the questionnaire, a first research sample (without any statistical relevance compared to the Italian context) has been created and the potential respondents were contacted via email. Data analysis discusses the workplace environmental factors that hinder and support the career development of women in ICT in this country. Understanding the limitations of this research project has given rise to some open points that deserve being analysed and further explored. Keywords: women & ICT, eSociety, gender issues in ICT 1 Introduction The debate about a possible career for women in ICT is not new, as many studies have addressed it (Michie, Nelson, 2006), but it has not reached a satisfactory conclusion, yet. Since long both institutions (EU commission, local governments, ICT professionals’ associations) and researchers have emphasized the limited presence of women in ICT jobs. 306 , For instance, in the survey “Women Active in the ICT sector” (2013) the European Commission found that getting more girls interested in a digital career, and getting more women into digital jobs would benefit the digital industry, women themselves, and Europe's economy. According to the study, there are now too few women working in the ICT sector. While women represent 59% of all tertiary graduates and 45.7% of total employees in Europe (European Commission, 2013), in the ICT sector though, women represent around 33% (Eurostat, 2011) of total graduates in science and technology and around 32% of employees of the ICT sector (Eurostat, 2012). According to the Labour Force Survey done by Eurostat in 2012, the first problem is that girls do not choose ICT-related studies, whereas the second is that they choose ICT careers to an even lesser extent. From all women in the labour market only 2% of them work in the ICT sector, compared to 3.6% of men. Additionally, one of the most important, and worrying, phenomena is the so-called “leaky pipeline”, defined as women abandoning the sector mid-career. The leaky pipeline is a known fact (Gras-Velasquez et.al, 2009; Griffiths and Moore, 2010; Hunt, 2012) common both to Europe and to the US. While 20% of women aged 30 with ICT- related degrees work in the sector, only 9% of women above 45 years of age do so. Studies (Quesenberry, Trauth and Morgan, 2006) attribute this situation largely to maternity. Underrepresentation of women in ICT is a research area that has received attention mostly in U.S.A., UK and some European countries. This phenomenon, termed “IT gender gap”, however, has not been explored in Italy, yet. Therefore, the purpose of this study is doing an “initial” research, understanding the characteristics of the IT workplace culture in Italy and investigating enablers and obstacles to an ICT career for women and the challenges in remaining in the industry. Based on international literature on gender gap we developed a research framework and a questionnaire, aimed at identifying possible barriers to an ICT career for women in Italy, taking into consideration local specificities. Framework and questionnaire have been tested within a small sample of 100 ICT female professionals. Data analysis shows that there are some cultural/environmental factors that hinder and others that play a positive role in the career development of ICT women in this country. Nonetheless, this research must be further extended as it is subject to some limitations. In the conclusions, the shortcomings are discussed as well as the updated version of the research kit. 2 Literature background The relation between women and ICT has been studied from a theoretical point of view since long. Three main theoretical approaches have been used to understand and explain the IT gender gap: 1) the essentialist theory; 2) the social construction theory; and 3) the individual differences theory of gender and IT. 307 Women and ICT: exploring obstacles and enablers of a possible career The essentialist theory is based on the assertion of fixed, unified, and opposed female and male natures (Trauth, 2002; Trauth et al., 2004; Wajcman, 1991). With regard to the IT gender gap research, the essentialist theory uses biological differences between men and women to explain differences in their relationship to technology. The social construction theory tends to reflect an interpretative epistemology as a lens to investigate the IT gender gap phenomenon. In this sense, gender is broadly viewed as two separate groups of men and women who are affected by different sets of sociological influences (e.g. family, school, colleagues, social networks, etc.). The individual differences theory of gender and IT rejects essentialism and offers refinement of various underexplored areas of the social construction theory. The theory examines the individual variations across genders as a result of both personal characteristics and environmental influences in order to understand the participation of women in the IT profession. Under this theory, several factors have been identified as affecting women retention in the IT workforce: cultural fit, mentors, role models, expectation gaps, role ambiguity, role conflict, career satisfaction and organizational commitment (Bartol, Williamson and Langa, 2006; Riemenschneider et.al, 2006; Tapia and Kvasny, 2004). Igbaria and Greenhaus (1992) reported that organizational commitment and job satisfaction are the most direct influences on turnover intentions among IT professionals. Another problem is that women are under-represented in managerial and decision- making positions (1) few of the women entering an ICT organisation, climb up the internal career ladder completely. This phenomenon has been labelled “glass ceiling” and it is clearly identified by women, particularly those with longer careers (63% of under 30s, 71% of 31–44s and 77% of over 45s acknowledge the barriers, according to Institute of Leadership and Management 2011). Lemons and Parzinger (2001) found that poor advancement opportunities for women in IT were due to limited gender socialization and corporate culture issues. Eventually it is necessary to understand the characteristics of the IT workplace culture (Denison, 1996; McLean, 2003), and in particular, workplace environmental factors that hinder and support the career development of women in IT. The IT computing culture has been described as having unique characteristics: largely white, male- dominated, competitive, individualistic, and antisocial (Trauth, Quesenberry and Yeo, 2008; Lemons and Parzinger, 2001; Ahuja, et.al, 2007). This culture has the potential to exclude women and minorities if they do not conform (von Hellens, Nielsen and Trauth, 2001; Roldan, Soe and Yakura, 2004; Trauth, Quesenberry and Yeo, 2008). Cultural expectations regarding women in the IS workforce have been found to often cause their talents to be overlooked and lead to increased level of work-family conflict (Aaltio and Huang, 2007). Such perceptions have been cited as a possible reason why women have begun to delay having children (Crump, Logan and Mcllroy, 2007). 1 worldwide, in 2010 4% of companies’ CEOs in the IT & Telecom sectors are female (Corporate Gender Gap Report, 2010). 308 , Unfortunately women are often confronted with a “chilly” or even “hostile” cultural climate in many IT work organizations (Margolis and Fisher, 2002; Roldan, Soe, and Yakura, 2004) and an occupational culture that seems to privilege male workers and their competencies, regardless of the skills possessed by women (Woodfield, 2002; Bagilhole, Powell, Barnard, and Dainty, 2007). In synthesis, literature suggests that a career in IT for a woman is a combination of individual decisions and a broad set of environmental factors. 3 Research framework As the focus of this research is Italy, where the IT gender gap phenomenon has not been studied in depth, we had to develop a research framework taking into account Italian specificities (cultural and social factors), which could explain the low presence of women in ICT in this country. A quick overview of women’s situation in Italy is necessary to understand the model. In 2012 the women’s employment rate is 48% (2), 12 percentage points below the EU- 27 average and women represent only 4% of the board of directors, whereas the European average is 11% (European Commission, 2012). Italy, in terms of gender equality, ranks 80 - out of 135 countries - and 126 in terms of wage gap, according to the Global Gender Gap report 2012 (World Economic Forum). This data reveals that there is a lot of work to be done at cultural and social level to overcome gender diversity in Italy. Besides, the official wage gap between men and women is 6.7% (Eurostat, 2012) and it depends on the sector and on the individual characteristics: age, presence of kids, and level of education. Italian women spend on family duties more time than any other European woman (5h20’). The family load is heavier in Italy (74%) than elsewhere and time devoted to domestic work represents the most evident element of gender difference in the use of daily time: women dedicate to domestic work on average 4h30’, while men only 1h28’ (Istat, 2008). Moreover, the Italian culture is considered to have a high masculinity degree (Hofstede, 1980) which might explain a lower propensity of women to enter and/or progress in ICT, a “man’s world”. According to a study (3) carried in 5 European countries (Italy, Poland, UK, Netherlands and France) in Italy there is no substantial difference in ICT knowledge and aptitudes between male and female students. Italian female students are competent in ICT and enjoy it but they do not intend to study ICT at tertiary level or pursue ICT career paths. Looking at these facts, maternity results as a key cause to a limited career, even if the birth rate for Italian women is very low (1.3 vs 1.8 Norway and Sweden, 1.9 France and 2 (Istat 2013) the official data varies significantly within Italian regions 3 Women and ICT: why are girls still not attracted to ICT studies and careers? (Gras-Velasquez, Joyce and Debry, 2009) 309 Women and ICT: exploring obstacles and enablers of a possible career Belgium) (World Economic Forum, 2012). The welfare regime in Italy relies largely on the family, and in particular, on women, that act as the main provider of care for children, sick and the elderly relatives. This has been labelled the Mediterranean welfare regime (Ferrera, 1996; Gosta Esping-Andersen, 1999). Women suffer of the “double burden” syndrome : the combination of work and domestic responsibilities, which is difficult to reconcile with another barrier: the “anytime, anywhere” performance barrier typical of the ICT world. Nevertheless, the most difficult part is the cultural prejudice: the idea that the conciliation of personal, social and affective life is something that only women have to deal with. Based on the “World Values Survey”, done between 2005 and 2008, Italy is characterized by having prejudices against the female presence in the economy and society. Social pressure does matter a lot: if there are few jobs, man has priority over woman; it is more important for a man than for a woman to go to university and to get a job soon, etc. The more these ideas are diffused, the less are women included in the job market. The conditions for working in the ICT sector seem very unattractive to women, as the industry looks extremely harsh and competitive: long working hours, no spare time, no holidays and individualized labor relations. Work life balance is therefore a key issue here. In addition, the ICT industry experiences a rapid obsolescence of know-how: software and technology constantly change and any specialist has to keep the pace with these changes to remain in the market place. To sum up, the topic of women in IT is a “highly complex cultural issue” (Svinth, 2006) with many faces, and it is affected by a range of subtle influences including the environmental context, gender, race, class, career decisions, work life balance issues, social networks, and organizational factors (Trauth, 2000). According to Coga and Chen (2007) women’s career development is more complex (than men’s), because they face a number of internal and external barriers that complicate and limit their career choices and advancement. Cultural expectations regarding women in the IS workforce have been found to often cause their talents to be overlooked and lead to increased levels of work family conflict. Families continue to be liabilities to women’s career development in organizations (O’Neil, Hopkins and Bilimoria, 2008). Unfortunately, women are often confronted with a “hostile” cultural climate in many IT work organizations (Margolis and Fischer, 2002; Roldan, Soe and Yakura 2004), an occupational culture that seems to privilege male workers and their competencies, regardless of the skills possessed by women. One of the top barriers that women face is the lack of role models. Studies reveal a persistent view that women “fit” better with the softer side of IS (Kuhn, Rayman, 2007) which may limit their career opportunities. All these issues influence the career options of women working in the ICT industry (see Figure 1). 310 , Figure 1: Internal and external barriers that affect women’s career development To understand the specificities of the Italian context, we have focused on the dimensions that can be strongly affected by local culture and social pressure (as seen at the beginning of this paragraph) and we propose a framework for Italy where four main factors influence the career progression in ICT (see Figure 2). Family obligations (children or parents to take care of, house work) are a great barrier for women’s career progression. Moreover, there is also the IT male dominated culture that hinders women in developing their career. We reckon that also personal choices might be a factor that influences women’s career progression. Italy is characterized by having prejudices against women presence in the economy and society. Apart from the prejudices there are also many stereotypes about the IT world and women’s ability to deal with it. Figure 2: Research framework for Italy A questionnaire has been developed to test the framework. It is composed of 33 questions divided in different parts: understanding the respondent profile (educational background, work context, marital status, family obligations), self-assessment of factors promoting and inhibiting the career (agreement or disagreement in a 10 points Likert scale), analysis of their vision at the beginning of their career and now. 311 Women and ICT: exploring obstacles and enablers of a possible career We decided to test the quality of the questionnaire by organising a data collection in a small sample of ICT women specialists. The questionnaire was sent to ICT organizations, women associations, direct contacts, without aiming at any statistical relevance. In general, the respondents were very enthusiastic about this study and asked the permission to send the questionnaire to their female friends working in other IT companies. The collection of data lasted for 3.5 months and it has reached 100 answers in total. 4 Pilot testing As explained in the previous chapter, the empirical step aimed at testing the questionnaire. The testing sample is composed of 100 women working in the ICT industry. Data collection was done via personal contacts or word of mouth. Within the sample, we can identify a cluster of 36 respondents belonging to two big ICT organizations. The remaining 64 work in small companies dealing with the ICT industry. Out of 100 respondents, 30 work in the IT department, and the remaining 70 in departments like Marketing, Sales or Consulting. One preliminary consideration is needed: based on the composition of our sample we can identify 4 main clusters based on the company size and on the job/position of the respondent in organization (see Figure 3). Figure 3: 4 sub clusters based on two considerations: size and technical profile Firstly, we profile the respondents from personal data and family status. The sample age is 37. More than half have a bachelor degree as their highest academic achievement. The pre-university educational background is technical for almost two thirds of the respondents. In a stable relation are 66% of the respondents while the others are single. Half of the respondents have no family obligation. The remaining half has children (37%), elder 312 , relatives (8%) and both children and elder relatives (5%). Another interesting observation is that in the “big companies” all women in the age range 25-35 have no family obligations, while in the “other companies” 40% of women in the same age range have already children or elder relatives to take care of. This is just a fact that cannot be statistically proven as a general path because the sample is very small. Nevertheless it would be interesting to study more in depth this observation and if significant to understand also the causes. Secondly, we asked the respondents to assess the impact of enablers and obstacles on their career. The respondents were asked to assess the impact of 24 factors on their professional career. It is a discretional assessment aimed at understanding the respondents’ perception. As shown in Figure 4, four factors seem to clearly support the women’s professional career: problem solving focus (5.78), teamwork oriented (5.77), challenging environment (5.71) and high accountability (5.63). Factors like - customer oriented, recognize excellence/contributions, open communication, high integrity, fast paced, employee people oriented, entrepreneurial, results driven culture and empowering - partially support the professional career of the respondents. Instead the characteristics that partially hinder the professional career of respondents are the following four: male dominated (3.31), very conservative (3.19), non consensus (3.08) and hostile/threatining (2.65). Figure 4: Factors that support, partially support and hinder the professional career As regards the career progression, 84% of women say that they are satisfied with it. If we consider the two subclusters – IT and non technical profiles - we notice that the satisfied women in the IT field (73%) are less than in the non technical fields (89%). The reasons for being satisfied are many. The most cited one (73%) is the continuous learning of new things. The other reasons cited by half of respondents are: working with very talented people (52%), being continuosly challenged (51%) and intellectually stimulated (51%). 313 Women and ICT: exploring obstacles and enablers of a possible career As regards women that are not satisfied with their career progression (16%) half of them cites as the most important reason the fact that the company did not encourage, support nor develop women for top level positions. Another reason is the fact that they are excluded from high level decision making within company. The gender discrimination is the least and not the most important reason, as we might have expected for career unsatisfaction. Only one quarter of non satisfied women reckon it. When asked if the company, where they work, does undertake any policies to sustain women, 49% of respondents say that they do, 28% say that they do not, while 23% of respondents does not know. An interesting observation based on the cluster of IT and non IT women is that more than half (57%) of non IT women say that their company does undertake some policies while a bit less than half (47%) of IT women say that their company does not. We were expecting to see a much higher percentage for the IT cluster as the work environment in the IT domain is more hostile than in the non IT one, and companies should undertake some policies to sustain somehow women. This result might be due to the fact that women in IT are only 30 out of 100. It should be analyzed how does this percentage change when the number of IT women is equal to or higher than the number of non IT women. The most cited policies undertaken are the options for flexible working condition and location, programs that encourage female networking and role models, support programs and facilities that help reconcile work and family life. The research framework tries to understand whether the perception about the gender issue has changed alongside the women’s career. At the beginning of their career, 53% of IT women versus 41% of non IT women thought that their gender would influence their career. In the past, they were not concerned about the “gender” issue. Moreover they were confident that gender would not influence their career opportunities and advancement. The fact that women thought that being a woman could influence their career is positively correlated with the fact that having a successful career would be more difficult for a woman (0.29) and the fact that they were aware/ concerned and this hampered them (0.27). Today the IT women say that having a successful career has been more difficult for them, being a woman (4.5). On the other hand, non IT women partially disagree with the fact that there's a "gender issue" in their department (3.42). This finding should not surprise us as we know that the IT environment is more hostile for women than are other departments. Nevertheless this finding should be tested in a bigger sample to be sure of the result. In our data analysis apart from performing the descriptive statistics we have also analyzed some correlations. In particular we have searched for correlations between being satisfied with the career progression and all 24 characteristics that were assessed by our respondents as hindering or supporting their career progression. The results are the following: the work environment that is male dominated and team 314 , work oriented has a positive correlation of 0.19 and respectively of 0.31 with women’s career progression satisfaction. The male dominated work environment has more to do with IT women and the team work oriented environment has more to do with the non IT women. If we take in consideration the two clusters based on the technical profile of the repondents, the IT and non IT women, we realize that the positive correlation with the characteristic “male dominated” exists in both clusters but is higher for the IT (0.3) than for the non IT (0.1). This finding suggests that the male dominated work environment has more to do with IT women’s career progression satisfaction than with the non IT one. We did not expect that the male dominated characteristic of the work environment might be positively correlated with women’s career progression satisfaction. This might suggest that the ones who replied to the questionnaire have developed some male characteristics and do enjoy the male dominated culture. Nevertheless this is just our observation that cannot be statistically proven as the sample is small and also the correlation even if positive is very small. The positive correlation with “team work oriented” characteristic is present in both clusters but with a different magnitude this time: for the IT cluster (0.12) is lower than for the non IT one (0.48). The work environment that is team work oriented has more to do with non IT women’s career progression than with the IT one. For IT women the work environment that is very competitive is highly and positively correlated (0.4) with their career progression satisfaction. This finding suggests that a very competitive work environment is a driver for the IT women’s satisfaction in their career progression. We have to admit though that we lack any information about the respondents’ personality. It might be that the women who chose ICT are very competitive by nature and therefore decide to make their career in this field. For non IT women instead the work environment that is employee and people oriented is positively correlated (0.28) with their career progression satisfaction . This finding suggest that the work environment that is employee and people oriented might have something in common with non IT women’s satisfaction. We have analyzed also the correlations between the fact that companies do undertake policies to sustain women and all 24 characteristics that were assessed by our respondents as hindering or supporting their career progression. The results are the following: the undertaking of policies, for all 100 respondents, is positively correlated with the following characteristics: team work oriented (0.28), male dominated (0.23), entrepreneurial (0.23), high accountability (0.2), open communication (0.2) and diversity not valued (0.19). This finding might be interpreted in the following way: the work environment that is team work oriented, male dominated, entrepreneurial, with high accountability and open communication might have something to do with the fact that the company undertakes some policies to sustain women. The negative correlations that exist are very low and therefore irrelevant. If we take in consideration the two clusters based on technicality of the job (IT and non IT profiles), we realize that the positive correlation with the characteristic “male dominated” exists in both clusters but is much higher for the IT (0.23) than for the non 315 Women and ICT: exploring obstacles and enablers of a possible career IT cluster (0.06). This might be interpreted in the following way: the work environment that is male dominated has to do with the fact that companies where IT women work undertake policies to sustain women. The positive correlation with “team work oriented” characteristic is present in both clusters but with a different magnitude this time: IT cluster (0.09) and the non IT one (0.17). This might be interpreted in the following way: the work environment that is team work oriented has to do with the fact that companies where non IT women work, undertake policies to sustain women. Moreover we have analysed the correlations between the fact that the respondents at the beginning of their career thought that being a woman could influence their career and their vision back then. The results are the following: the fact that women thought that being a woman could influence their career is negatively correlated with the fact that they thought gender was irrelevant (-0.59), that gender would not influence their career opportunities and advancement (-0.42) and that they were not aware of any “gender issue” (-0.33). The positive correlations are instead with the fact that having a successful career would be more difficult for a woman (0.29) and the fact that they were aware/ concerned and this hampered them (0.27). As regards the clusters based on size and those based on the technicality of the job, there are no relevant differences. In addition we have also analyzed the correlations between the fact that the respondents at the beginning of their career thought that being a woman could influence their career and their vision today. The results are the following: The fact that women thought that being a woman could influence their career is negatively correlated with the fact that today the gender is not influencing their career opportunities and advancement (-0.24). This statement may be interpreted in the following way: as today’s belief, that gender is not influencing career opportunities, “increases” (becomes stronger), the past belief that gender could influence career, “decreases” (becomes less strong). As regards the two clusters based on the technicality of the job (IT and non IT), the only difference from the general path is that the IT cluster is not correlated at all (0.00) with the consideration that “my gender is giving me the possibility to make a difference” while the non IT cluster (0.31) and also the whole sample (0.23) is positively correlated with this characteristic. We have also tried to verify some answer patterns. For example it would have been interesting to see how those who are satisfied with their career progression evaluate their vision back then about the gender relevance and their today’s vision. Unfortunately, the results obtained in doing the regressions were not significant at all. This was due to the small size of the sample. 316 , 5 Conclusions The data analysis shows that there are some cultural and environmental factors that hinder (male dominated culture and the conservative, non consensus and hostile work environment) and some others that support (challenging work opportunities and employee oriented workplace culture) women’s professional career. Male dominated characteristic of the work environment is positively correlated with career progression satisfaction of women from the ICT related cluster. Policies that companies undertake to sustain women: the percentage in the ICT related cluster is lower than in the non ICT related one. Women have satisfying careers and enjoy the work they do. Nevertheless the satisfaction of the two clusters is correlated with opposite work environment characteristics (a very competitive work environment is positively correlated (corr=0.4) with the IT cluster while the team work oriented characteristic of the work environment is positively correlated (corr=0.48) with the non IT cluster). In the past, the two clusters perceived the “gender issue” in the same way. They were not concerned about the “gender issue”. Moreover they were confident that gender would not influence their career opportunities and advancement. Now, the issue is perceived differently. The IT cluster is more affected by gender, as they say that having a successful career has been more difficult for them, being a woman. On the other hand, the non IT cluster disagrees with the fact that there’s a “gender issue” in their department. The results of this first effort to explore the gender issue in the ICT industry in Italy are surely interesting, but the present research presents clear limitations. First of all the sample, as we have only 30% of the respondents working properly in the IT department. The other 70% of respondents work for ICT companies, but in departments different from the IT one. Extending the survey and focusing mainly to women working in the IT department would be helpful both statistically and in terms of contents. Another relevant future step would be considering the intrinsic characteristics of women working in IT departments and comparing them with women in other departments of the companies. By exploring this topic, many of our not statistically strong observations will become clearer. In fact, we might understand better how the “gender issue” is perceived and how it has influenced the career path and to what extent it is linked to the personal profile of each woman involved in the survey. Another modification suggested regards the questionnaire, where some of the questions have resulted not relevant or too detailed for the research. Additional questions exploring possible links between career satisfaction/progression and the marital status or family obligations of a woman could also help in the answers. 317 Women and ICT: exploring obstacles and enablers of a possible career In synthesis, this research is just the first step to start understanding an interesting topic whose relevance is high also for the new generations and their career choices and models. References  Reports World Economic Forum’s 2012 Global Gender Gap report. Corporate Gender Gap Report 2010. World Economic Forum. Geneva, Switzerland. Institute of Leadership and Management (2011): Ambition and gender at work. pp5, London, UK.  Government publications Eurostat (2011): Human Resources in Science and Technology (HRST). Eurostat (2012): Labor Force Survey2012Q4. Data for EU27. European Commission (2013) Women Active in the ICT sector. European Commission: database on women and men in economic decision making. European Union Labor force survey- annual results 2012. Istat: Time Use in Daily Life (2008): A Multidisciplinary Approach to the Time Use's Analysis, Argomenti No. 35. Istat: BES 2013 Il benessere equo e sostenibile in Italia.  Journal Articles Aaltio, I. I. and Huang J. (2007). Women managers' careers in information technology in China: high flyers with emotional costs? Journal of Organizational Change Management, 20(2): 227-244. Ahuja, M., Chudoba, K., Kacmar, C., McKnight, H. and George, J. (2007). IT Road Warriors: Understanding the Role of Work-Family Conflict, Autonomy and Work Overload on Turnover Intention. MIS Quarterly, 31(1): 1-17. 318 , Bagilhole, B., Powell, A., Barnard, S. and Dainty, A. (2007).Researching Cultures in Science, Engineering and Technology: An analysis of current and past literature. Research Report Series for UKRC 7. Bartol, K. M., Williamson, I. O., and Langa, G. A. (2006). Gender and professional commitment among IT professionals: The special case of female newcomers to organizations. Coga, P. A., and Chen, C. P. (2007). Career development and counseling for women: Connecting theories to practice. Counseling Psychology Quarterly, 20(2): 191-204. Crump, B. J., Logan, K. A. and Mcllroy A. (2007). Does Gender Still Matter? A Study of the Views of Women in the ICT Industry in New Zealand. Gender, Work and Organization, 14(4): 349-370. Denison, D. R. (1996). What is the difference between organizational culture and organizational climate? A native’s point of view on a decade of paradigm wars. Academy of Management Review, 21(3): 619-655. Esping-Andersen, G. (1999). Social foundations of postindustrial economies. Oxford: Oxford University Press. Ferrera, M. (1996). The Southern Model of Welfare in Social Europe. Journal of European Social Policy 6(1), 17-37. Gras-Velazquez, A., Joyce A., A. and Debry, M. (2009). Women and ICT: why are girls still not attracted to ICT studies and careers? EUN White Paper, European Schoolnet (EUN Partnership AISBL). Griffiths, M. and Moore, K. (2010): Disappearing Women: A Study of Women Who Left the UK ICT Sector. Journal of Technology Management and Innovation. Hofstede, G. (1980): Cultures’ Consequences: International Differences in Work- Related Values. Beverly Hills, CA: Sage. Hunt, J. (2012): Why Do Women Leave Science and Engineering? Discussion Paper No. 6885. IZA Institute for the Study of Labor. Igbaria, M., and Greenhaus, J. (1992). Determinants of MIS employees’ turnover intentions: A structural equation model. Communication of the ACM, 35(2): 35- 49. 319 Women and ICT: exploring obstacles and enablers of a possible career Kuhn, S and Rayman, P (2007). Women on the edge of change: Employees in United States information technology companies. In C. Burger (Ed.), Reconfiguring the firewall: a cross-cultural context for recruiting women to IT. A.K. Peters, Ltd. Lemons, M. A., and Parzinger, M. J. (2001). Designing Women: A Qualitative Study of the Glass Ceiling for Women in Technology, Advanced Management Journal: 4-11. Margolis, J., and Fisher, A. (2002). Unlocking the Clubhouse: Women in Computing. Cambridge, MA: MIT Press. McLean, D. (2003). Workplaces that work: Creating a workplace culture that attracts, retains and promotes women. The Centre of Excellence for Women’s Advancement, The Conference Board of Canada. Ontario, Canada. Michie S., Nelson D.L. (2006). Barriers women face in information technology careers: Self-efficacy, passion and gender biases, Women in Management Review, Vol. 21 (1), pp. 10 – 27. O’Neil, D. A., Hopkins, M. M., and Bilimoria, D. (2008). Women’s Careers at the Start of the 21st Century: Patterns and Paradoxes. Journal of Business Ethics, 80(4): 727-743. Quesenberry, J., Trauth, E.M. and Morgan, A. (2006). Understanding the ‘Mommy Tracks’: A Framework for Analyzing Work-Family Issues in the IT Workforce. Information Resources Management Journal, Volume 19, Number 2: 37-53. Riemenschneider, C. K., Armstrong, D. J., Allen, M. W., and Reid, M. F. (2006). Barriers facing women in the IT workforce. The DATA BASE for Advances in Information Systems, 37(4): 58-78. Svinth, L. (2006). The leaky pipeline’ – to be or not to be a useful metaphor in understanding the under-representation of women in science academia. Paper presented at the 6th European Gender Research Conference. Roldan, M., Soe, L., and Yakura, E.K. (2004). Perceptions of chilly IT organizational contexts and their effect on the retention and promotion of women in IT. Proceedings of the 2004 SIGMIS conference on Computer personnel research: Careers, culture, and ethics in a networked environment: 108-113. 320 , Tapia, A. H., and Kvasny, L. (2004). Recruitment is never enough: Retention of women and minorities in the IT workplace. Proceedings of the 2004 SIGMIS Conference on Computer Personnel Research: Careers, Culture, and Ethics in a Networked Environment: 84-91. Trauth, E. M. (2002). Odd girl out: An individual differences perspective on women in the IT profession. Information Technology and People, 15(2):98-118. Trauth, E. M., Quesenberry, J. L., and Morgan, A. J. (2004). Understanding the under representation of women in IT: Toward a theory of individual differences. Proceedings of the ACM SIGMIS Computer Personnel Research Conference, Tucson, AZ. Trauth E.M., Quesenberry J.L., and Yeo, B. (2008). Environmental influences on gender in the IT workforce. The DATA BASE for Advances in Information Systems, 39(1): 8-32. Von Hellens, L., Nielsen, S. H., and Trauth, E. M. (2001). Breaking and entering the male domain: Women in the IT industry. Proceedings of the ACM SIGCPR Conference, San Diego, CA. Wajcman, J. (1991). Feminism confronts technology. University Park, PA: The Pennsylvania University Press. Woodfield, R. (2002). Women and information systems development: Not just a pretty (inter)face? Information Technology and People, 15(2): 119-138. 321 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Understanding Online Channel Expansion in an SME Context: A Business Model Perspective John Jeansson Linnaeus University, Kalmar, Sweden John.jeansson@lnu.se Shahrokh Nikou IAMSR/Åbo Akademi University, Turku, Finland Shahrokh.nikou@abo.fi Rune Gustavsson KTH, Stockholm, Sweden runeg@kth.se Siw Lundqvist Linnaeus University, Sweden Siw.lundqvist@lnu.se Leif Marcusson Linnaeus University, Sweden Leif.marcusson@lnu.se Anna Sell IAMSR/Åbo Akademi University, Turku, Finland Anna.sell@åbo.fi Pirkko Walden IAMSR/Åbo Akademi University, Turku, Finland Pirkko.walden@abo.fi Abstract The purpose of the paper is to study, from a business model perspective, value creating activities taken by SMEs when making a transition to an online multichannel context by 322 John Jeansson et al. adopting and adding e-commerce and/or m-commerce. 16 SMEs in Sweden are studied using a basic qualitative research approach and an e-transit business model configuration. Main results of the study are the existence of primary and secondary transition activities and the existence of a discrepancy between actions taken and their perceived degree of importance. One main conclusion is that the combination of value creating activities an SME should focus on during different stages of an online channel expansion differ depending on transition category and will change over time. Keywords: Business models, e-commerce, m-commerce, retail, small and medium sized enterprise 1 Introduction Over the last decades the traditional way of creating and capturing value in retail has gone through tremendous changes. It is no longer sufficient for a small and medium- sized enterprise (SME) to merely rely on face-to-face customer interaction as online channels (electronic and mobile commerce) have grown in importance and provide means to virtually interact with customers (Chen, et al. 2014). Customers in mature e- commerce markets such as UK, France, and Sweden now expect retailers to provide an integrated experience throughout several channels (Ecommerce-Europe, 2014; Postnord, 2014). Expanding into online channels could mean greater opportunities for SMEs. Studies on SMEs and e-commerce adoption suggest a positive influence on average sales growth rates (Abebe, 2014), financial gains both in terms of revenue growth and cost reduction (Johnston, et al. 2007), access to a wider range of markets, enhanced communication, and improved customer service (Stockdale & Standing, 2004). However, channel expansion is also very much a time of transition and new structures, adding greater complexity to how business is done. As SMEs are faced with new opportunities through online channels, the model by which they create and capture value is challenged, making the adoption of a business model that fits the organisation a crucial strategic decision (Chatterjee, 2013; Li, Troutt, et al., 2011). Zott, et al. (2011) present business models as a theoretically robust construct useful to researchers and practitioners alike in their quest to understand the realities of doing business in a highly complex and connected world. They provide four central themes describing the characteristics of business models: one, a business model seeks to understand the logic of how value is created and captured, two, a business model sets focus on activities performed by internal as well as external stakeholders, three, a business model emphasizes a holistic, system-level approach when explaining how SMEs do what they do, four, a business model is a unit of analysis that is centred around a specific SME yet its boundaries are wider and includes business partners as well as customers (Zott & Amit, 2013; Zott et al., 2011). The purpose of this paper is to study online channel expansion of SMEs from a business model perspective. Such a perspective enables a holistic approach in order to better understand value creating activities taken by SMEs during online channel expansion. In order to do so, eight business model components from three different existing business model frameworks have been placed together into what, for the purpose of this paper, is named an e-transit business model configuration. The research question this paper addresses is: what value creating activities are taken by SMEs when expanding to online channels? 323 Understanding Online Channel Expansion in an SME Context 2 Related Work This paper is positioned between the phenomenon of online channel expansion, the context of SMEs, and the business model as a unit of analysis. As these three fields converge they constitute the theoretical backdrop of the e-transit business model configuration. 2.1 Online Channel Expansion In this paper the concept of online channels encompasses e-commerce and m- commerce. E-commerce is defined as: “the process of buying, selling, or exchanging products, services, or information via computer networks” (Turban, et al. 2006, p.4), and m-commerce is defined as: “any transaction, involving the transfer of ownership or rights to use goods and services, which is initiated and/or completed by using mobile access to computer-mediated networks with the help of an electronic device.” (Tiwari, et al. 2006, p.40). Any combination of e-commerce, m-commerce, and the physical store is often referred to as a multichannel retail landscape (Zhang et al., 2010). When making a transition from a single channel to a multichannel environment there are several issues to consider. Weill and Vitale (2001) stress the need of business model configuration, managing points of customer interaction, understanding targeted customer segments, and creating the IT infrastructure capability. Valos (2009) emphasises the need of a strategic approach to market communication and Thomas and Sullivan (2005) stress the importance of leveraging enterprise-level data in order to understand and predict customersćhannel choices. Zhang et al. (2010) suggest that organisations build a structure where they manage multiple channels instead of each channel on their own. They argue that organisations need to balance what to offer in all channels and to what extent each channel with its distinct characteristics should be allowed to be unique. In the end, such a balance is quite unique and something for each organisation to find (Avery, Steenburgh, Deighton, & Caravella, 2012; Wagner, Schramm-Klein, & Steinmann, 2013). 2.2 Business Model Perspective Business model research within an online context has increased over the years (Zott et al., 2011). There are several definitions of what constitutes a business model. In this paper, a business model is considered to be a model describing the rationale of how an organisation creates, delivers, and captures value, as well as a unit of analysis that enables a holistic understanding of the activities a focal organisation, including their partners, conducts in order to “do business” (Osterwalder & Pigneur, 2010; Weill & Vitale, 2001; Zott & Amit, 2013; Zott et al., 2011). As a response to environmental changes, such as making a transition to a multichannel environment, organisations often find themselves innovating existing business models (Schneider & Spieth, 2013). Gunzel and Holm (2013) argue that such a business model innovation process is quite differentiated. Innovation at the front-end of the business model tends to be more trial- and-error oriented where back-end tends to be more coupled with a linear approach. Moingeon and Lehmann-Ortega (2010) speak of business model creation as a learning process, where identifying core objectives and developing business specific profit logic are needed in order to successfully design business models (Chatterjee, 2013). Business model innovation processes, especially during online transitions, are highly information 324 John Jeansson et al. driven and depend to a large extent on how well organisations are able to connect technology and strategy (Ja-Shen & Ching, 2002; Weill & Vitale, 2001). 2.3 The Context of SMEs The studied SMEs in this paper are located in Sweden. SMEs constitute the backbone of both the European and the Swedish economy, accounting for 99.8% of all enterprises and 66.8% of total employment in the European non-financial business sector, and for 99.9% of all enterprises and 65.8% of employments in the Swedish non-financial business sector (EuropeanCommission, 2014a). In this paper the European Union definition of SMEs is used, which encompasses micro, small, and medium-sized enterprises. A SME employs fewer than 250 persons and has an annual turnover not exceeding EUR 50 million or an annual balance sheet total not exceeding EUR 43 million (EuropeanCommission, 2014b). SMEs possess specific characteristics that are of interest when to understand their online channel expansion, see table 1. When adopting online channels SMEs tend to be heavily influenced by the level of IT skills within the organisation, and access to help outside it. SMEs risk propensity also affects adoption, having a direct impact during pre-adoption decision and more of an indirect impact when deciding to continue using online channels (Wilson, et al. 2008). Karjaluoto and Huhtamäki (2010) argue that SMEs utilize online channels in various ways in order to reach different goals and that there is not one ideal level of online channel usage that would be the same for each company. However, Mustaffa and Beaumont (2004) found that SMEs are especially keen to utilize online channels in order to extend their geographical reach. Overall, studies of European companies online channel adoption show that SMEs are late adopters and laggards (Vlachos, 2011). Perspectives SME Characteristics Reference Owner is the CEO, highly visible and close to point of delivery, (Zach et al., 2014; Management centralized and intuitive decision making, time constraints, Wong and Aspinwall, modest in skills and competence. 2004; Ghobadian and Gallear, 1997) Mostly local and regional markets-few international, (Zach et al., 2014; Customers and dependent on small customer base, close and frequent Wong and Aspinwall, market contact with customers, many known customers personally 2004; Ghobadian and and socially. Gallear, 1997) (Zach et al., 2014; Processes and Flexible, adaptable and less complicated processes, low Wong and Aspinwall, procedures degree of standardization, focus on operational processes- less focus on strategic processes, mostly people dominated. 2004; Ghobadian and Gallear, 1997) Flat with fewer levels of hierarchy, flexible, often rapid (Zach et al., 2014; response to environmental changes, low degree of Wong and Aspinwall, Structures specialization, multi-tasked owner-managers, limited and 2004; Ghobadian and unclear division of activities, organic and fluid culture, low Gallear, 1997) resistance to change, influenced by owner-manager. Flexible information flows, limited knowledge of IS, limited Information systems managerial expertise and attention, lack of strategic planning, (IS) and information limited in-house IT expertise, greater reliance on third parties, (Zach et al., 2014) technology (IT) emphasis on packaged applications, IS function in its earlier stages. Modest financial resources, limited human capital, more (Zach et al., 2014; Resources versatile employees, no specific training budget, training and Wong and Aspinwall, staff development likely to be ad hoc and in small scale, 2004; Ghobadian and Gallear, 1997) Table 1: General characteristics of SMEs. 325 Understanding Online Channel Expansion in an SME Context 3 E-transit Business Model Configuration A generic business model framework has its value, however, the existence of several different frameworks with different components could suggest that certain components might be more suitable in certain contexts. Here we identify and use a configuration of business model components corresponding to the studied context, rather than advocating for a new generic business model framework. Components of the e-transit business model configuration have their theoretical foundation in the studies of Weill and Vitale (2001), Osterwalder and Pigneur (2010) and El Sawy and Pereira (2013). Osterwalder, et al. (2005) make an effort to identify the most common business model components from literature and compile them into one framework, the business model Canvas, which includes work from Weill and Vitale (2001) but not from El Sawy and Pereira (2013). They identify nine components: customer segments, value proposition, channels, customer relationships, revenue streams, key resources, key activities, key partnerships, and cost structures (Osterwalder & Pigneur, 2010). The nine components could be divided into two categories, where key resources, key activities, key partnerships, and cost structures conduct the back-end of the business model emphasising efficiency, and where customer segments, value proposition, channels, customer relationships, and revenue streams conduct the front-end of business models emphasising value (Gunzel & Holm, 2013; Osterwalder & Pigneur, 2010). However, the business model Canvas does not take the online context into specific considerations. Weill and Vitale (2001), on the other hand, take an e-business model archetype perspective when studying organisations transition from a physical context to an e- business context. They propose eight components: customer segments, value proposition, channels, customer relationship, sources of revenue, core competencies, key information, and IT infrastructure. They emphasize the importance of information and the ability to capture, share, and exploit key information in order to be successful (Weill & Vitale, 2001). However, Weill & Vitale (2001) do not - for obvious reasons - take the mobile, often service oriented platform, channel into consideration, which calls for yet another framework. El Sawy and Pereira (2013) propose the VISOR business model framework, which has a digital platform perspective and an ecosystem approach. The framework has five components: value proposition, interface, service platforms, operational model, and revenue model. They especially emphasize the importance of interface, which is the user interface experience including both hardware and software. It links the value proposition with the IT infrastructure that delivers it. They also stress the importance of service platforms, which enable, shape, and support business processes and relationships needed in order to deliver proposed value as well as to improve the same (El Sawy & Pereira, 2013). Together these three existing business model frameworks provide specific components and perspectives regarded as suitable to online channel expansion in a SME context. 3.1 Configuration Perspectives Adopting a business model perspective is very much to set focus on SMEs value creating activities. SMEs, in general, tend to be activity focused (Ghobadian & Gallear, 1997; Wong & Aspinwall-Roberts, 2004), and each selected e-transit configuration component is given an activity oriented name and described from an activity perspective, see table 2. The eight perspectives are: knowing the customer, offering 326 John Jeansson et al. value, creating points of interactions, finding new ways, making money, building networks, informating, and optimizing resources. Some of the perspectives correspond to all three existing business model frameworks and some correspond to only one or two of the frameworks. Configuration perspective Activity description Existing business model framework component Knowing the Activities taken by an SME in order to identify Customer segments, customer and describe its customers, as well as to Customer relationships, build relationships through different channels (Osterwalder and Pigneur, 2005). using different tools. Customer segments, Customer relationships, (Weill and Vitale, 2001). Value proposition, (El Sawy and Pereira, 2013). Offering value Activities taken by an SME in order to provide Value proposition, (Osterwalder and a package of both a solution to a perceived Pigneur, 2005). problem as well as a specific product/service Value proposition, (Weill and Vitale, to a specific customer segment through a 2001). specific channel. Value proposition, (El Sawy and Pereira, 2013). Creating points of Activities taken by an SME in order to design Channels, (Osterwalder and interactions and facilitate places where the company and Pigneur, 2005). the customer could interact and where Channels, (Weill and Vitale, 2001). transactions of information, money, products Interface, (El Sawy and Pereira, and services could take place. 2013). Finding new ways Activities taken by an SME in order to add Key activities, (Osterwalder and and create value to and through their product, Pigneur, 2005). service and information that a specific NA, (Weill and Vitale, 2001). customer is willing to pay for. Organizing model, (El Sawy and Pereira, 2013). Making money Activities taken by an SME in order to Revenue streams, cost structures, increase revenue streams as well as to (Osterwalder and Pigneur, 2005). manage costs of doing business. Sources of revenue, (Weill and Vitale, 2001). Revenue model, (El Sawy and Pereira, 2013). Building networks Activities taken by an SME in order to Key partnerships, (Osterwalder and establish a value network of different actors Pigneur, 2005). to be part of the companyś value proposition. NA, (Weill and Vitale, 2001). Organizing model, (El Sawy and Pereira, 2013). Informating Activities taken by an SME in order to gain NA, (Osterwalder and Pigneur, access to key information and to exploit that 2005). information in order to make informed Key information, (Weill and Vitale, decisions regarding existing and future 2001). success of chosen business model. NA, (El Sawy and Pereira, 2013). Optimizing Activities taken by an SME in order to acquire Key resources, (Osterwalder and resources and maintain resources in a way that makes Pigneur, 2005). the most out of them. Resources could be IT infrastructure, core competencies, physical, financial, intellectual or human, (Weill and Vitale, 2001). including core competencies, IT- EService platforms, (El Sawy and infrastructure, and digital service platforms. Pereira, 2013). Table 2: Configuration perspectives used in the e-transit business model configuration with corresponding components from existing business model frameworks. 328 Understanding Online Channel Expansion in an SME Context 4 Research Method Conducted research uses what Merriam (2009) refers to as a basic qualitative research approach. In qualitative research a holistic view is desired of that which is studied and to understand a phenomenon from a participantś perspective (Merriam, 2009; Miles & Huberman, 1994). 4.1 Sample Selection A purposeful sampling strategy was used when selecting participating SMEs. The intention of such a strategy is to select SMEs that could provide rich information of the topic at hand (Patton, 2001). Based on Patton’s (2001) 16 sampling strategies this study used, what best could be described as, a combination between intensity, criterion, and convenience sampling. In order to be included in the study SMEs had to meet four criteria: first, to qualify as an SME according to the European commission definition (EuropeanCommission, 2014a), second, to have conducted at least one expansion including either or both e-commerce or m-commerce as a retail channel within the last five years prior the study, third, to sell products or services to consumers, fourth, to operate in Sweden. In total 16 companies were included and categorised based on three transition categories: PEM, which includes SMEs that made a transition from having a physical (P) channel to include both e-commerce (E) and m-commerce (M), PE, which includes SMEs that made a transition from having a physical channel to include e- commerce, and EM, which includes SMEs that made a transition from having an e- commerce channel to include m-commerce, see table 3. Respondents were selected based on their position and role within the company. In order to be selected they had to be either the owner-manager or a high-level decision maker or project manager who had been involved in making strategic decision regarding the channel expansion. Choosing respondents in such positions enables a rich picture of how decisions on channel expansion are made, factors influencing such decisions, value creating activities, and results of performed actions. 4.2 Data Collection Interviews were the main data collecting method. On average, each interview took 60- 90 minutes; they were conducted on site at each companyś head office, except at two occasions when a telephone interview had to be conducted due to practical issues. Each interview was recorded and transcribed verbatim shortly afterwards. The interviews were semi-structured and an interview guide was used at all interviews. Interview questions were carefully designed to cover each business model configuration perspective. Questions were asked with the purpose of capturing the perspective and worldview of respondents and did not necessarily follow the same order and wording in all interviews (Merriam, 2009). Each respondent were also asked to grade on a 5 point Likert scale, from strongly disagree to strongly agree, each business model configurationś perceived degree of importance to the success of performed channel expansion, see table 4. Data were also collected through available documentation and company websites (Creswell, 2007). 4.3 Data Analysis Initially the interview transcripts were printed and read carefully by at least two researchers. A first version of a list of codes was developed which had a descriptive 328 John Jeansson et al. character. The list was then revised during the different rounds of coding. Codes that did not work were deleted and codes that overlapped were merged (Miles & Huberman, 1994). Each interview transcript was manually open coded in order to categorise data according to the e-transit business model configuration perspectives. During this time codes and notes were made in each margin of the printed transcript (Merriam, 2009; Miles & Huberman, 1994). A second round of coding was conducted in which activities taken by SMEs in order to achieve their channel expansion were coded and categorised. A third round of coding was conducted in order to ensure that assigned codes actually denoted the meaning of underlying quotations (Miles & Huberman, 1994). As themes of activities emerged the following questions were asked in order to gain a deeper understanding: What activities can be identified and which stakeholders are involved and how? How often do identified activities occur and how are they spoken of in terms of importance? How are identified activities organised and how do they evolve over time? What initiates identified activities and what are their results? (Lofland, et al.2006). Transition category Transition category Transition category PEM PE EM Number of Companies 5 7 4 Size of companies (Micro/Small/ 2/2/1 4/1/2 3/1/0 Medium) Business Direct to customer, Direct to customer, model Direct to customer Franchise, Producer, Intermediary Hotel and hospitality, Industry home and furniture, IT- Clothes, handicraft, home Events, phone- accessories, fashion and and furniture, toys, hotel and accessories, hardware and styling hospitality tools Product/ Service 3/2 5/2 4 P→E Transition E→M/E→P E→M P E→P →E/M Table 3: Participating SMEs categorised based on transition categories. Transition describes the character of expansion (arrows indicate different structures of transition that were present within each category, e= e-commerce, m= m-commerce, p= physical store). 5 Results In this section respondents’ descriptions of online channel expansion activities are presented together with how they perceived each configuration perspectives’ degree of importance in relationship to undertaken online channel expansion. Together actions taken and perceived importance paint a rich picture of what participating SMEs regarded as important to do in order to succeed with their channel expansion. The results are categorised according to participating SMEs transition category. The proportion of identified activities and the level of perceived importance of each 329 Understanding Online Channel Expansion in an SME Context configuration perspective often varied within a transition category as well as aggregated between transition categories, see table 4. 5.1 Transition Category PEM SMEs that made a transition from having a physical channel to include both e- commerce and m-commerce described a large amount of activities. Activities varied in character and corresponded to seven of the eight e-transit configuration perspectives. The most frequently described configuration perspectives were: building networks, making money, and optimizing resources. The only perspective to which no description of activity could be assigned was the perspective of finding new ways. When asked to grade the importance of different configuration perspectives there were no perspective that all SMEs in this category perceived alike. However, aggregated there was a clear line between the three perspectives that most SMEs perceived to be of high importance (creating points of interactions, finding new ways, informating) and the five perceived to be of low importance (knowing the customer, offering value, making money, building networks, optimizing resources). Transition Transition category Transition category Transition category category PEM PE EM Proportion Perception Proportion Perception Proportion Perception Configuration of activities of of activities of of of perspective importance importance activities importance Knowing the 11% Low 4% High - High customer Offering value 3% Low 4% Low - Low Creating point 3% High 4% High - High of interactions Finding new - High - High - Low ways Making money 21% Low 18% High 43% Low Building 24% Low 33% High 14% Low networks Informating 7% High 4% High - Low Optimizing 31% Low 33% High 43% High resources Table 4: The table shows proportion of described activities by SMEs together with SMEs perception of each configuration perspectiveś degree of importance (High = a majority of SMEs perceived the perspective to be of importance, Low = a majority of SMEs perceived the perspective not to be of importance). 5.2 Transition Category PE SMEs that expanded from a physical channel to include an e-commerce channel described fewer activities than SMEs in the PEM category, but substantially more than SMEs in the EM category. Activities varied in character and corresponded to seven of the eight e-transit configuration perspectives. The configuration perspectives of making money, building networks, and optimizing resources were by far the most frequently described. SMEs within the PE category perceived four of the eight configuration perspectives alike, all four being perceived to be of high importance (knowing the customer, creating points of interactions, informating, optimizing resources). 330 John Jeansson et al. Aggregated, only the configuration perspective of offering value was by a majority of SMEs perceive to be of low importance. 5.3 Transition Category EM SMEs that made a transition from an e-commerce setting to include m-commerce described a significantly lesser amount of activities than SMEs within the other two categories. Activities showed a limited variation in character and corresponded only to three configuration perspectives, making money, building networks, and optimizing resources. Within the EM transition category there were two configuration perspectives that SMEs perceived alike, finding new ways and making money, which were perceived to be of low importance. Just as within the PEM category there was a clear line between the three perspectives that most SMEs within the EM category perceived to be of high importance (knowing the customer, creating points of interactions, optimizing resources) and the five perceived to be of low importance (offering value, finding new ways, making money, building networks, informating). 5.4 Cross-category Connections The configuration perspectives of making money, building networks and optimizing resources were the most frequently described perspectives in all three transitions categories. Only two configuration perspectives were perceived alike in all transitions categories: offering value was perceived to be of low importance in all transition categories, and creating points of interactions was perceived to be of high importance. Apart from that, SMEs within the PEM and PE category shared the same perceived degree of importance (high) in two configuration perspectives (finding new ways, informating); PEM and EM shared the same perceived degree of importance (low) in two configuration perspectives (making money, building networks); PE and EM shared the same perceived degree of importance (high) in two configuration perspectives (knowing the customer, optimizing resources). 5.5 Activities of Value Creation Activities of value creation are descriptions made by SMEs of how they acted in order to overcome challenges when conducting their channel expansion. Activities are categorised based on configuration perspectives and grouped into primary and secondary activities. Primary activities correspond to configuration perspectives present in all three transition categories, and secondary activities correspond to perspectives not present in all transition categories. 5.6 Primary Activities Ÿ Activities corresponding to making money; SMEs in the PEM transition category described managing payment solutions when conducting business in other countries as an important activity. SMEs in the PE category also spoke of payment systems but stressed the importance to reach a good agreement with payment service providers. All transition categories described how they had to manage demands from suppliers to have a certain price level for their products online, across channels and between countries. Managing increased cost was also an activity within this perspective that was described in all transition categories. Ÿ Activities corresponding to building networks was described in all transition categories but with great variety. Within the PEM category SMEs frequently spoke of 331 Understanding Online Channel Expansion in an SME Context managing their logistic partners. Both within the PEM and PE category SMEs described how they needed to manage and overcome suppliers fear or scepticism of using the e- commerce channel. SMEs in both the EM and PEM category spoke of the importance to be precise in ones requirements and to avoid lock-in when managing their relationships with developers of technical solutions. SMEs in the PE category emphasized the challenge to find the right partner within their network to get support. SMEs in the PEM category spoke of finding partners that would be willing to contribute to value creation and not just try to profit for themselves. SMEs within the PEM category also mentioned the management of third party booking platforms and to gain influence over its future development. Ÿ Activities corresponding to optimizing resources; there were three activity areas that dominated SMEs descriptions. One, activities of competence was described by SMEs in both the PEM and PE transition category in terms of finding the right competence and to integrate new competence within the organisation. Two, activities of technical infrastructure were described in all three transition categories and consisted of activities in order to develop and integrate the different technical platforms with each other, especially integration between enterprise systems and e-commerce platforms. Three, activities of content were described in all three transition categories. SMEs in both the PEM and EM transition category spoke of challenges when managing content in a mobile context having to adapt to a smaller screen size. SMEs in the PE category spoke of creating a well structured website that could be easily navigated. 5.7 Secondary Activities Ÿ Activities corresponding to knowing the customer; SMEs in the PEM category described activities such as, supporting customer adoption of new technical solutions and managing customer loyalty. SMEs in the PE category spoke of building closer relationship with customers to prevent them from moving to other retailers. SMEs in the EM category did not explicitly mention activities related to this configuration perspective. Ÿ Activities corresponding to offering value; SMEs in the PEM category spoke of the importance of explaining to customers in a pedagogic way the company’s delivering times and shipment costs. SMEs in the PE category described the importance of being able to visualize products digitally in a way that made them justice. SMEs in the EM category did not explicitly mention activities related to this configuration perspective. Ÿ Activities corresponding to creating points of interactions; SMEs in the PEM category mentioned creating the same customer experience no matter which channel customers chose. PE category SMEs described the importance of integrating their web store and physical store. SMEs in the EM transition category did not explicitly speak of creating point of interactions. Ÿ Activities corresponding to informating; SMEs in the PEM category described managing product information updates from suppliers and managing customer reviews and ratings on websites and social media sites in a transparent fashion as important informating activities. SMEs in the PE category expressed difficulties in having the time to take hold of all the data/information available. SMEs in the EM category did not explicitly mention activities related to this configuration perspective. Ÿ Activities corresponding to finding new ways was the only perspective to which no activities described by SMEs could be explicitly related. 332 John Jeansson et al. 6 Discussion The results of the study indicate that there are three perspectives: making money, building networks, and optimizing resources, that regardless of studied transition type need the most attention when SMEs expand to an online multichannel setting. Interesting to notice is that none of the three perspectives is customer oriented, which often is stressed as an important area to work with (Thomas & Sullivan, 2005; Weill & Vitale, 2001). Both activities of building networks and optimizing resources are by Osterwalder and Pigneur (2010) and Gunzel and Holm (2013) related to back-end business model components having a strong efficiency drive. This suggests that during the initial stages of online transitions SMEs are more occupied with building and creating the functionality of the new channel, rather than getting to know and interacting with customers and developing existing value proposition beyond that which the channel itself provides. The identified difference between SMEs descriptions of activities and their perceived degree of importance is also quite interesting. It could suggest that the level of attention a certain area in a business model requires changes over time. When respondents described activities taken they reflected on what they actually did, retrospectively, in order to conduct the transition. When asked to grade degree of importance they did so not only in retrospective but also with current state in mind, when having accomplished the transition. The results then indicates that as the transition comes into place and the initial phase is over the scope of SMEs changes and to some degree broadens, which the high grading of creating points of interactions could be an articulation of. Creating points of interactions is by Osterwalder and Pigneur (2010) and Gunzel and Holm (2013) related to front-end business model components having a value drive. Studied SMEs could then be argued to have shifted the character of their activities from an initial efficiency-focus to an expanded value-focus. This emerging picture of SMEs activities corresponds to some degree with identified SME characteristics as they tend to be managed more at an operative than a strategic level, have a limited number of employees, be highly dependent on external technology competencies, scarce on resources, and mostly have a local and well known customer base (Ghobadian & Gallear, 1997; Wong & Aspinwall-Roberts, 2004; Zachet al. 2014). It is also interesting to notice the difference in described amount of activities between transition categories. SMEs that included an m-commerce channel did only describe a moderate amount of activities with an even greater efficiency-focus. This does not suggest that an m-commerce transition is easily done. It could however suggest that when SMEs already have made one online transition (in our case e-commerce) some of the challenges of the online context have already been faced. Left are activities needed to address specific issues regarding m-commerce, for example adapting content to a smaller screen size. The use of the e-transit business model configuration contributed to a rich picture of activities taken by SMEs during their channel expansion. The chosen configuration perspectives corresponded well with described activities and all activities could be placed within a configuration perspective. However, the configuration perspectives of informating and creating points of interactions that were explicitly or heavily influenced by the more specific online business model frameworks, did not correspond as expected. Activities of informating, which was proposed to be of great importance by Weill and Vitale (2001) in an online context, supported by Thomas and Sullivan (2005), was not 333 Understanding Online Channel Expansion in an SME Context mentioned by SMEs to any great extent, although graded to be of high importance by SMEs expanding from a physical channel. When spoken of by SMEs it was often in the context of them knowing the importance and possibilities of the configuration perspective but at the moment not having the time or resources to act on that understanding. The same pattern could be seen with the perspective of creating points of interactions, which generated even less corresponding activities. However, this perspective was the only one that all transition categories graded to be of high importance. These results indicate that both configuration perspectives were relevant in the e-transit business model configuration they were however not as evident during the initial stages of studied SMEs channel transition. 7 Conclusions, Limitations, and Further Research As studied SMEs embarked on their online channel expansion they not only initiated a technology change, they also initiated a change in how to create and offer value to their customers. In other words, their decision to pursue possible benefits of online channel expansion changed their business model. Stated research question targeted the understanding of what value creating activities SMEs take when expanding to online channels. By using a business model perspective a rich picture of two main sets of activities has been gained, primary and secondary transition activities. A discrepancy was also found between activities SMEs actually took and what they perceived to be of importance to take. The practical implication of this is knowledge offered to SMEs, whom are thinking of conducting an online channel expansion, of which kind of issues to expect and to some degree when to expect them. Theoretical implications of the study are both the suggested e-transit business model configuration, and a richer picture of the characteristics of SMEs actions during times of business model change and online channel transition. The study is not without limitations and should be read and understood based on its context. Interviews were made with owner-managers, high-level decision makers, or project managers within a company and not with external stakeholders or partnering companies. Including external stakeholders might have given a more holistic understanding of activities needed in order to conduct an online channel expansion. However, the purpose of the study was to study actions taken by participating SMEs. Further research could be aimed at testing proposed e-transit business model configuration in a larger setting and with a different methodological approach. Another interesting area could be to further study how value-creating activities change over time. Acknowledgements The authors gratefully acknowledge the financial support for the research from Peter Wallenbergs Foundation. 334 John Jeansson et al. References Abebe, Michael. (2014). Electronic commerce adoption, entrepreneurial orientation and small- aand medium-sized enterprise (SME) performance. Journal of Small Business Management, 21(1), 100-116. Avery, Jill, Steenburgh, Thomas J., Deighton, John, & Caravella, Mary. (2012). Adding Bricks to Clicks: Predicting the Patterns of Cross-Channel Elasticities Over Time. Journal of Marketing, 76(3), 96-111. doi: 10.1509/jm.09.0081 Chatterjee, Sayan. (2013). Simple Rules for Designing Business Models. California Management Review, 55(2), 97-124. Chen, Jing Elaine, Pan, Shan L., & Ouyang, Tao Hua. (2014). Routine reconfiguration in traditional companies’ e-commerce strategy implementation: A trajectory perspective. Information & Management, 51(2), 270-282. doi: http://dx.doi.org/10.1016/j.im.2013.11.008 Creswell, John. (2007). Qualitative Inquiry and Research design, choosing among five traditions (2 ed.). Thousand oaks: Sage Publications. Ecommerce-Europe. (2014). European B2C E-commerce Report 2014. Retrieved 2014/12/10, 2014, from http://www.ecommerce-europe.eu/ El Sawy, Omar A., & Pereira, Francis. (2013). Business modelling in the dynamic digital space, an ecosystem approach. New York: Springer. EuropeanCommission. (2014a, 11/07/2014). Facts and figures: SME definition. Enterprise and Industry. Retrieved 9/10/2014, 2014 EuropeanCommission. (2014b). SME Performance Review. Enterprise and Industry. Retrieved 21/10/2014, 2014 Ghobadian, Abby, & Gallear, David. (1997). TQM and organization size International Journal of Operations, 17(2), 121-163. Gunzel, Franziska, & Holm, Anna B. (2013). One size does not fit all: understanding the front-end and back-end of business model innovation. International Journal of Innovation Management, 17(1), 1340002-1340001-1340002-1340034. doi: 10.1142/S1363919613400021 Ja-Shen, Chen, & Ching, Russel K. H. (2002). A proposed framework for transitioning to an e-business model Quarterly Journal of Electronic Commerce, 3(4), 375. Johnston, David, Wade, Michael, & McClean, Ron. (2007). Does e-Business matter to SMEs? A comparison of the financial impacts of internet business solutions on European and North American SMEs. Journal of Small Business Management, 45(3), 354-361. Karjaluoto, Heikki, & Huhtamäki, Maija. (2010). The Role of Electronic Channels in Micro-Sized Brick-and-Mortar Firms. Journal of Small Business & Entrepreneurship, 23(1), 17-38. Li, Xiaolin, Troutt, Marvin, Brandyberry, Alan, & Wang, Tuo. (2011). Decision factors for the adoption and continued use of online direct sales channels among SMEs. Journal of the Association for Information Systems, 12(1), 1-31. Lofland, John, Snow, David, Anderson, Leon, & Lofland, Lyn H. (2006). Analysing social settings. Belmont: Wadsworth, Cengage Learning. Merriam, B. Sherran. (2009). Qualitative Research, a guide to design and implementation. San Francisco: Jossey-Bass. Miles, Matthew, & Huberman, Micheal. (1994). Qualitative Data Analysis: An Expanded Sourcebook. (2 ed.). Thousand Oaks: Sage publications. 335 Understanding Online Channel Expansion in an SME Context Moingeon, Bertrand, & Lehmann-Ortega, Laurence. (2010). Creation and Implementation of a New Business Model: a Disarming Case Study. Management, 13(4), 266-297. Mustaffa, S., & Beaumont, N. (2004). The effect of electronic commerce on small Australian enterprises. Technovation, 24, 85-95. doi: 10.1016/S0166- 4972(02)00039-1 Osterwalder, A, & Pigneur, Y. (2010). Business model generation: John Wiley and Sons. Osterwalder, A, Pigneur, Y, & Tucci, Cristopher. (2005). Clarifying business models: origins, present, and future of the concept Communications of AIS, 16, 1-25. Patton, Michael Quinn. (2001). Qualitative Research & Evaluation Methods (3 ed.). Thousand Oaks: Sage Publications. Postnord. (2014). E-commerce in the Nordics 2014. Retrieved 2014/12/10, 2014, from http://www.postnord.com/en/media/publications/e-handel/ Schneider, Sabrina, & Spieth, Patrick. (2013). Business model innovation: towards an integrated future research agenda. International Journal of Innovation Management, 17(1), 1340001-1340001-1340001-1340034. doi: 10.1142/S136391961340001X Stockdale, Rosemary, & Standing, Craig. (2004). Benefits and barriers of electronic marketplace participation: an SME perspective. The Journal of Enterprise Information Management, 17(4), 301-311. Thomas, Jacquelyn S., & Sullivan, Ursula Y. (2005). Managing Marketing Communications with Multichannel Customers. Journal of Marketing, 69(4), 239-251. doi: 10.1509/jmkg.2005.69.4.239 Tiwari, Rajnish, Buse, Stephen, & Herstatt, Cornelius. (2006). From electronic to mobile commerce, Opportunities through technology convergence for business services. TECH MONITOR(Sept-Oct), 38-45. Turban, Efraim, King, David, Viehland, Dennis, & Lee, Jae. (2006). Electronic Commerce, a managerial perspective 2006. Upper Saddle River: Pearson Prentice Hall. Valos, Michael J. (2009). Structure, people and process challenges of multichannel marketing: Insights from marketers. Journal of Database Marketing & Customer Strategy Management, 16(3), 197-206. doi: 10.1057/dbm.2009.21 Vlachos, Ilias. (2011). SMEs E-Business Behaviour: A Demographics and Strategic Analysis. Journal of Enterprise Resource Planning Studies, 2011, 1-21. Wagner, Gerhard, Schramm-Klein, Hanna, & Steinmann, Sascha. (2013). Effects of cross-channel synergies and complementarity in a multichannel e-commerce system – an investigation of the interrelation of e-commerce, m-commerce and IETV-commerce. International Review of Retail, Distribution & Consumer Research, 23(5), 571-581. doi: 10.1080/09593969.2013.832697 Weill, Peter, & Vitale, Michael. (2001). Place to Space, migrating to e-business models. Boston: Harvard Business School Press. Wilson, Hugh, Daniel, Elisabeth, & Davies, Ian (2008). The diffusion of e-commerce in UK SMEs. Journal of Marketing Management, 24(5-6), 489-516. Wong, Kuan Yew, & Aspinwall-Roberts, Elaine. (2004). Characterizing knowledge management in the small business environment. Journal of knowledge management, 8(3), 44-61. 336 John Jeansson et al. Zach, Ondrej, Munkvold, Bjorn Erik, & Olsen, Dag Håkon. (2014). ERP system implementation in SMEs: exploring the influences of the SME context. Enterprise Information Systems, 8(2), 309-335. Zhang, Jie, Farris, Paul W., Irvin, John W., Kushwaha, Tarun, Steenburgh, Thomas J., & Weitz, Barton A. (2010). Crafting Integrated Multichannel Retailing Strategies. Journal of Interactive Marketing, 24(2), 168-180. doi: http://dx.doi.org/10.1016/j.intmar.2010.02.002 Zott, Christoph, & Amit, Raphael. (2013). The business model: A theoretically anchored robust construct for strategic analysis. Strategic Organization, 11(4), 403-411. Zott, Christoph, Amit, Raphael, & Massa, Lorenzo. (2011). The Business Model: Recent Developments and Future Research. JOURNAL OF MANAGEMENT, 37(4), 1019-1042. 337 BACK 28th Bled eConference #eWellBeing June 7-10, 2015; Bled, Slovenia Integration of machine learning insights into organizational learning A case of B2B sales forecasting Marko Bohanec Salvirt d.o.o., Slovenia Marko.Bohanec@salvirt.com Mirjana Kljajić Borštnar University of Maribor, Faculty of Organizational Sciences, Slovenia Mirjana.Kljajic@fov.uni-mb.si Marko Robnik - Šikonja University of Ljubljana, Faculty of Computer and Information Science, Slovenia Marko.RobnikSikonja@fri.uni-lj.si Abstract Business to Business (B2B) sales forecast can be described as a decision-making process, which is based on past data (internal and external), formalized rules, subjective judgment, and tacit organizational knowledge. Its consequences are measured in profit and loss. The research focus of this paper is aimed to narrow the gap between planned and realized performance, introducing a novel model based on machine learning techniques. Preliminary results of machine learning model performance are presented, with focus on distilled visualizations that create powerful, yet human comprehensible and actionable insights, enabling positive climate for reflection and contributing to continuous organizational learning. Keywords: B2B sales modeling, machine learning, visual data mining, organizational learning, forecasting error reduction, knowledge engineering. 1. Introduction Organizational ability to grasp knowledge and transform it into continuous learning curve is a multifaceted problem (Gronhaug and Stone, 2012; Kljajić Borštnar, Kljajić, Škraba, Kofljač and Rajkovič, 2011). Learning ability is needed to preserve capacity to adapt to changes in the environment in order to achieve organizational goals and purpose. Learning curve increases from ignorance and converges towards full understanding while approaching the goal (Kljajić Borštnar et al., 2011). This logic is rooted in Locke’s philosophy (1924) 338 postulating that knowledge of the world can only be gained by experience (and its generalization by reflection). The learning is characterized by the change of behavior as a result of an individual and/or group exposure to experience (Kljajić Borštnar et al . , 2011). Two types of learning are distinguished: the single-loop and the double-loop learning (Argyris, 1996; DiBella and Nevis, 1998; Gephart, Marsick, Mark, VanBuren and Spiro, 1996, Nonaka and Takeuchi, 1995). The double-loop learning refers to not just changing the behavior in order to achieve the stated goal (single loop), but changing mental models, visions and beliefs, and therefore organizational knowledge. With the proposed approach we build a foundation to achieve the double-loop learning – as a basis to establish new premises (i.e. paradigms, schemes, mental models or perspectives), with potential to override existing ones (Nonaka and Takeuchi, 1995). Same authors are fully aware that an effort to question and rebuild existing perspectives, interpretation of frameworks or decision premises can be very difficult to implement in an organization; it requires persistent activities. Organizational learning presents ongoing effort of creating organizational knowledge. Team learning, personal mastery and mental models principles (Senge, 1990) are built-in into organizational knowledge. In this paper we propose a classification model, which builds on insights from B2B sales professionals. Insights are presented in a form of sales history described with features reflecting attributes of sales process and B2B relationships (Bohanec et al., 2015). Machine learning techniques are applied to build the classification model, which is capable to classify future, unseen sales opportunities. The classification model represents the organizational knowledge which is presented and visualized in a human comprehensible form to support the double-loop learning process within an organization. Our aim is to investigate whether it is possible to develop such a model, based on B2B sales history, which supports process of forecasting and transparent reasoning. 2. Literature review and methodology Comprehensive research by Ngai, Xiu and Chau (2009) reveals that application of data mining techniques is widespread in the field of Customer Relationship management (CRM). Applicants are benefiting from customer data and past purchase behavior. However, in specific field of B2B sales forecasting, which is influenced by dynamics of markets, loosely structured information and possible noise in data, lack of academic approach and modeling is obvious (Monat, 2011). As Monat concludes, a final step of making the forecast is left to a decision maker. Despite improved statistical and organizational learning capabilities, Rieg (2010) identified environmental uncertainty as a significant reason why there is no evidence about increased forecasting performance. Different approaches and solutions are proposed as forecasting support systems (FSS), however it was reported that they are not delivering on their promise, mostly pointing out low trust in FSS recommendations and suggesting improvements in explanations, work on better past perception of FSS systems and more comprehensible format of information delivery (Alvarado-Valencia and Barrero, 2014). To represent the B2B sales domain knowledge is an important first step in order to build a learning data set. Following Monat (2011) conclusions, an overview of attributes (features) 339 was compiled using research articles from academic databases and sales professionals’ additions (Bohanec, Kljajić Borštnar and Robnik-Šikonja, 2015). A selection of attributes, which sales experts can reflect upon for cases from their sales history, defines “descriptive language” of specific sales organization. This creates a foundation to describe sales context of both successful and failed sales opportunities. To secure a high quality learning data set, literature is outlining different data preparation techniques, e.g. focus on outlier’s detection, data normalization, handling of missing data, noise detection and noise reduction, feature enhancements, data reduction or generation etc. A decision which techniques to use is highly dependent on a particular problem. The approach is guided by a process of data preparation based on insights, created knowledge in the process of building learning data set, and selected machine learning algorithms (Maaß, Spruit and Waal, 2014). Therefore builders of data set need to pay an extra attention to secure high data quality and enable good performance of machine learning techniques. Lack of attention to data quality can possibly lead to “Garbage In, Garbage Out” problem. Machine learning (ML) in our context is interpreted as an acquisition of structural descriptions from examples (Witten, Eibe and Hall, 2011). The fact that it leverages different models and algorithms to approximate complex theories which are difficult to be exactly represented with other mathematical tools, connects it to the field of artificial intelligence. ML has been successfully applied in different fields, e.g. medical diagnostics, spam filtering, OCR, internet browsers etc. (Liao, Chu and Hsiao, 2012; Ngai et al., 2009; Bose and Mahapatra, 2001). ML techniques take training data set to learn relationships needed to categorize new, yet unseen, objects to target categories (Witten et al., 2011; Robnik-Šikonja and Kononenko, 2008). Some classification models produced are able to explain their decisions, which can help in better adoption of ML techniques in practice due to participant’s faster understanding of ML insights (Robnik-Šikonja and Kononenko, 2008; Collopy, Adya and Armstrong, 2006). 2.1. Methodology Our research methodology is best described as the action design research (Sein, Henfridsson, Purao, Rossi and Lindgreen, 2011). Selected method supports our goal to create an IT artifact within the context of organization both in development and use phase. The nature of organizational learning requires continuous process to maintain organizational agility as a response to internal and external market dynamics. In Figure 1 we present the research framework, combining machine learning methods for model building and introduction of extracted knowledge to the forecasting learning loop. Each cycle of forecasting is evaluated (forecasts, supported by the classification model, are compared to actual results) and the feedback is used twice: first in the machine learning model and second in a decision maker. The process is iterative and each plan-act-reflect cycle is done in a natural setting of the organization. 340 Figure 1: The research framework The proposed approach was presented to selected companies participating in the research. They expressed the interest and some shared their CRM data for an initial review. Practical experience shows that this approach is quite challenging, because it seems that typical CRM implementations are missing sales opportunity’s attributes reflecting relationship dynamics, individual and organizational attributes (Bohanec et al., 2015) and thus a context of particular case is not described in a suitable way for machine learning techniques to perform adequately as the existing CRM attributes have low information and prediction value. CRM systems should therefore be updated from technical point of view and sellers shall adapt their process of information collection. This created new level of complexity and effectively stalled participation of companies. Therefore we leveraged anonymized sales history from the company founding the research. We applied R package semiArtificial (Robnik-Šikonja, 2014) to generate sufficient number of instances to allow modeling and development. A full CRISP-DM process (Chapman, Clinton, Kerber, Khabaza, Reinartz, 2000) was followed in combination with visual data mining as portrayed in Figure 2. Such approach supports well the modeling presented in this paper, and can serve as a reference for inclusion of new external organizations with their B2B sales historical data in the future. Figure 2: Visual data mining process - Simoff et al. (2008) 341 At the beginning we have a roughly defined set of real-world sales opportunities historical data with their outcomes, minus sign representing failure to close the deal and plus sign reflecting successful closure of it (Figure 3). This represents the starting point where business understanding (thru expert’s reflection) needs to be demonstrated. As a next step, a subset of most influential features needs to be defined, which at the beginning is determined by the judgment of sales team. Starting point could be (but not limited to) a list of features proposed in Bohanec et al . (2015). Based on the selection of features, real world cases reflecting sales history need to be described with values of these features. In Figure 3, table view reflects selection of N features and M cases, with different type of values to illustrate versatility of the approach. Figure 3: Real-world training cases transformation to learning data set. Frequently the question about the number of cases needed arises. According to Shmuelli, Patel and Bruce (2007) some authors recommend as a rule of thumb ten records for every feature. They are also citing Delmar and Hancock, which propose the number of cases equal to 6 * m * p, where p represents the number of selected features and m represents the number of class values (two in our case). In case of p=10 and m=2, we would therefore need 120 past cases, which is practically possible, provided several experienced sellers participate in the construction of training data set. It is important to note iterative nature of this process. In case ML ranking and classification techniques will not immediately perform at the expected level, we should “go back” and select additional attributes, not yet part of the subset. In this phase our goal could be to add new attributes, while keeping existing ones. We might reconsider values of particular attributes with the help and argumentation of participating sales experts. Our aim is to build a compact, comprehensible model, intended for use by sales force and sales management as a source of insight. Therefore the final selection of features should be limited to a number, which can be cognitively handled. According to Miller’s (1956) recommendation, this would be 7 ± 2 features. In this way the complexity of the model is still cognitively feasible, and overfitting of data set is unlikely. However, the parsimony of the model has to be balanced with model’s performance on real world cases, to prevent excessive elimination of attributes at the expense of model accuracy. Machine learning techniques need sufficient information to expose valuable information. An efficient way to achieve this goal is by ranking features. 2.2. Ranking features Ranking features helps us to identify which features are the most important in the training data set, by ranking them according to how informative they are. To estimate feature 342 significance several different scoring techniques are available, some of them are used in Table 1. We use Orange data mining suite (Demšar and Zupan, 2013) for majority of ML techniques applied in this paper. Attribute ReliefF Inf. gain Gain Ratio Gini RF Negotiations 0,961 0,819 0,523 0,215 2,078 Reaction 0,913 0,848 0,332 0,223 0,085 Prospect_authority 0,900 0,802 0,802 0,213 11,717 S_A_Pilot 0,822 0,830 0,362 0,220 0,876 Need_defined 0,783 0,775 0,775 0,207 17,405 Product 0,490 0,324 0,137 0,099 0,000 Client_growth 0,323 0,098 0,056 0,032 1,007 Other_solution 0,308 0,110 0,122 0,037 1,041 Source 0,246 0,226 0,156 0,072 0,137 Owned 0,233 0,183 0,110 0,055 0,055 Budget_limits 0,226 0,020 0,018 0,007 0,400 Existing_client 0,146 0,174 0,305 0,047 1,169 Familiary_wVendor 0,132 0,030 0,027 0,009 0,125 External_svcs 0,106 0,026 0,060 0,009 0,191 Competitors -0,005 0,034 0,089 0,010 0,232 Deal_size -0,013 0,083 0,041 0,028 0,437 Table 1: Features evaluation – some different techniques from Orange suite. A scoring technique ReliefF measures attribute’s ability to detect conditional dependencies between attributes and provides a unified view on the attribute estimation in regression and classification. In addition, its attribute importance estimates have natural interpretation (Robnik-Šikonja and Kononenko, 2008). The Inf. gain is measuring information entropy of attributes conditioned upon class with a downside that this measure prefers features with more values. Gain Ratio prevents this by normalization with attribute entropy (Quinlan, 1993). Similarly, Gini index estimates purity of conditional class values split by the values of attribute. RF represents Random Forest, an ensemble learning technique, which can also output attribute importance score estimated on internal set of instances. Random forest method is fast, robust to noise, does not overfit and offers possibilities for explanation and visualization of its output (Breiman, 2001). In our case, looking across all scoring approaches, a general observation from B2B sales domain perspective is that there are approximately 5 informative features out of 16, namely Negotiations, Reaction, Prospect authority, S_A_Pilot (how easy was to get a pilot of a solution) and Need defined (by a buyer). The rest of the features show low importance, except for features Source and Existing Client for which some methods indicate contribution. We present insights into analysis and topics of discussion with a team of experts when modeling the description of sales history. On one side it is expected that some features are considered very important; however, it is rather surprising that others produce so low estimates. For example, in presented case it looks that familiarity with vendor (seller’s company), Deal size and competition have very little relevance. These scores shall encourage sales experts to thoroughly analyze each surprising result and eventually update learning data set with additional cases, providing values for missing values or introducing different features into the training data set. Alternatively, they can accept the insight as a learning 343 opportunity and update their beliefs and mental models. Continuous movement back and forth between phases Analytical Reasoning and Data preparation in Figure 2 is therefore required. McCarthy Bryne, Moon and Mentzer (2011) have shown in their research that inclusion of sales professionals is important; however they need to be included in such a way to maintain a positive attitude towards the learning potential of the model in the light of the forecasting task. When sales experts agree that the model seems to be correct and no obvious or hidden flaws are identified, we can proceed to the next step and verify how well machine learning techniques are capable to learn based on the produced data set. The training data set is considered stable from the point of view of accepted features, number of missing values, and number of provided training instances. 2.3. Building and testing machine learning model Based on prepared training data set different machine learning techniques can be utilized to build an automated reasoning system. Results of learning are presented in Table 2 and show that three selected classification techniques are performing quite well, as they are exceeding 70% classification accuracy (CA), which can be considered as good performance taking into account difficulty of this real-world problem. The performance estimates were computed using “Leave-one-out” cross-validation (Elisseeff and Pontil, 2003). AUC stands for Area Under the ROC curve, which is a standard machine learning assessment measure giving information about quality of the produced classification probabilities (Witten et al., 2011). Method CA AUC Random Forest 0,9624 0,9934 Naive Bayes 0,9474 0,9764 Classification Tree 0,9398 0,9315 Table 2: Performance of some classification techniques (classification accuracy (CA) and area under the ROC curve (AUC)), produced by Orange system. Random Forest technique is performing best with the CA of 96%. The case study training set has 133 examples, 73 classified as NO (lost deal), and 60 classified as YES (won deal). Additional analysis of Confusion matrix reveals that 5 opportunities which should be classified into the class NO, were incorrectly classified as YES. For the class YES all cases were classified correctly. 2.4. Testing machine learning model with a limited set of features As evident from Table 1, top ranked attributes represent less than a half of all features. We indicated that we would prefer to work with less attributes to maintain simplicity and parsimony of a model in a balance. The question is how ML performance will be affected if selected techniques would use only top 7 attributes (as ranked by ReliefF). In Table 3 results of this limited feature set are presented. Comparison of CA results in Table 2 and Table 3 reveals that Random Forest and Classification tree were not affected and (surprisingly) Naïve Bayes has even slightly improved its score compared to Table 2, indicating some noise in the excluded features. 344 Method CA AUC Random Forest 0,9624 0,9852 Naive Bayes 0,9549 0,9588 Classification Tree 0,9398 0,9311 Table 3: Evaluation using only top 7 features (from Table 1, as ranked by ReliefF). 3. Results - representation of B2B sales knowledge In this section we apply different ML techniques and visualizations to emphasize the insights and to create an input to the double-loop learning within B2B forecasting task. The results are preliminary and are based on conceptual model building, its evaluation and validation. Sales domain interpretations are related to the case study presented in this paper; for a different organization they could be completely different, reflecting their training data set. 3.1. Sieve Multigram projection Sieve Multigram shows how features are correlated. Red color indicates negative correlation and blue color indicates positive correlation. Thickness of lines indicates how strong the correlation is. For example, from Figure 4 we can see that value Yes for the feature Need defined is negatively correlated to the value Mid for the feature Prospect authority, however it is positively correlated to the value High of the same feature. There are some other interesting relationships revealed, for example that client stability (value Stable) and presence of competitors are positively correlated, meaning stable organizations look for offers from more vendors, compared to organizations in transition. Figure 4: Sieve multigram - correlations among selected features (Orange). 345 3.2. Generating association rules Using RStudio (www.rstudio.com) and R library arules, we created a set of association rules (Table 4). For each rule we present three standard evaluation metrics: support, confidence and lift. The support measures the proportion of transactions in the training data set which contain the lhs items (left side of a rule). The confidence reflects the proportion of cases satisfying rule preconditions (i.e., lhs), that also satisfy rule consequences, i.e., rhs (right side of a rule). The lift reports the ratio of the observed support to that expected lhs and rhs were independent (Witten et al . , 2011). To illustrate: the first rule from Table 4 reveals that buyer’s need accompanied by high prospect authority occurs in 45% of all cases in training data set (therefore 45% support), and when this is true, it yields a 95% confidence for this rule to lead to a contract signature. The lift value of 2.1 reports that support of a rule “lhs implies rhs” (i.e. its confidence) is 2.1 times more likely than support of rhs and lhs estimated independently in the whole population. This (shortened) list of association rules is transparently reveling “preconditions” for sales opportunity to be closed or not in the next observational period. Transparent representation of rules creates a solid foundation for an individual or group review of forecasts. lhs rhs support confidence lift 1 {Need_defined=Yes, Prospect_authority=High} => {Signed=YES} 0.4511278 0.9523810 2.111111 2 {Competitors=No, Prospect_authority=High} => {Signed=YES} 0.4436090 0.9516129 2.109409 3 {Need_defined=Yes, Competitors=No, Prospect_authority=High} => {Signed=YES} 0.4436090 0.9516129 2.109409 4 {External_svcs=Yes, Need_defined=Yes, Prospect_authority=High} => {Signed=YES} 0.3834586 0.9444444 2.093519 a {Need_defined=Info_Gathering, Prospect_authority=Mid, Negotiations=Not_started} => {Signed=NO} 0.4812030 1.0000000 1.821918 b {Existing_client=No, Need_defined=Info_Gathering, Negotiations=Not_started} => {Signed=NO} 0.4962406 1.0000000 1.821918 c {External_svcs=Yes, Need_defined=Info_Gathering, Negotiations=Not_started} => {Signed=NO} 0.4736842 1.0000000 1.821918 d {Existing_client=No, Need_defined=Info_Gathering, Prospect_authority=Mid} => {Signed=NO} 0.4887218 1.0000000 1.821918 Table 4: Association rules for positive (YES) and negative (NO) outcome (RStudio, library arules). 3.3. Classification tree Based on the case study training data set, classification tree presented in Figure 5 was built. It reveals an importance of the attribute Negotiations, ranked highest by the ReliefF algorithm. First insight is that it’s good to be in some kind of negotiations; when negotiations did “not_started” yet, no deal was closed (quite obvious, though). Second insight reveals that success is high when moderate negotiations take place and there is a possibility of some other solution. Here it’s important to make a distinction between competing with other providers and competing with other solutions. In the context of our case study, other solution could be an old solution, which needs to be replaced, or a manual solution, which 346 needs to be upgraded; therefore a client is not really comparing different external providers, competing for the business but investigating different alternatives within our portfolio. Figure 5: Decision tree with ReliefF attribute selection criterion (Orange). Next insight reveals that when sellers are involved in moderate negotiations and they offer a solution to the existing client, they always win. This insight reflects organizational strength for cross selling and could be recognized as one of key sales approaches for future growth. However, when sellers offer the same solution to a new prospect, they always fail. This indicates the importance of the client relationship for the analyzed case study. 3.4. Parallel Coordinates – Visualization When dealing with multiple features, a parallel view of training data is presented in Figure . Each training instance is presented with one connected line and we can see frequently occurring patterns. The whole graph reveals a “broader story” of what is working and which scenarios should be avoided. Blue lines represent won deals and red represent lost deals. For our case study, we can learn that selling to existing clients who have some level of dynamics from growth perspective, with their business needs clearly expressed by a person with high authority to secure the budget and who is eager to start negotiations with us only (no competitors), creates a great likelihood for a success. Sellers are used to storytelling practice and such interpretation is compelling to them to position their sales into the context of what need (or can) be changed to fit into the “story”. For different training data set, a different story could emerge. 347 Figure 6: Parallel coordinates (Orange). 3.5. Scatter plot In comparison to parallel coordinates the scatter plot deals with fewer interacting features; however those are compared against each other and thus create new perspective. Figure 7 shows the relationship between ability to secure pilot (or trial) testing of offered solution and a prospect’s authority to execute the deal. One insight from this visualization is prevailing – sellers should work hard to secure a person with high authority on the other side of a table, who needs to be keen to try the solution. Other scenarios do not close the deals. Figure 7: Scatter plot (Orange). 348 Such visualization is important, as it helps to quickly assess new sales opportunities; if an opportunity fails this “test”, it is safe to go for “No” in the forecasting task. If a positive answer is secured in the “test”, the opportunity needs to be further analyzed before a conclusion about the forecast is reached, taking into account different models and visualizations. For example, a different scatter plot could be produced comparing familiarity with vendor with budget limits, enriched with what was offered to the prospect ( Product) and the size of the used shape defined with the relative size (Deal size) of an offer. From this perspective our case study offers mixed results. It might indicate low flexibility on price or other client’s priorities have higher budgetary attention – the case has to be further researched with sales professionals. Comparing insights in Figure 4 to Figure 7, we can see that different insights were presented in a transparent way, revealing the important knowledge for development of broader understanding of sales dynamics. These different views contribute to and stimulate strong, facts based dialog and assessment of sales opportunities and create an improved understanding needed for a more realistic B2B forecasting, with a goal to decrease the forecasting error. 5. Discussion and Conclusions In this paper a novel approach towards organizational learning using machine learning and specific domain knowledge from the field of B2B forecasting is proposed. Our goal is to develop a model supporting the decision-maker in the process of forecasting by transparent reasoning, based on the real-world data. Insights provided by the ML model must be presented in a comprehensible way, enabling decision-makers to reflect on in the double- loop learning and so to update their sales knowledge. New knowledge should give direction to adapt or change behaviors thereby create new sales opportunities and thus improve decision-makers ability to decrease their forecasting error. We applied transparent ML techniques capable to explain their recommendations, which is essential for the double-loop learning. The training data set consisting of 133 unique cases was built from anonymized sales history of a company providing services to their clients. Initial list of 16 features (attributes), describing sales history, was assessed by 5 different ML attribute evaluation techniques. Based on the results we selected 7 out of 16 features which contain important information for the development of the prediction model. Using several feature evaluation techniques we reduced the probability of biased and subjective prioritization, frequently observed when working with individuals. Several visualization techniques were evaluated within Orange and RStudio data mining suites but only a small subset of interesting visualizations is presented in this paper. For a given case study, selected ML prediction models achieved high classification accuracy. Altogether the presented methods reveal several important insights enabling modification of seller’s behavior and challenge their traditional intuitive forecasts of new opportunities by comparing them with insights of ML techniques. The goal of this paper was to investigate the possibility to develop a classification model, based on B2B sales history, which supports forecasting process and provides transparent 349 reasoning, supported by machine learning techniques. We are convinced that the research presented in this paper has positively confirmed viability of the novel approach. To further develop the model, we need additional companies from different industries to participate with their data. However, getting well-structured tabulated data with attributes describing B2B sales opportunities presents the biggest challenge, mainly because of the fact that currently companies and sales personnel put no attention to specific attributes, needed to create domain knowledge reflection framework. Further research is also needed on the machine learning process: number of cases needed to create acceptable level of forecasting accuracy, effect of adding new attributes to training data set, statistical quantification of predictive confidence etc. The work presented in this paper establishes a basis for a further research in domains other than sales and stimulates a critical discussion about selection of powerful and insightful visualizations used with machine learning and data mining techniques. This will increase trust in forecasts and so bridge the gap between users and technology. 6. Acknowledgement We are grateful to the company Salvirt ltd. for funding the research and development of the model, presented in this paper. 7. References Argyris C. (1996): “On Organizational Learning”, Blackwell, New York. Alvarado-Valencia J.A., Barrero L. H. (2014): “Reliance, Trust and Heuristics in Judgmental forecasting”, Computers in Human Behavior, Vol. 36, pp. 102-113. Bohanec M. (2014a): “Modeling knowledge for reducing opportunity based forecasting error in B2B scenario with help of machine learning methods”, Proceedings of 33rd Conference of Organizational science developments, Portorož, Slovenia. Bohanec M. (2014b): “Reducing sales forecast error by leveraging machine learning techniques for B2B opportunity-based forecasting”, Proceedings of 27th Bled eConference, Bled, Slovenia. Bohanec M., Kljajić Borštnar M., Robnik-Šikonja M. (2015); “Modeling attributes for forecasting B2B opportunities acquisition”, Proceedings of 34th Conference of Organizational science development, Portorož, Slovenia. Bose I., Mahapatra R.K. (2001): “Business data mining - a machine learning perspective”, Information & Management, Vol. 39, No.3, pp. 211-225. Bradley A. P. (1997): “The use of the area under the ROC curve in the evaluation of machine learning algorithms”, Pattern Recognition, Vol. 30, No. 7, pp. 1145-1159. Breiman L. (2001): “Random forests”, Machine Learning Journal, Vol. 45, pp. 5–32. 350 Collopy F., Adya M., Armstrong J.S. (2001): “Expert Systems for Forecasting”, In Principles of Forecasting: A Handbook for Researchers and Practitioner, Kluwer. Chapman P., Clinton J., Kerber R., Khabaza T., Reinartz T., Shearer C., Wirth R. (2000): CRISP- DM 1.0 Step-by-step data mining guides” , ftp://ftp.software.ibm.com/software/analytics/spss/ support/Modeler/ Documentation/ 4/UserManual/CRISP-DM.pdf Demšar J., Zupan B. (2013): “Orange: Data Mining Fruitful and Fun – a Historical Perspective”, Informatica, Vol. 37, pp. 55–60. http://orange.biolab.si DiBella AJ, Nevis EC. (1998): “How Organizations Learn”, Jossey‐Bass, San Francisco, USA. Elisseeff A., Pontil M. (2003): “Leave-one-out error and stability of learning algorithms with applications”, Advances in Learning Theory: Methods, Models and Applications, J.A.K. Suykens, ed., IOS Press, 2003. Gephart MA, Marsick VJ, Mark E, VanBuren ME, Spiro MS. (1996): “Learning organizations come alive”, Training Development, Vol.50, No.12, pp. 36-41. Gronhaug K., Stone R. (2012):”The learning organization”, Competiveness review: An international business journal, Vol. 22, No. 3. Katz D.M., Bommarito M.J., Blackman J. (2014): “Predicting the Behavior of the Supreme Court of the United States: A General Approach”, Social Science Research Network, http://ssrn.com/abstract=2463244 (27.2.2015). Kljajić Borštnar, M., Kljajić, M., Škraba, A., Kofjač, D., Rajkovič, V. (2011): “The relevance of facilitation in group decision making supported by a simulation model”, System dynamics review, ISSN 0883-7066, vol. 27, no. 3, pp. 270-293. Liao SH., Chu PH., Hsiao PY. (2012): “Data mining techniques and applications – a decade review from 2000 to 2011”, Expert systems with Applications, Vol. 39. Locke J. (1924): “An Essay Concerning Human Understanding (Slovenian translation)”, Slovenska šolska matica, Ljubljana. Quinlan, J.R. (1993): “C4.5: Programs for Machine Learning”, Morgan Kaufmann, San Francisco. Rieg, R. (2010): “Do forecast improve over time?”, International Journal of Accounting and Information Management, Vol. 18, no. 3. Robnik-Šikonja M. (2014): “semiArtificial: Generator of semi-artificial data”, R package version 1.2.0. http://cran.r-project.org/package=semiArtificial Robnik-Šikonja M., Kononenko I. (2008): “Explaining Classification for Individual Instances”, IEEE Transactions on Knowledge and Data Engineering, Vol. 20, no. 5, pp. 589-600. Maaß, D., Spruit, M., Waal, P. D. (2014): “Improving short-term demand forecasting for short-lifecycle consumer products with data mining techniques”, Decision Analytics, Springer, pp. 1–17. 351 McCarthy Byrne T.M., Moon M.A., Mentzer J.T. (2011): “Motivating the industrial sales force in the sales forecasting process”, Industrial Marketing Management, Vol. 40, pp. 128-138. Miller, G.A. (1956): “The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information”, Psychological Review, Vol. 63, pp. 81-97. Monat, J. P. (2011): “Industrial sales lead conversion modeling”, Marketing Intelligence & Planning, Vol. 29, Iss: 2, pp.178 – 194. Ngai E.W.T., Xiu Li, Chau D.C.K. (2009): “Application of data mining techniques in CRM: a literature review and classification”, Expert Systems with Applications, Vol. 36, pp. 2592– 2602. Nonaka I., Takeuchi H. (1995): “The knowledge creating organization”, Oxford University Press, New York. Senge P. (1990): “The Fifth discipline: The Art & Practice of the Learning Organization”, Doubleday Currency, New York. Sein M.K., Henfridsson O., Purao S., Rossi M., Lindgreen R. (2011): “Action Design Research”, MIS Quarterly, Vol 35, pp. 37-56. Shmueli G., Patel N.R., Bruce P.C. (2007): “Data Mining for Business Intelligence: Concepts, Techniques, and Applications in Microsoft Office Excel with XLMiner “, John Wiley & Sons. Simoff S. J., Böhlen M. H., Mazeika A. (eds.) (2008): “Visual data mining“, Lecture Notes in Computer Science, Springer. Witten, I.H., Eibe F., Hall M.A. (2011): “Data mining – Practical Machine Learning Tools and Techniques”, third edition, Elsevier. 352 BACK 28th Bled eConference #eWellBeing June 7 -10, 2015; Bled, Slovenia Ontology based Multi-Agent System for the Handicraft Domain E-Bartering Rahma Dhaouadi SOIE, Institut Supérieur de Gestion, Université de Tunis, Tunisia rahma.dhaouadi@fsegn.rnu.tn Kais Ben Salah SOIE, Institut Supérieur de Gestion, Université de Tunis, Tunisia Kaisuria@yahoo.fr Achraf Ben Miled SOIE, Institut Supérieur de Gestion, Université de Tunis, Tunisia achraf.benmiled@gmail.com Khaled Ghedira SOIE, Institut Supérieur de Gestion, Université de Tunis, Tunisia Khaled.ghedira@isg.rnu.tn Abstract The online supply requirement management within a heterogonous environment represents a real challenge. While the traditional e-purchase is widely adopted, the e-barter seems to be an ambitious alternative. It is mainly solicited when suppliers might be unavailable or the delivery timeouts are important. Moreover, it reinforces the communication between the producer and his professional network. In this paper, we propose an emulation of the handicraft women e- procurement process based on the power of multi-agent paradigm and ontology formalism. Indeed, we establish an e-barter framework which targets to recommend in real time, under different circumstances and regarding the handicraft woman situation the suitable exchange partners. Likewise, we established several producing rules in order to deduce automatically the best sourcing moment. Furthermore, the handicraft woman which is the decision maker might drive an e-barter auction (e-BA) process in order to choose the best exchange opportunities and 353 Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled, Khaled Ghedira then minimize her expenses. We consider the e-BA as a new concept merging the barter and auction notions. Keywords: E-procurement, E-barter, E-auction, multi-agent system, recommender system, handicraft domain, business ontology 1 Introduction HanDicraft Women (HDWs) within developing countries constitute an important workforce. However, they encounter different issues during their business activities. This paper is project- driven which targets to improve the HDWs socio-eonomic level dealing with their heterogeneous environment and different contexts. Indeed, they can be expert or beginner, illiterate or cultivated, autonomous or dependant, integrated or isolated etc. To handle these issues, we should understand HDWs needs to better fit their requirement. The recommendation of suitable supply opportunities promotes significantly the producing process aspects. Dealing with highly demanding users such as HDWs within a heterogeneous and inconstant environment, leads us to look for suitable suppliers in order to meet their expectations. Different researches evoked the best suppliers selection within the context of traditional e-procurement such as (Lee et al. 2009; Wang et al. 2012; Seungsup Lee et al. 2013). We are based particularly on (Dhaouadi et al. 2014a), where the authors treated the supply chain agentification in order to automatically recommend suitable suppliers for the HDWs communities during their business activities. The e-barter alternative can improve the HDWs satisfaction mainly when different issues may arise such as the suppliers' unavailability, the far delivery deadlines, the rare and seasonal raw material nature. These circumstances may affect critically the stock state and then threaten the producing activities progress. Indeed, an under-storage and/or over-storage situations may emerge. The over-storage phenomenon occurs when the raw material quantity exceeds the recommended level. This situation causes probably a poor workshop space management and a raw material quality degradation over time. The under-storage means that a particular raw material is out of stock. If the raw material in question is hardly procured (rare or non available), the situation will be increasingly unfavourable for the producing tasks. Taking into account these issues, the raw material exchange (e-bartering) within the professional network will be hence very efficient. In this paper, we aim at providing consistent procurement opportunities via suitable exchange peers. The proposed e-barter system is considered as an extension of a previous proposed traditional e-commerce system in (Dhaouadi et al. 2014a). In fact, the barter is "an exchange of two items" (Küpçü and Lysyanskaya. 2012). According to our case, it consists in exchanging raw materials among handicraft network members. Similarly to (Núñez el al. 2005), we adopt the multi-agent paradigm so to model the e-barter system. The latter is considered as "A set of agents performing exchanges of products" (Núñez el al. 2005). The recommendation of exchange peers is not a trivial task. It takes into account the HanDicraft Woman (HDW) situation such as her locality. In fact two relevant barter peers have to reach an agreement about the raw material to exchange and have to be also situated in 354 Ontology based Multi-Agent System for the Handicraft Domain E-Bartering adjacent localities. Besides, in real world, the HDW, which is the barter initiator, may deals with different exchange alternatives where she has to opt for the best choice. In this paper we define a new concept "barter auction" (e-BA) merging the barter and the auction notions. Note that a classic auction operation consists in trading an item between a seller and different buyers. The seller is considered as the auction initiator and is expected to precise a starting price (Christidis et al. 2013). The buyers compete against each other in real time by proposing several bids. Each bid should be higher than the previous one. In our research work, we define the barter auction as an exchange operation between the different HDWs. It targets to facilitate the procurement process without the implication of money. The barter auction initiator (HDW) seeks to acquire an expected raw material and give back the less useful one. She is hence able to minimize her expenses by opting for the best exchange offer. The remainder of this paper is organized as follows: Section 2 represents related works. Then, section 3 introduces the proposed e-barter system. A case study is then undertaken in order to validate our proposal in section 4. Finally, we conclude this paper in section 5. 2 Related Works In the literature, different research works have undertaken multi-agent systems when dealing with the e-barter process. In (Cavalli & Maag. 2004), the authors present a formal specification of an e-barter system based on intelligent agents. The proposed formalization uses the utility function to represent customer preferences and considers the transaction and shipping costs. They also validated their proposal via the test generation method application which reduces the design ambiguities and errors. In (Haddawy et al. 2005), an emulator for trade brokers practices is developed. The proposed system aims at matching buyers and sellers. Besides it uses the greedy heuristic to reduce the sellers number affected to a buyer regarding a particular product. Their solution is modeled as an optimization problem which targets to maintain the trade balance. However, the authors assume that the barter entries are already existent. Effectively, the implemented algorithm deals with real exchange demands taken during a week. Barter data is structured as a requirements matrix where the rows represent the trade members, the columns represent the product nature and the matrix content represents quantities to buy or to sell. The commercial barter problem is considered hence as the minimum cost circulation problem within a network. In despite of the method efficiency, it handles static data picked in a previous period of time. Nevertheless, the online treatment of the barter exchange demands is interesting. Furthermore, the proposed system proposes many suppliers in order to meet one good need which is far to be feasible in real life. (Küpçü & Lysyanskaya 2012) provide an efficient optimistic p2p fair exchange mechanism for bartering digital files using cryptology primitives. The system is able to be used in real bartering applications with high competence level. Most of the time, it is functional without the implication of a third party (Arbiter). However, when a conflict arises, the implication of a trustworthy arbiter is recommended. In the other side, multiple papers handled the e-sourcing process based on the auction transactions. Effectively, there is a rising interest concerning the adoption of auction 355 Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled, Khaled Ghedira mechanism during the procurement process. In (Lin et al. 2011) the authors propose an agent- based price negotiation system for online auctions. Mainly, three agents are used in the study: seller agent, buyer agent and mediator agent. The proposed system helps sellers and buyers to personalize their price negotiation strategies based on the fuzzy rules. (Huang et al. 2013) present a hybrid mechanism for e-supply procedure including two phases. The first phase concerns a multi-attribute combinatorial auction. The second consists in bargaining with the auction winners. The system improves the transactional social surplus. In (Pla et al. 2014), an auction mechanism is introduced. It considers different attributes such as price, service time, quality tolerance in order to better fit the auctioneer expectations. In this research, a multi-agent system framework is proposed in order to facilitate the barter and auction transactions of raw materials within the professional handicraft networks. Indeed, it enables the complainant HDW firstly to search for suitable exchange alternatives and secondly to conduct an exchange auction if the proposed barter candidates are competitors. Our proposed system considers the benefits of existing scientific works in this regard. In fact, it recommends online and automatically the best trade alternatives. The recommendation is based on the exchange pairs preferences and profiles matching. Actually, the product nature in need and the geographic proximity are fundamental parameters during the barter opportunities suggestion. Hence, the selection process decreases the HDWs exchanged messages number (since the addressed community of interest is reduced) which is an important aspect in multi-agent paradigm. Likewise, we deal with several issues presented previously such as the necessity of a trustworthy arbiter which is no longer needed. In fact, the e-bartering pairs are communicating together without a mediator agent during the exchange and auction transactions. In order to maximize the HDW (barter initiator) benefit, a barter auction procedure is performed where she is able to select the best exchange offer. From another perspective, our system promotes the HDWs communication within the same professional network. Actually, the proposed system enables the HDWs to elaborate professional communities collaborating together. However, the system does not consider the shipping costs resulted from the exchange operation. It reduces only the delivery expenses since it takes into account the barter peers geographic adjacency. 3 E-bartering Within The Handicraft Domain During her business activities, the HDW has to explore different supply alternatives in order to acquire the desired raw materials. Usually, she looks for suitable suppliers providing the needed good with lowest price and highest quality. Besides, when it is necessary, she is allowed to exchange several products with another member from her professional network. The exchange target is to get the desired product and in return providing another. Once the required raw materials are available, the HDW deals with the production. She may collaborate with other HDW in the purpose of satisfying the customer command. As seen, the HDW environment is open, heterogeneous, dynamic and distributed. Thus, the multi-agent paradigm adoption looks to be appropriate to the problem modeling. Indeed, we handled the e-procurement process agentification (Figure 1). Here, the HDW receives the customer command (1). Then she checks her stock state through the Stock Notifier (2). If the necessary 356 Ontology based Multi-Agent System for the Handicraft Domain E-Bartering quantities are not yet available then the HDW might ask the Purchase and Barter agents (3) to search for the suitable supply opportunities. As decision maker, she is then able to select radically the convenient procurement alternative. Figure 1: The e-procurement agentification Our proposed system deals with HDW from different handicraft fields. It targets to recommend relevant supply opportunities in spite of the HDW context heterogeneity. That is why the specification of the handicraft domain via an ontology formalism is necessary. Moreover, the integration of the ontology enables agents to better represent knowledge (Wang et al. 2012). Likewise, it facilitates their interoperability and coordination (Thanh et al. 2004) and reinforces their confidence (Rosaci & Sarnè, 2014). In the following, we introduce more details about the agents as well as the implemented ontology. 3.1 The Agent Specification We defined different agents dedicated to the agentification of the HDW professional network. The next table summarize the different roles played by each agent. Agent Behavior The HDW is the decision maker and the final user of the recommender system. In order to satisfy her clients and minimize her expenses, the HDW HDW must check the available quantities within her stock. In the presence of an over or under-storage, she orders the barter and the purchase agents to call for the best e-procurement opportunities. The interface agent is considered as a middleware between the HDW agent and the remaining agents. It facilitates their interactions. Interface 357 Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled, Khaled Ghedira It transmits a command to the HDW agent. The command includes one product or plus with different quantities. Customer It is the stock manager. It manages the entrees of new raw materials and updates their quantities if it is necessary. Whenever an under or over-storage case emerges, it notifies the HDW continuously. Stock Notifier Once receiving the HDW order, it looks for suitable suppliers based on three levels: selecting only providers disposing of the required goods, Purchase then only those having a seller profile successfully matched with the HDW one and finally selecting the trustworthy ones. Once definitively selected by the HDW, the supplier agent is called to negotiate with her about the required good quantity, quality, costs and delivery timeout. Supplier It broadcasts exchange demand within the professional network of the HDW in question. Indeed, it calls for HDW potentially interested in this Barter specific exchange and of course satisfying the current HDW preferences. Likewise, it is attentive to the circulation of the diverse exchange initiatives. It means that it examines the received exchange demands from other HDW and selects the interesting ones (those meeting the current HDW needs). Table 1: Agents' behaviours definition In the purpose to cover the handicraft domain details and thus facilitate the agents interactions, the business knowledge formalization is required. In fact, several business ontologies are implemented so to represent the handicraft business particularities. In the next subsection, we take as example the tapestry business ontology design. 3.2 The Handicraft Domain Specification In order to join the handicraft woman expectations, we need to specify her business environment. A suitable way is to represent her professional world via the ontology formalism. An ontology is defined as a part of the real-world knowledge representation (Guarino 1998). The adoption of an ontology within the e-barter system ensures a successful communication between its agents and enables their interoperability and coordination (Sadeh et al. 2003). Actually, we integrated different business ontologies related to different handicraft areas such as tapestry, ceramic, embroidery etc. Every business ontology identifies and highlights the concepts and relationships specific to a particular handicraft field (Dhaouadi et al. 2014b). In the following, we propose a short overview on the business ontology dealing with the tapestry production. In the following figure (Figure 2), three business processes are depicted namely procurement, producing and commercialization processes. The procurement process, for instance, may be achieved through an eventual E-purchase or E-barter procedures. Each one includes different activities. Indeed, the E-barter has the "Need Specification", "Peer Selection", "Barter Auctions" and finally the "exchange procedure" as activities. Regarding the producing process, it comprises different phases such as "Raw Material Preparation" and "Realization". Each producing phase may be divided into diverse activities which each requires 358 Ontology based Multi-Agent System for the Handicraft Domain E-Bartering the use of several "Tools" (e.g. Loom) and has as input diverse "Raw Materials" (e.g. Colorant, Wool). Figure 2: The business ontology related to the tapestry production 3.3 The E-Barter Interaction Protocols In this subsection, we highlight the agents' interactions through the introduction of several communication protocol diagrams. The next figure (Figure 3) depicts the impact of a new customer command arrival. Figure 3: Communication protocol diagram for the stock checking In order to satisfy the client demand, the HDW would like to check automatically the available quantities within the stock. The interface agent plays the role of middleware between the HDW and the Stock Notifier. The latter is responsible of the checkout of the recommended raw materials for the production. Indeed, one interesting aspect in our work is to deduce automatically the supply requirement. Note that the under/over-storage verification is executed continuously and is not only expected when a new command is received. Different 359 Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled, Khaled Ghedira business rules are used to infer an over or under-storage situations. These rules are considered as input of the Stock Notifier verification technique. More details about this technique are presented in the Algorithm 1. Before moving to the latter, several basic notions will be defined in the following: Definition 1 (Minimal stock). The minimal stock includes the raw materials quantities necessary for the production. MinQ_RMj, for instance, represents the minimal quantity relatively to the raw material RMj. Definition 2 (Safety stock). The safety stock represents the extra stock level necessary for the production sustainability under critical circumstances. Particularly, SQ_RMi depicts the safety quantity of the raw material RMi. Definition 3 (Maximal stock). The maximal stock represents the maximal raw materials quantities able to be stored. The producers target to respect the maximal stock level due to the storage cost and the raw materials degradation over time. MaxQ_RMi, for example, expresses the maximal quantity regarding the raw material RMi. Definition 4 (Customer command). The customer command includes different artisanal products with the respective quantities. It is formulated as follows: C (customeri) = NB_P1 * P1 +....+ NB_Pi * Pi +...+ NB_Pn * Pn (1) With : C (customeri) : is the command related to the customeri. Pi: is the product of category "i". NB_Pi: is the number of pieces of the product Pi. Definition 5 (Business Rule). The business rule , in our context, expresses approximately the involvement degree of the different raw materials within an artisanal product fabrication. These rules are extracted from the HDWs answers when they are face to face interviewed. Actually, we conducted an important number of interviews in the purpose to gather relevant producing techniques. Firstly, a questionnaire is performed by a sociologist which is a project member. The questionnaire includes different questions such as: " Who designs the products for you?"," What are the needed raw materials for your production and what are the recommended quantities?", "What are the needed tools for your production?", "Are there any procurement difficulties?", "What are the transformation process stages?" etc. As project members, we conducted about 100 interviews with HDWs having different profiles issued from Tunisia and Algeria. The HDWs answers provide however a good basis for the business rules extraction. A business rule is formulated as follows: R (Pi) = t1 * RM1 + t2 * RM2 +...+ tj * RMj + ...+ tm * RMm (2) With : R(Pi) : is the business rule related to the product Pi fabrication. RMj: is a particular raw material required for the product Pi fabrication ti : is the involvement degree (needed quantity) of the raw material RMj within the product Pi fabrication. Example. Let a business rule R1 regarding the product P1 creation: 360 Ontology based Multi-Agent System for the Handicraft Domain E-Bartering R1 (P1) = 5 * RM1 + 2 * RM2 + 0.25 * RM3 Literally, this rule notes that 5 units of RM1, 2 units of RM2, 0.25 unit RM3 must be available in order to produce P1. We introduce right now the check stock state algorithm which uses the terms previously defined. ALGORITHM 1. Check Stock State Algorithm Input: Node {NB_P1, NB_P2...NB_Pi... NB_Pn}// the commanded products numbers {SQ_RM1, SQ_RM2...SQ_RMi... SQ_RMm}// the safety quantities of the raw materials RMi {MaxQ_ RM1, MaxQ_ RM2... MaxQ_ RMi... MaxQ_ RMm}// the maximal quantities of the raw materials RMi {Q_RM1, Q_RM2...Q_RMi... Q_RMm} // the current quantities of the raw materials RMi {R (P1), R (P2)... R (Pi)... R (Pn) }// the business rules regarding the product Pi production Output: The notifications concerning raw material out of stock or on surplus with respective quantities; */ We initialize the minimal quantity MinQ_RMj with the safety quantity SQ_RMj value. for j from 1 to m do MinQ_RMj ← SQ_RMj end for */ If a raw material RMj is included into the commanded products {P1, P2, ..., Pn} fabrication then we compute its minimal quantity relying on the defined business rules and the demanded articles pieces number*/ for i from 1 to n do // scroll-on the products' list {P1,P2,...Pi,...Pn} for j from 1 to m do // scroll-on the raw materials' list {RM1,RM2,...RMj,..RMm} if RMj ∈ R (Pi) then QMin_RMj ← QMin_RMj + tj * NB_Pi end if end for end for */ Check the presence of an under or over-storage situation regarding the raw material RMj */ for j from 1 to m do if (QMIN_RMj > Q_RMj) then Write ("Under-storage Alert concerning", RMj, "Quantity in deficit is", QMIN_RMj - Q_RMj); end if if (Q_RMj > QMAX_RMj ) then Write ("Over-storage Alert concerning", RMj, " Quantity in excess is", Q_RMj - QMAX_RMj); end if end for end Once, the check stock state algorithm is executed, three alternatives may arise regarding each raw material namely available quantity, under-storage or over-storage cases as showed in the next diagram (Figure 4). 361 Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled, Khaled Ghedira Figure 4: Communication protocol diagram for the alerts management As illustrated in (Figure 4), the checking process is repeated for each raw material. For each process iteration, the Stock Notifier agent deduces and displays the state of the stored raw material on the HDW interface. If the quantity is available, the HDW can proceed for the producing phases normally. If an under-storage case is occurred, the HDW asks both the purchase and barter agents to search for relevant supply opportunities. However, if an over- storage situation emerges, the HDW is only allowed to look for exchange possibilities through the barter agent recommendations. As said before, we do not consider the procurement alternatives suggested by the purchase agent which are undertaken in (Dhaouadi et al. 2014a). Besides, in this paper, we deal particularly with the e-bartering scenarios. To do so, the barter agent is the responsible for driving the research process for suitable peers. In order to initialize the exchange process (Figure 5), the barter agent calls for exchange opportunities within the HDW network. It addresses accurately the different barter agents relative to the remaining HDWs. Note that Barter[i] and HDW Interface[i] mean respectively the barter and the HDW interface agents related to the HDW i-th instance. Once the exchange demand is received and displayed in her interface, the HDW may refuse or accept it. In the second case, the acceptation is propagated until the initial barter agent. The latter has to rank the received positive answers through the execution of the matching profile algorithm. The proposed algorithm adopts different rules which guides the selection of suitable barter peers. In fact, two HDWs having similar profile parameters, such as living at the same locality, are disposed to be relevant exchange peers. The ranked list is then communicated to the current HDW. 362 Ontology based Multi-Agent System for the Handicraft Domain E-Bartering Figure 5: Communication protocol diagram for the exchange call As shown in (Figure 6), three cases may arise relying on the number of received positive answers. If the ranked list in empty (answers number = 0) thus the current HDW has the choice to look for new procurement opportunities. If only one positive response is received (answers number = 1) then the HDW responsible for the decision making is able to initialize a negotiation procedure with the proposed exchange candidate. The negotiation begins with a call for proposal from the current HDW to the eventual exchange peer (HDW [i]). The latter proposes the proprieties of the good to exchange and can ask for another in return. Of course, the proposal is propagated until the current HDW. If she disagrees with this proposal, it does not mean the negotiation is failed. Nevertheless, the negotiation steps will be repeated until the mutual satisfaction of both barter peers occurs. This loop is broken when the HDW in question accepts or refuses the proposal definitively. The third scenario occurs when the number of received positive answers exceed 1 (answers number > 1). Here, the HDW bids on the barter auction in order to select the best exchange offer. As a barter auction initiator, the HDW targets to acquire the needed raw material and give back the less useful one. In the opposite side, each HDW, which seems to be an eventual barter candidate, sends her proposal towards the barter initiator (Figure 6). When the proposal list is displayed within her interface, the HDW opts for the promising alternative. Sometimes, she calls for proposal for a second or third time until receiving a convenient offer: an offer which maximizes the needed raw material quantity and minimizes the quantity of the asked raw material. In other words, the barter auction scenario (frequent in real life) maximizes considerably the HDW (barter initiator) benefit. Likewise, it affords her more flexible and relevant choices. 363 Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled, Khaled Ghedira Figure 6: Communication protocol diagram for the exchange scenarios 4 Validation In order to validate the proposed approach, we emulate several e-bartering scenarios. The simulation objective is to show the real scenarios feasibility. First-of-all, we settled up different business ontologies related to tapestry, ceramic and embroidery productions based on Protégé software. We specified likewise, several business rules as extracted from the interviews where each rule is dealing with a specific product regarding its nature as well as its characteristics (dimensions, weight etc.). The system as described above is implemented using the jade platform under eclipse tool. Moreover, the bean ontology generator plug-in is used to generate the Java files representing the developed ontologies (easily used by the jade environment). During the simulation, we created 2 customer agents in addition to a community of 10 HDW agents having different profile parameters (geographic locality, handicraft domain) and their respective barter, purchase and interface agents. The agent communication is ensured via the FIPA-ACL language. The Contract-Net Protocol is followed during the bidding and negotiation procedures. In the following, we expose a simulation scenario (Figure 7) where The interfaces express the process evolvement. 364 Ontology based Multi-Agent System for the Handicraft Domain E-Bartering "Sofia" (a HDW agent), received a customer command (Interface 1) asking for the fabrication of the "margoum" product with precise dimensions (2 meters of length and 1.5 meters as width). The "margoum" is a wool weaving used as floor carpet whose origins are Arab-Berber. After the automatic stock checking (Interface 2), it has been found that the "wool" (required raw material) is out of stock. Hence, the HDW looks for suitable procurement opportunities. Effectively she receives different supply alternatives from both barter and purchase agents (Interface 3). Since the HDW is rather interested in an exchange procedure, she focuses on the ranked list of the eventual exchange peers recommended by the barter agent. The list comprises 2 tapestry makers (HDWs) namely Sarah and Linda which are living at the same locality as "Sofia". A barter auctions scenario is held then (Interface 4). While Sarah proposes the "colorant" raw material in exchange for "wool", Linda asks for cotton. The colorant is less interesting according to Sofia that is why she opts for Sarah as the interesting exchange peer. Figure 7: The HDW interactions with the system 5 Conclusion We proposed a recommender system which assits the HDW during the procurement process. It offords her relevant supply opportunities through suggesting suitable barter partners. The trade peers suitability relies on their profile similarities and their expenses minimization. The e-barter system is recurrent supply solution which replaces or extends classical procurement transactions. E-bartering consists in exchanging goods between a network members without the implication of money. Our approach proposes online solutions. It guarantees also the barter initiator (HDW) benefit on two levels. The first level renforces her relationships with the network members. The second level promotes her economic profit via the barter auctions 365 Rahma Dhaouadi, Kais Ben Salah, Achraf Ben Miled, Khaled Ghedira module. In further research we propose to test our approach in real context where data regarding profile parameters and preferences are learned automatically. Likewise, we propose to define new ways for assessing the bartered goods values. Acknowledgement We are very thankful to the Tunisian Algerian Project (BWEC) dealing with the improvement of handicraft women business in emerging countries through affordable technologies and social networks. References  Proceedings from conferences Cavalli Ana R, Maag Stéphane.(2004).Automated test scenarios generation for an e-barter system. In SAC, 14-17 March (795-799). Nicosia, Cyprus Guarino Nicola.(1998).Some Ontological Principles for Designing Upper Level Lexical Resources. In Proceedings of the First International Conference on Lexical Resources and Evaluation. Dhaouadi Rahma, Ben Miled Achraf, Ghédira Khaled.(2014)a.Ontology based Multi Agent System for Improved Procurement Process: Application for the Handicraft Domain. In the 18th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems, 15-17 September (251-260). Gdynia-Poland: Elseiver. Dhaouadi Rahma, Ben Miled Achraf, Ghédira Khaled.(2014)b.Système multi-agents à base d’ontologies pour la recommandation d’opportunités d’approvisionnement. In the fifth French Days on Ontology (JFO), 14-16 November (139-144). Hammamet-Tunisia Thanh Le Bach, Dieng-Kuntz Rose, Gandon Fabien.(2004).On Ontology Matching Problems - for Building a Corporate Semantic Web in a Multi-Communities Organization. In ICEIS, 14-17 April (236-243). Porto - Portugal.  Journal Articles Pla Albert, López Beatriz, Murillo Javier, Maudet Nicolas.(2014).Multi-attribute auctions with different types of attributes: Enacting properties in multi-attribute auctions. Expert Syst. Appl. 41(10), 4829-4843 Küpçü Alptekin, Lysyanskaya Anna. (2012).Usable optimistic fair exchange.Computer Networks 56(1), 50-63. Huang He, Kauffman Robert J., Xu Hongyan, Zhao Lan.(2013).A hybrid mechanism for heterogeneous e-procurement involving a combinatorial auction and bargaining.Electronic Commerce Research and Applications 12(3), 181-194. Christidis Konstantinos, Mentzas Gregoris.(2013).A topic-based recommender system for electronic marketplace platforms.Expert Syst. Appl. 40(11), 4370-4379. 366 Ontology based Multi-Agent System for the Handicraft Domain E-Bartering Lee Carman Ka Man, Lau Henry C. W, Ho George T. S, Ho William.(2009).Design and development of agent-based procurement system to enhance business intelligence.Expert Syst. Appl. 36(1), 877-884. Lin Che-Chern, Chen Shen-Chien, Chu Yao-Ming. (2011).Automatic price negotiation on the web: An agent-based web application using fuzzy expert system.Expert Systems with Applications. 38(5), 5090–5100. Núñez Manuel, Rodríguez Ismael, Rubio Fernando.(2005).Formal specification of multi-agent e- barter systems. Sci. Comput. Program. 57(2), 187-216. Haddawy Peter, Cheng Christine, Rujikeadkumjorn Namthip, Dhananaiyapergse Khaimook.(2005).Optimizing ad hoc trade in a commercial barter trade exchange. Electronic Commerce Research and Applications. 4(4), 299-314. Rosaci Domenico, Sarnè Giuseppe. M. L.(2014).Multi-agent technology and ontologies to support personalization in B2C E-Commerce.Electronic Commerce Research and Applications. 13(1), 13-23. Sadeh Norman M., Hildum David W., Kjenstad Dag.(2003).Agent-based e-supply chain decision support. Journal of Organizational Computing and Electronic Commerce. 13(3-4), 225- 241. Lee Seungsup, Choi Keunho, Suh Yongmoo.(2013).A personalized trustworthy seller recommendation in an open market. Expert Syst. Appl. 40(4), 1352-1357. Wang Xiaohuan, Wong T. N., Wang Gong.(2012).An ontological intelligent agent platform to establish an ecological virtual enterprise.Expert Syst. Appl. 39(8), 7050-7061. 367 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia SOCIAL CRM PERFORMANCE DIMENSIONS: A RESOURCE-BASED VIEW AND DYNAMIC CAPABILITIES PERSPECTIVE Nicolas S. Wittkuhn University of St. Gallen, Switzerland nicolas.wittkuhn@student.unisg.ch Tobias Lehmkuhl University of St. Gallen, Switzerland tobias.lehmkuhl@unisg.ch Torben Küpper University of St. Gallen, Switzerland torben.kuepper@unisg.ch Reinhard Jung University of St. Gallen, Switzerland reinhard.jung@unisg.ch Abstract Social Customer Relationship Management (SCRM) is a new paradigm to manage and engage customers via Social Media and should be treated as a holistic business strate- gy. Despite convincing reference cases by scholars and practitioners, there is still skep- ticism and reservation towards SCRM. Scholars are applying the resource-based view and the dynamic capabilities perspective for their exploratory and explanatory research to provide insights backed by these proven theories. This paper examines contemporary research and juxtaposes it to current business needs within a holistic SCRM perfor- mance dimension framework. The results are obtained through interactive research. The paper provides new and validated definitions of infrastructure and process compo- nents related to SCRM and develops propositions regarding customer-centric resources and capabilities. It further reveals research gaps within the literature regarding SCRM performance measurement and provides suggestions for further research. Keywords: Social CRM, customer-centricity, resource-based view, dynamic capabili- ties, performance dimensions, infrastructure, processes SCRM performance dimensions: a resource-based view and dynamic capabilities perspective 1 Introduction SCRM has arrived and is perceived as a means to reach new level of customer interac- tion, engagement and co-creation. Based on the use of Social Media and Web 2.0 prin- ciples1, it is achieving recognition both in the academic and in the business world (Lehmkuhl & Jung, 2013a; Paniagua & Sapena, 2014). Introducing term and concept, Greenberg labels SCRM as “a philosophy and a business strategy, supported by a sys- tem and a technology, designed to engage the customer in a collaborative interaction that provides mutually beneficial value in a trusted and transparent business environ- ment” (2010, p. 414). SCRM is thus a new paradigm to manage and engage customers via Social Media (Askool & Nakata, 2010) and consequently should be treated as a ho- listic integrated business strategy, rather than an (IT-focused) extension of existing CRM concepts (Lehmkuhl, 2014; Malthouse, Haenlein, Skiera, Wege, & Zhang, 2013; Woodcock, Green, & Starkey, 2011). As User Generated Content (UGC) accounts for around 11% of global internet traffic (Hennig-Thurau et al., 2010) and UGC-related websites are some of the most popular sites in the Internet (Dylko, 2014), businesses have been long-way into Social Media and Web 2.0 technologies. Around 80% of executives perceive Social Media as highly relevant for their business (Choudhury & Harrigan, 2014). More specifically, several studies have highlighted the usability of Social Media and Web 2.0 in pushing direct sales (Paniagua & Sapena, 2014), improving marketing effectiveness (Alt & Reinhold, 2013), installing customer support communities (Lehmkuhl & Jung, 2013b) or offering new service channels (Bock, Ebner, & Rossmann, 2013). Studies also provide proof regarding beneficial financial effects of Social Media and Web 2.0. For example, cus- tomer “engagement in Social Media brand community leads to a significant increase in consumer purchases” (Goh, Heng, & Lin, 2013, p. 103) and customer support commu- nities may reduce service costs by about 90% compared to call centers (Ang, 2011, p. 36). Yet, despite convincing reference cases and scholars’ calls to establish SCRM stra- tegically, there is still reservation in practice. Market analyses reveal that merely 11% percent of organizations have a formal SCRM program in place (Dickie, 2013). Most of this is likely to be related to a customer service in which customers handle other cus- tomers’ service requests based on Social Media. There is a paucity of evidence for com- prehensive SCRM programs spanning throughout entire businesses. That is, there are more watchers than actors and the actors are still experimenting with Social Media ap- plications to find the optimal leverage. Scholars are focusing their research on practical, implementable results to help “develop and deploy [the] new technologies and capabilities” (Trainor, 2012, p. 319). Building on the resource-based view and a dynamic capabilities perspective (Eisenhardt & Martin, 2000), these scholars hypothesize and confirm positive relationships between technical and organizational resources, SCRM capabilities and business performance (Choudhury 1 Definition:  Web 2.0 is a set of economic, social and technology trends that collectively form the basis for the next generation of the Internet – a more mature, distinctive medium characterized by user participation, openness and network effects (Musser & O’Reilly, 2006, p. 4)  Social Media are a group of Internet-based applications that build on the ideological and techno- logical foundations of Web 2.0, and that allow the creation and exchange of User Generated Content (A. M. Kaplan & Haenlein, 2010, p. 61) 369 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung & Harrigan, 2014; Trainor, Andzulis, Rapp, & Agnihotri, 2014). Especially in relation to business performance, existing frameworks such as the service profit chain (Heskett & Schlesinger, 1994) or the balanced scorecard (R. S. Kaplan & Norton, 1992, 1996) seem appropriate to capture SCRM in a holistic or cross-functional setting (Payne & Frow, 2005, p. 172). Research on these frameworks within the context of CRM (H.-S. Kim & Kim, 2009; Llamas-Alonso, Jiménez-Zarco, Martínez-Ruiz, & Dawson, 2009) might serve as a suitable basis in this regard. A recent literature review by Küpper et al. (2014) provides an overview of current advances on performance measures in the con- text of SCRM. From a business perspective, performance measures are most viable once measurable and relevant objectives for SCRM operations are defined. Lehmkuhl et al. (2015) provide first guidelines, developing a CRM-based scorecard approach into a SCRM framework which might support businesses in diagnosing and improving SCRM practices. This paper builds on and combines these research results and proposes a solution to- wards the dilemma of scholars calling for comprehensive SCRM programs with busi- nesses using Social Media for - at best - service issues. To help closing this gap between theory and practice, an interactive research approach is chosen (Gummesson, 2001, 2002), accompanied by a literature review. Two research questions (RQ) have been de- fined for this paper:  RQ1: What are up-to-date and comprehensive definitions for resources and capabili- ties-related performance dimensions of SCRM?  RQ2: What propositions support the resources and capabilities-related performance dimensions of SCRM? The paper is organized as follows: First, the authors present the conceptual background and detail the research objectives. The next section introduces the methodology, fol- lowed by a presentation and later discussion of the results. The paper concludes with highlighting contributions and implications to theory and practice. 2 Theoretical Background and Conceptual Model 2.1 The Concept of SCRM With the arrival of Social Media and Web 2.0, the traditional notion of CRM as “a cross-functional strategic approach concerned with creating improved shareholder value through the development of appropriate relationships with key customers and customer segments” (Frow & Payne, 2009, p. 11) has been shaken and reshaped. Customers now have a direct link to businesses, bypassing call centers and field agents (Alt & Reinhold, 2012, p. 281). Customer interaction with other customers regarding a business’s prod- ucts is now easier and this word of mouth creates a more believable and trustable source of information than corporate advertising (Acker, Gröne, Akkad, Pötscher, & Yazbek, 2011, p. 6). Businesses have to realize that in order to maintain good customer relation- ships, they have to switch from a parent (company) – child (customer) relationship to- wards collaborative efforts based on what Greenberg coined “Social CRM” (2010, p. 414). Recently, scholars have started pushing towards a more strategic definition of the term and provided blueprints for SCRM adoption or implementation (Choudhury & Harrigan, 2014; Lehmkuhl & Jung, 2013a; Lehmkuhl, 2014; Malthouse et al., 2013). In this paper, SCRM is understood as a holistic business strategy based on and driven by 370 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective integrated Social Media in order to focus customer-facing activities, processes, systems and technologies on engaging customers in collaborative communication and co- creation in order to optimize customer relationships (Greenberg, 2010; Trainor et al., 2014; Trainor, 2012). To further advance conceptual understanding of SCRM, researchers have built on the resource-based view and the dynamic capabilities perspective. Schaupp and Bélanger (2014) build on the resource-based view to analyze Social Media value for small busi- nesses. Apart from underlining the relevance of applying the resource-based view in Social Media research, they derive four different dimensions of potential Social Media value. Within their study, Social Media has the highest impact on internal organization- al operations, followed by impact on sales, marketing and customer service, potentially leading to operative efficiency gains, increased sales areas, reduced marketing costs and improved customer satisfaction. Trainor (2010) focuses his research exclusively on the capabilities part and has developed a capabilities-based perspective on SCRM. He com- bines a traditional view on CRM resources, capabilities and processes with new cus- tomer-centric technologies and processes for SCRM, linking both views to performance outcomes. He lists five capabilities and three performance outcomes, highlighting the need for further academic research on this topic (2012, p. 328). Building on this, Trai- nor et al. elaborate on the conceptualization and measurement of SCRM capabilities (2014, p. 1201). Expanding the traditional CRM capability of relational information processing into a SCRM capability comprising information generation and dissemina- tion, they postulate that this capability is “a unique combination of emerging technolog- ical resources and customer-centric management systems that can lead to customer sat- isfaction, loyalty, and retention” (2014, p. 1202). Finally, Choudhury and Harrigan (2014) develop a new construct labelled customer engagement initiatives to stress the change in communication between businesses and customers as caused by Social Me- dia. Embedded within the context of the resource-based view and the equity theory, they show the positive impact of this construct on customer relationship performance. Fur- thermore, they also develop updated definitions on CRM technology use and relational information processes. Although all these articles provide a vantage point in scope of this paper, some limita- tions have to be mentioned. Mainly, the research does not apply a holistic SCRM per- spective with regard to the definition of resources, capabilities and related performance measures. Viewing SCRM as “an extension” (Trainor, 2012, p. 319) limits identifica- tion, analysis and development of distinct and unique resources and capabilities for SCRM, as the scope of such research often focuses on specific organizational functions only, instead of incorporating the complete organization with all its production factors and in its entire context. Regarding potential performance dimensions for SCRM, Kim and Kim (2009) have shown that a detailed and balanced set of performance measures is necessary to capture the effects of capabilities and to provide strategic guidance to oper- ate SCRM cross-functionally. The research by Lehmkuhl et al. (2015) provides a useful framework in this regard, but lacks clear definitions of the proposed dimensions and objectives and does not specify factors to reach the different performance objectives. This paper is thus motivated by this research gap and aims at developing the paradigm of SCRM towards a more holistic perspective. 371 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung 2.2 Conceptual Model and Research Objectives Building on the resource-based view, the dynamic capabilities perspective and also the balanced-scorecard approach within the context of SCRM is valuable for three reasons. First of all, research needs “theory for guidance, but not for obedience; we should go back to the “classics” to get a perspective, but for application today most of the “clas- sics” are in need of upgrading or replacement” (Gummesson, 2002, p. 347). Secondly, researchers such as Trainor state that “one of the most pressing challenges […] is relat- ed to capability measurement” (2012, p. 328) – thus calling for both research on capa- bilities and on measurement approaches for SCRM. Lastly, research results based on these abovementioned well-proven concepts and theories (i.e. the resource-based view and the dynamic capabilities perspective) might diminish practitioners’ uncertainty re- garding the overall benefits of SCRM. Integrating a detailed performance dimension framework with a specific focus on resources and capabilities might thus create a new impetus for research and application of SCRM. Building on Trainor (2014; 2012), Kim and Kim (2009), Küpper et al. (2014) and Lehmkuhl et al. (2015), the conceptual model for this paper is depicted in Figure 1. Figure 1: Conceptual model and research focus Numerous studies have already proven the link between resources, capabilities and their effects on performance dimensions (Bharadwaj, 2000; Drnevich & Kriauciunas, 2011; Huselid, Jackson, & Schuler, 1997; Zott, 2003). Within the performance dimensions, the two dimensions Infrastructure and Processes contain components related to actions based on resources and capabilities, whereas the dimensions Business Performance and Customer hold components related to measuring the end results of those actions (Küpper et al., 2014; Lehmkuhl et al., 2015). As the focus of this paper is on resources and capabilities-related SCRM performance dimensions, the latter two dimensions are excluded from the current research focus. The dimensions Infrastructure and Processes are thus at the core of the research questions for this paper, together with the focus on identifying resources and capabilities. Given the scarcity of studies on SCRM operation as well as taking into account the limi- tations of existing research, this paper is built on two research objectives (RO), which are in line with the guiding research questions. 372 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective  RO1: Provide validated definitions for components within the SCRM performance dimensions Infrastructure and Processes.  RO2: Develop propositions regarding required customer-centric SCRM resources and capabilities to support the performance dimensions Infrastructure and Processes. 3 Methodology This paper builds on and extends current research on SCRM by integrating research on SCRM performance dimensions with SCRM research on resources and capabilities. The research approach is based on a literature review and interactive research (Gummesson, 2001, 2002), as detailed in Figure 2. Figure 2: Research methodology Defining the components of the SCRM performance dimensions (RQ1) is a necessary step to enable the interactive research. First of all, the scope for the literature review was determined, based on the work of Cooper (1988). Consequently, for each of the components, an exhaustive literature research was conducted, applying the guidelines concerning rigor as laid out by Vom Brocke et al (2009).2 Following this, and still in scope of RQ1, the interactive research then focused on validating the developed compo- nent definitions as well as grading them in terms of necessary adaptation with regards to existing and known definitions. Based on these validated definitions, the scope of the interactive research then moved to RQ2 in order to identify, discuss, review and finally approve propositions regarding customer-centric resources and capabilities supporting the performance dimensions in scope of this research. The interactive research concept has been chosen as the underlying methodology for the research at hand, because as a mixed method, it is able to provide answers to both re- searchers and practitioners (Ballantyne, Frow, Varey, & Payne, 2011; Frow, Payne, & Storbacka, 2011; Payne, Ballantyne, & Christopher, 2005; Payne, Storbacka, Frow, & Knox, 2009; Payne & Frow, 2006). The research questions for this paper are set within 2 More details regarding the literature review process and its results can be obtained directly from the authors. 373 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung a broader research program in which four business organizations (telecommunication and insurance) take part to professionalize their Social Media management with a focus on CRM. Included herein is a panel of four executives working in CRM, customer ser- vice and IT, a group of 10 operative and strategic Social Media experts (customer ser- vice, communication, direct / digital / strategic marketing, IT), and a team of six re- searchers and external consultants. The research was conducted in a series of focus group workshops and expert interviews and took place between March and November 2014. 4 Results 4.1 Infrastructure Dimension While elaborating on the literature as input for the definition of the components, super- ordinate similarities between different components could be identified. For the first four components, this superset was labelled Culture, for the last three components, it was labelled Information Management. Both supersets of the respective components are introduced as new sub-dimensions beneath the Infrastructure dimension. Tables 1a and 1b present these new sub-dimensions, the validated definition for the components, the degree of necessary adaptation of existing definitions3 as well as related research. Sub- Component Definition Degree of Related Research Dimension adaptation Open-minded A holistic organizational culture, defined by a poten- Little van Bentum and Stone, culture tially technology-supported / -based willingness of the 2005; Frow and Payne, organization and its employees to actively share 2009; King and Burgess, knowledge and information across internal functions 2008 and communication channels in order to create open and transparent communication Management Active and explicit support, encouragement and No Alt and Puschmann, 2004; commitment involvement of top management regarding the intro- Becker et al., 2009; duction, usage and development of SCRM Bohling et al., 2006; Dong et al., 2009; King and Burgess, 2008 Culture Integration of Integration of back-office functions (employees, IT Full Buehrer and Mueller, back-office applications) into direct customer communication to 2002; Bull, 2003; functions leverage existing and relevant knowledge and skills Finnegan and Currie, in order to speed up communication processes and 2010; Karimi et al., 2001; to ensure ‘first time right’ answers or solutions Payne and Frow, 2005; Stefanou et al., 2003 Customer- Direct, personal, interactive and multi-directional Little Hartline et al., 2000; Kim centric com- communication between an organization and its et al., 2012; Schultz et al., munication customers, irrespective of the communication chan- 2012 nel, built on common organizational values regarding customer-orientation or -centricity Table 1a: Results overview on Culture components of the Infrastructure dimension 3 No adaptation = existing definition from literature was used; little adaptation = slight changes to or update of existing definitions from literature; full adaptation = no definition within existing literature found, new definition developed 374 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective Sub- Component Definition Degree of Related research Dimension adaptation Social Media A selection of relevant customer-facing, front-office Little Kaplan and Haenlein, / CRM and back-office applications focusing on supporting 2010, p. 61; Mohan et al., applications Social Media and CRM-related processes and inter- 2008; Payne and Frow, actions 2005 SCRM A tailored, integrated and efficiently set-up IT- Little Alt and Reinhold, 2012, IT- architecture, consisting of tools and systems covering 2013; Greenberg, 2009; infrastructure the main SCRM functionalities (e.g. social search Payne and Frow, 2005; tools, social media monitoring, business intelligence, Reinhold and Alt, 2012 Information CRM systems, social network analysis, social media Management management and community management) while supporting process standardization SCRM data Descriptive and predictive analytics of (customer Little Reinhold and Alt, 2011; management segment, - behavior, -value and product-related) data Stieglitz et al., 2014 gathered and classified manually / semi-automatically / automatically through Social Media / Web 2.0 appli- cations to support key processes, decision-making, communication, sales, marketing, service functions for an optimized B2C-interaction Table 1b: Results overview on Information Management components of the Infrastructure dimension Within the Culture sub-dimension, three of the four components have good coverage within the contemporary literature and are also known and perceived as relevant by the participating research partners. However, a notable outcome concerns the component Integration of back-office functions. As of the time of this research, there was no current direct research on this component, a new definition fitting to the context of SCRM had to be developed. The idea of back-office employees having first-hand contact with cus- tomers was accepted as relevant by the research partners, because this component might be a relevant lever to achieve a customer-centric culture within an organization. Con- cluding, future research should try to identify examples and generate more insights into this topic. Within the Information Management sub-dimension, the component SCRM data man- agement was discussed most intensely. Especially the term ‘predictive analytics’ within the definition resonated with the participating business experts. Big expectations are placed into big data, or rather smart data, as one participant put it: “we need clean data on the customer, we need a 360° perspective to be able to provide the right information in the right time through the right channel before the customer asks for it”. Research and discussions regarding required resources and capabilities for both the Cul- ture and the Information Management components resulted in the following proposi- tions: 375 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung  Proposition 1: The higher the share of digital natives within a business’s work- force, the easier it will be to achieve a customer-centric culture under the roof of a strategic holistic SCRM.  Proposition 2: Information Management for SCRM is especially successful if human and technological analytical resources are strongly embedded within a business.  Proposition 3: Human and technological analytical SCRM resources are opti- mally embedded within a business and can create unique capabilities, if they are fully incorporated within and central to the value chain for SCRM processes. 4.2 Processes Dimension The Processes dimension components were structured into a group focusing on Internal business processes and another group relating to Customer-oriented processes. The val- idated definitions, indications regarding the degree of adaptation and information re- garding related research can be found in Tables 2a and 2b. Sub- Component Definition Degree of Related Research Dimension adaptation Strategy & Strategic framework comprised of vision, mission Little Bohling et al., 2006; added value statement, functional strategies and objectives, ensur- Malthouse et al., 2013; ing that SCRM is perceived and accepted as a benefi- Payne and Frow, 2006; cial, cross-functional holistic organizational program Wirtz et al., 2010; Woodcock et al., 2011 Co- Partnering with selected 3rd parties (e.g. digital start- Little Blomqvist, Kyläheiko, & operations ups, content providers, sponsors), not with customers, Virolainen, 2002; to fill internal organizational resource / capability gaps Constantinides, Romero, & in order to provide digital content and services for Boria, 2008; Day, 2011 achieving organizational differentiation in the percep- Internal tion of the target customers business Governance Relevant formal and informal rules, practices and Little Deans, 2011; De Hertogh et processes mechanisms needed to determine decision-making, al., 2011; Jutla et al., 2001; monitor decision execution, escalate problems, meas- Prasad et al., 2012 ure and control results of decisions, exercise empow- erment for decision-making and deal with accountabil- ity of decision makers Value Statements of the organizations towards its customer Little Agnihotri et al., 2012; Payne proposition (segments) regarding the specific physical or service- and Frow, 2005; Prahalad based offerings, defining the received specific value, and Ramaswamy, 2004; based on insight and interaction and aimed at co- Vargo et al., 2008 creating unique experiences for each customer (seg- ment) Table 2a: Results overview on Internal business processes components of the Processes di- mension Sub- Component Definition Degree of Related research Dimension adaptation Trigger- Pro-active and event-related execution of fitting ‘next- Full Academic research: Zeng based best’ actions across different organizational functions et al., 2010 actions within a specific situation involving the customer in Business-oriented publica- order to maximize customer satisfaction and retention tion: Pugh and Chessell, Customer- 2013 oriented Engagement A multi-dimensional concept defining the psychological Little Baird and Parasnis, 2011; processes state of a customer regarding emotional bonds and Brodie et al., 2011; Ray et relational exchange with a company, based on an al., 2014; Sashi, 2012 interactive and iterative process of co-creating cus- tomer experience and having different context-specific outcomes of engagement levels 376 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective Consistent / Create ‘perfect customer experience’ by ensuring and Little Baird and Parasnis, 2011; seamless using optimal customer knowledge, providing personal Frow and Payne, 2007; customer communication and ultimately customer experience Lemke et al., 2010; Payne experience across all communication / interaction channels by and Frow, 2005; Schmitt consistent use of technology and processes across all and Zarantonello, 2013 channels in order to increase the relationship with the customer and ultimately lead to full customer engage- ment Table 2b: Results overview on Customer-oriented processes components of the Processes di- mension In general, the components within the Internal business processes dimension were per- ceived as rather generic by the research partners. Terms such as strategy or governance convey a very broad meaning in a normal business sense unless specified very exactly. A first step before reviewing the performance dimensions thus was to have the research partners define specific mission statements and other strategic items for their respective businesses. Based on this work, the validation of the components and their definition was successfully accomplished. Within the sub-dimension Customer-oriented process- es, the component Trigger-based actions had to be defined first-hand. The relation of this process-related component to the infrastructure-related component of SCRM data management in terms of predicting customer behavior and communication is evident. As one research partner stated, “if we knew in advance, like if our IT could predict based on medical bills, when a customer’s baby was about to be born – if we knew how to and were legally allowed to do that – then with a great story to tell, how close could we get to this customer?” The discussion within the interactive research related to the Processes dimension focused mainly on capabilities and thus, the following proposi- tions were developed:  Proposition 4: A cross-functional governance capability is required to success- fully operate holistic SCRM within a business.  Proposition 5: Successful implementation and execution of trigger-based actions within customer-facing processes requires analytics-driven decision manage- ment.  Proposition 6: Of all components within the Processes dimension, Seamless and consistent customer experience has the highest direct impact on the Customer and Business Performance dimensions. 5 Discussion 5.1 Theoretical Contributions Firstly, the definitions of the components within the performance dimensions provide a sound basis to study SCRM as a holistic framework. Each definition is based on a thor- ough and rigorous literature review and could be used for specific research questions further advancing SCRM theory. Regarding the components Integration of back-office functions and Trigger-based actions, the definitions as introduced by this paper cover previously uncharted territory and advance the conceptual understanding of SCRM as a new paradigm. As Payne and Frow have shown in their research (Payne & Frow, 2005), in order to successfully support the establishment of a new paradigm, clear and accepta- ble definitions are a highly relevant factor to advance conceptual understanding and growth of knowledge. 377 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung Secondly, additional insights regarding customer-centric SCRM resources and capabili- ties are generated. Existing terms regarding customer-centric resources are clarified and refined. The notion of customer-centric technologies is enhanced by explicitly adding human and technical analytical SCRM resources, combining formerly unconnected re- search by Trainor (2012) and Reinhold & Alt (2011). Existing research on customer- centric capabilities is expanded: the strong focus on process capabilities (Choudhury & Harrigan, 2014; Trainor, 2012) is broadened by proposing the need for an analytics- driven decision management capability strongly embedded within SCRM infrastructure resources. Thirdly, this paper establishes new departure points for further SCRM research. The role of a cross-functional governance function within a holistic SCRM framework has been strengthened. Although within the SCRM performance dimensions, governance is but one component of many, research should further look into its potentially pivotal role in making SCRM operations successful. Lastly, the interactive research has also shown that by generally linking customer- centric SCRM resources and capabilities to the performance dimensions, specific rela- tionships between individual resources, capabilities and components can be identified. Strengthening these relationships through further research will advance SCRM theory and consequently its acceptance and transfer into the business world. 5.2 Managerial Implications This paper has shown that by focusing on understanding customer-centric SCRM re- sources and capabilities, a first step towards a holistic understanding of SCRM as a stra- tegic concept is taken. By achieving and having such a mindset, SCRM resources and capabilities can be potentially turned into sustainable competitive advantages – doing it half-heartedly will probably be noticed by the customers and could generate a loose- loose scenario. Some further remarks are necessary: a cultural SCRM mindset will be easier to imple- ment within a business if its workforce already has a high affinity to Social Media and Web 2.0 principles. A business striving to achieve a customer-centric SCRM culture should thus, next to training its existing workforce, put special emphasis on SCRM af- finity of applicants when filling vacant positions. This is also in line with current rec- ommendations by practitioner-oriented industry research (Hirt & Willmott, 2014). The results of the interactive study provide practical proof that developing a holistic SCRM perspective is not an impossible task. The first step is breaking down and modu- larizing this holistic perspective – here, the dimensions, sub-dimensions and compo- nents provide a sound basis for a first step. While this is a more top-down approach, having a look at required resources and capabilities and trying to identify opportunities to develop and implement those resources and capabilities represents a bottom-up ap- proach. Businesses should follow both approaches simultaneously in order to derive implementation measures to move forward on the road towards holistic SCRM. Of rele- vance in this context is designing a smart governance structure for all SCRM related activities, resources and processes, as is also shown by other studies (Baird & Parasnis, 2011a; Lehmkuhl, 2014). Depending on a business’s maturity with regard to the use of Social Media, Web 2.0 and SCRM, cross-functional governance elements have to be adapted or even newly designed. 378 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective 6 Conclusion, Limitations and Further Research 6.1 Conclusion In light of the research questions and the research objectives, the conceptual understand- ing of SCRM is advanced by the research results of this paper. Drawing on the re- source-based view and the dynamic capabilities perspective, this paper generates valua- ble insights into relevant SCRM resources and capabilities for infrastructure and process performance dimensions. Nonetheless, research relating to SCRM as a holistic strategic concept is still in its infancy, as a lot of research is undeniably still settled deeply within the CRM domain. The skepticism of practitioners to implement and operate SCRM pro- grams on holistic levels hinders the development of a SCRM domain as a relevant part of scientific literature. This paper only represents a first move to bring SCRM as a ho- listic strategy closer to implementation and operation within businesses. The perfor- mance dimension framework, the definitions of its components and the propositions regarding required resources and capabilities break down the big picture into smaller understandable compartments. These results add scientific credibility to the long-issued claims from market research institutes and the postulated need of mapping out a “social CRM capabilities-building plan” (Band & Petouhoff, 2010, p. 6). Consequently, this should motivate practitioners to drop their hesitancy and approach SCRM holistically. Doing this will create an opportunity to bring their businesses closer to their customers and ahead of their competition. 6.2 Limitations This paper is limited in several ways. First of all, the interactive research relied only on a small sample of research partners from only two branches of industry. And even though the research partners are acknowledged as Social Media pioneers in their respec- tive industry, the research results rely strongly on the statements of and discussions with the research partners. Currently, it is also mainly this expertise and little actual literature on which the new definitions for some components are based. Further research could prove them valuable, in need of adaption or not distinct enough from existing resources and capabilities to be regarded as new stand-alone results. Lastly, only parts of the con- ceptualized performance scorecard framework have been tested, as no focus was laid on the Business performance and Customer dimensions. Expanding the research scope might yield new insights into Social CRM resources and capabilities and discover new mediating or dependency relationships between resources, capabilities and outcomes. 6.3 Further Research With regard to future research, the authors would like to explicitly propose two areas. The first area concerns the role of Governance: research literature and practitioners mainly put SCRM either within the authority of marketing, customer service or IT. However, SCRM in a holistic perspective should be considered separate from or super- ordinate to these functions. In terms of cross-functional SCRM governance, a synthesis of existing literature and more research in terms of case studies, grounded - or action research is recommended. The second area focuses on the dimensions Customer and Business performance. Additionally to developing scientific definitions to build a basis 380 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung for mutual understanding between scholars and practitioners, future research should focus on providing examples for measurable items for each component. 380 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective Acknowledgement The authors wish to acknowledge the contributions by all research partners involved in the re- search program and their valuable input in scope of this research. Furthermore, the authors ex- tend their gratitude to the anonymous reviewers who provided valuable feedback to this paper. References Acker, O., Gröne, F., Akkad, F., Pötscher, F., & Yazbek, R. (2011). Social CRM: How companies can link into the social web of consumers. Journal of Direct, Data and Digital Marketing Practice, 13(1), 3–10. doi:10.1057/dddmp.2011.17 Agnihotri, R., Kothandaraman, P., Kashyap, R., & Singh, R. (2012). Bringing “Social” into Sales: The Impact of Salespeople’s Social Media Use on Service Behaviors and Value Creation. Journal of Personal Selling and Sales Management, 32(3), 333–348. doi:10.2753/PSS0885-3134320304 Alt, R., & Puschmann, T. (2004). Successful practices in customer relationship management, (November). Retrieved from http://works.bepress.com/cgi/viewcontent.cgi?article=1095&context=thomas_pusc hmann Alt, R., & Reinhold, O. (2012). Social-Customer-Relationship-Management (Social- CRM). Wirtschaftsinformatik, 54(5), 281–286. doi:10.1007/s11576-012-0330-6 Alt, R., & Reinhold, O. (2013). How Companies are Implementing Social Customer Relationship Management : Insights From Two Case Studies. In 26th Bled eConference (pp. 206–221). Ang, L. (2011). Community relationship management and social media. Journal of Database Marketing & Customer Strategy Management, 18(1), 31–38. doi:10.1057/dbm.2011.3 Askool, S., & Nakata, K. (2010). A conceptual model for acceptance of social CRM systems based on a scoping study. Ai & Society, 26(3), 205–220. doi:10.1007/s00146-010-0311-5 Baird, C. H., & Parasnis, G. (2011a). From social media to Social CRM: reinventing the customer relationship. Strategy & Leadership, 39(6), 27–34. doi:10.1108/10878571111176600 Baird, C. H., & Parasnis, G. (2011b). From social media to social customer relationship management. Strategy & Leadership, 39(5), 30–37. Retrieved from 10.1108/10878571111161507 Ballantyne, D., Frow, P., Varey, R. J., & Payne, A. (2011). Value propositions as communication practice: Taking a wider view. Industrial Marketing Management, 40(2), 202–210. doi:10.1016/j.indmarman.2010.06.032 381 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung Band, W., & Petouhoff, N. L. Topic overview: Social CRM goes mainstream. , Forrester Research Inc (2010). Retrieved from http://www.yieldopedia.com/paneladmin/reports/7278fa7637eb90752661b3670cb8 703d.pdf Becker, J. U., Greve, G., & Albers, S. (2009). The impact of technological and organizational implementation of CRM on customer acquisition, maintenance, and retention. International Journal of Research in Marketing, 26(3), 207–215. doi:10.1016/j.ijresmar.2009.03.006 Bharadwaj, A. S. (2000). A Resource-Based Perspective on Information Technology Capability and Firm Performance: An Empirical Investigation. MIS Quarterly, 24(1), 169. doi:10.2307/3250983 Blomqvist, K., Kyläheiko, K., & Virolainen, V.-M. (2002). Filling a gap in traditional transaction cost economics: towards transaction benefits-based analysis. International Journal of Production Economics, 79(1), 1–14. doi:10.1016/S0925- 5273(00)00095-5 Bock, A., Ebner, W., & Rossmann, A. (2013). Telekom hilft — Kundenservice via Social Media. Marketing Review St. Gallen, 30(3), 74–85. doi:10.1365/s11621- 013-0237-7 Bohling, T., Bowman, D., LaValle, S., Mittal, V., Narayandas, D., Ramani, G., & Varadarajan, R. (2006). CRM Implementation Effectiveness Issues and Insights. Journal of Service …, 9(2), 184–197. Retrieved from http://jsr.sagepub.com/content/9/2/184.short Brodie, R. J., Hollebeek, L. D., Juric, B., & Ilic, a. (2011). Customer Engagement: Conceptual Domain, Fundamental Propositions, and Implications for Research. Journal of Service Research, 14(3), 252–271. doi:10.1177/1094670511411703 Buehrer, R., & Mueller, C. D. (2002). Approach To Overcome Existing Limitations For CRM-implementation. In ECIS (pp. 1066–1076). Retrieved from http://is2.lse.ac.uk/asp/aspecis/20020132.pdf Bull, C. (2003). Strategic issues in customer relationship management (CRM) implementation. Business Process Management Journal, 9(5), 592–602. doi:10.1108/14637150310496703 Choudhury, M. M., & Harrigan, P. (2014). CRM to social CRM: the integration of new technologies into customer relationship management. Journal of Strategic Marketing, 22(2), 149–176. doi:10.1080/0965254X.2013.876069 Constantinides, E., Romero, C. L., & Boria, M. A. G. (2008). Social Media: A new frontier for retailers? European Retail Research, 22, 1–28. doi:10.1007/978-3- 8349-8099-1 382 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective Cooper, H. M. (1988). Organizing knowledge syntheses: A taxonomy of literature reviews. Knowledge in Society, 1(1), 104–126. doi:10.1007/BF03177550 Day, G. S. (2011). Closing the Marketing Capabilities Gap. Journal of Marketing, 75(4), 183–195. doi:10.1509/jmkg.75.4.183 De Hertogh, S., Viaene, S., & Dedene, G. (2011). Governing Web 2.0. Communications of the ACM, 54(3), 124–130. doi:10.1145/1897852.1897882 Deans, C. P. (2011). The impact of social media on c-level roles. MIS Quarterly Executive, 10(4), 187–200. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=buh&AN=67793227&site =ehost-live Dickie, J. (2013). How Marketers use Social CRM. CRM Magazine, (3). Dong, L., Neufeld, D., & Higgins, C. (2009). Top management support of enterprise systems implementations. Journal of Information Technology, 24(1), 55–80. doi:10.1057/jit.2008.21 Drnevich, P. L., & Kriauciunas, A. P. (2011). Clarifying the conditions and limits of the contributions of ordinary and dynamic capabilities to relative firm performance. Strategic Management Journal, 32(3), 254–279. doi:10.1002/smj.882 Dylko, I. (2014). Using Technological Attributes to Study Online Media: The Case of User-Generated Content. Journal of Broadcasting & Electronic Media, 58(4), 501–521. doi:10.1080/08838151.2014.966358 Eisenhardt, K. M., & Martin, J. a. (2000). Dynamic capabilities: what are they? Strategic Management Journal, 21(10-11), 1105–1121. doi:10.1002/1097- 0266(200010/11)21:10/11<1105::AID-SMJ133>3.0.CO;2-E Finnegan, D. J., & Currie, W. L. (2010). A multi-layered approach to CRM implementation: An integration perspective. European Management Journal, 28(2), 153–167. doi:10.1016/j.emj.2009.04.010 Frow, P., & Payne, A. (2007). Towards the “perfect” customer experience. Journal of Brand Management, 15(2), 89–101. doi:10.1057/palgrave.bm.2550120 Frow, P., & Payne, A. (2009). Customer Relationship Management: A Strategic Perspective. Journal of Business Market Management, 3(1), 7–27. doi:10.1007/s12087-008-0035-8 Frow, P., Payne, A., & Storbacka, K. (2011). Co-creation: a typology and conceptual framework. In Proceedings of the 2011 Anzmac Conference (pp. 1–6). Retrieved from http://202.154.59.182/ejournal/files/a typology and conceptual framework.pdf 383 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung Goh, K., Heng, C., & Lin, Z. (2013). Social media brand community and consumer behavior: Quantifying the relative impact of user-and marketer-generated content. Information Systems Research, 24(1), 88–107. doi:10.1287/isre.1120.0469 Greenberg, P. (2009). Social CRM comes of Age. Sponsored by Oracle. Retrieved from http://www.oracleimg.com/us/products/applications/crmondemand/industry- solutions/wealth-management/036062.pdf Greenberg, P. (2010). The impact of CRM 2.0 on customer insight. Journal of Business & Industrial Marketing, 25(6), 410–419. doi:10.1108/08858621011066008 Gummesson, E. (2001). Are current research approaches in marketing leading us astray? Marketing Theory, 1(1), 27–48. doi:10.1177/147059310100100102 Gummesson, E. (2002). Practical value of adequate marketing management theory. European Journal of Marketing, 36(3), 325–349. doi:10.1108/03090560210417156 Hartline, M. D., Maxham III, J. G., & McKee, D. O. (2000). Corridors of influence in the dissemination of customer-oriented strategy to customer contact service employees. Journal of Marketing, 64(2), 35–50. Retrieved from http://journals.ama.org/doi/abs/10.1509/jmkg.64.2.35.18001 Hennig-Thurau, T., Malthouse, E. C., Friege, C., Gensler, S., Lobschat, L., Rangaswamy, a., & Skiera, B. (2010). The Impact of New Media on Customer Relationships. Journal of Service Research, 13(3), 311–330. doi:10.1177/1094670510375460 Heskett, J., & Schlesinger, L. (1994). Putting the service-profit chain to work. Harvard Business Review, 72(2), 164 – 174. Retrieved from http://www.tgli.cl/students/intranet/content/Lecturas/Putting the Service-Profit Cahin to work.pdf Hirt, M., & Willmott, P. (2014). Strategic principles for competing in the digital age. McKinsey Quarterly, (May). Retrieved from http://www.mckinsey.com/insights/strategy/strategic_principles_for_competing_in _the_digital_age Huselid, M. A., Jackson, S. E., & Schuler, R. S. (1997). Technical and strategic human resources management effectiveness as determinants of firm performance. Academy of Management Journal, 40(1), 171–188. doi:10.2307/257025 Jutla, D., Craig, J., & Bodorik, P. (2001). Enabling and measuring electronic customer relationship management readiness. In Proceedings of the 34th Hawaii International Conference on System Sciences (Vol. 00). IEEE Computer Society. Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of Social Media. Business Horizons, 53(1), 59–68. doi:10.1016/j.bushor.2009.09.003 384 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective Kaplan, R. S., & Norton, D. P. (1992). The balanced scorecard - measures that drive performance. Harvard Business Review. Retrieved from http://moodle.bridgend.ac.uk/newvle/pluginfile.php/129783/mod_resource/content/ 0/Strategy Execution.pdf Kaplan, R. S., & Norton, D. P. (1996). The Balanced Scorecard: Translating Strategy Into Action. Harvard Business Press. Karimi, J., Somers, T., & Gupta, Y. (2001). Impact of information technology management practices on customer service. Journal of Management Information …, 17(4), 125–158. Retrieved from http://mesharpe.metapress.com/index/A06UE7VQ5V2UGAVW.pdf Kim, H.-S., & Kim, Y.-G. (2009). A CRM performance measurement framework: Its development process and application. Industrial Marketing Management, 38(4), 477–489. doi:10.1016/j.indmarman.2008.04.008 Kim, M., Park, J. E., Dubinsky, A. J., & Chaiy, S. (2012). Frequency of CRM implementation activities: a customer-centric view. Journal of Services Marketing, 26(2), 83–93. doi:10.1108/08876041211215248 King, S. F., & Burgess, T. F. (2008). Understanding success and failure in customer relationship management. Industrial Marketing Management, 37(4), 421–431. doi:10.1016/j.indmarman.2007.02.005 Küpper, T., Jung, R., Lehmkuhl, T., Walther, S., & Wieneke, A. (2014). Performance Measures for Social CRM : A Literature Review. In Proceedings of the 27th Bled eConference (pp. 125–139). Bled, Slovenia. Lehmkuhl, T. (2014). Towards Social CRM - A Model for Deploying Web 2.0 in Customer Relationship Management. University of St. Gallen. Lehmkuhl, T., & Jung, R. (2013a). Towards Social CRM - Scoping the concept and guiding research. In eInnovations: Challenges and Impacts for Individuals, Organizations and Society (pp. 190–205). Bled: Faculty of Organizational Sciences. Lehmkuhl, T., & Jung, R. (2013b). Value creation potential of Web 2.0 for SME - Insights and lessons learnt from a european producer of consumer electronics. International Journal of Cooperative Information Systems, 22(03), 1–22. doi:10.1142/S0218843013400030 Lehmkuhl, T., Jung, R., Küpper, T., & Wieneke, A. (2015). Zieldimensionen von Social CRM. Controlling, 27(7). Lemke, F., Clark, M., & Wilson, H. (2010). Customer experience quality: an exploration in business and consumer contexts using repertory grid technique. Journal of the Academy of Marketing Science, 39(6), 846–869. doi:10.1007/s11747-010-0219-0 385 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung Llamas-Alonso, M. R., Jiménez-Zarco, A. I., Martínez-Ruiz, M. P., & Dawson, J. (2009). Designing a Predictive Performance Measurement and Control System to Maximize Customer Relationship Management Success. Journal of Marketing Channels, 16(1), 1–41. doi:10.1080/10466690802147896 Malthouse, E. C., Haenlein, M., Skiera, B., Wege, E., & Zhang, M. (2013). Managing Customer Relationships in the Social Media Era: Introducing the Social CRM House. Journal of Interactive Marketing, 27(4), 270–280. doi:10.1016/j.intmar.2013.09.008 Mohan, S., Choi, E., & Min, D. (2008). Conceptual Modeling of Enterprise Application System Using Social Networking and Web 2.0 “Social CRM System.” In 2008 International Conference on Convergence and Hybrid Information Technology (pp. 237–244). IEEE. doi:10.1109/ICHIT.2008.263 Musser, J., & O’Reilly, T. (2006). Web 2.0 principles and best practices. O’Reilly Radar. Retrieved from http://radar.oreilly.com/research/web2-report.html Paniagua, J., & Sapena, J. (2014). Business performance and social media: Love or hate? Business Horizons. doi:10.1016/j.bushor.2014.07.005 Payne, A., Ballantyne, D., & Christopher, M. (2005). A stakeholder approach to relationship marketing strategy: The development and use of the “six markets” model. European Journal of Marketing, 39(7/8), 855–871. doi:10.1108/03090560510601806 Payne, A., & Frow, P. (2005). A strategic framework for customer relationship management. Journal of Marketing, 69(4), 167–176. Retrieved from http://journals.ama.org/doi/abs/10.1509/jmkg.2005.69.4.167 Payne, A., & Frow, P. (2006). Customer Relationship Management: from Strategy to Implementation. Journal of Marketing Management, 22(1-2), 135–168. doi:10.1362/026725706776022272 Payne, A., Storbacka, K., Frow, P., & Knox, S. (2009). Co-creating brands: diagnosing and designing the relationship experience. Journal of Business Research, 62(3), 379–389. Retrieved from http://www.sciencedirect.com/science/article/pii/S0148296308001562 Prahalad, C. K., & Ramaswamy, V. (2004). Co-creation experiences: The next practice in value creation. Journal of Interactive Marketing, 18(3), 5–14. doi:10.1002/dir.20015 Prasad, A., Green, P., & Heales, J. (2012). On IT governance structures and their effectiveness in collaborative organizational structures. International Journal of Accounting Information Systems, 13(3), 199–220. doi:10.1016/j.accinf.2012.06.005 386 SCRM performance dimensions: a resource-based view and dynamic capabilities perspective Pugh, D., & Chessell, M. (2013). Smarter Analytics: Driving Customer Interactions with the IBM Next Best Action Solution. IBM Redbooks. Ray, S., Kim, S. S., & Morris, J. G. (2014). The Central Role of Engagement in Online Communities. Information Systems Research, 25(3), 528–546. doi:10.1287/isre.2014.0525 Reinhold, O., & Alt, R. (2011). Analytical Social CRM: Concept and Tool Support. In Proceedings of the 24th Bled eConference (pp. 226–241). Reinhold, O., & Alt, R. (2012). Social Customer Relationship Management: State of the Art and Learnings from Current Projects. In Proceedings of the 25th Bled eConference (pp. 155–169). Sashi, C. M. (2012). Customer engagement, buyer-seller relationships, and social media. Management Decision, 50(2), 253–272. doi:10.1108/00251741211203551 Schaupp, L. C., & Bélanger, F. (2014). The Value of Social Media for Small Businesses. Journal of Information Systems, 28(1), 187–207. doi:10.2308/isys- 50674 Schmitt, B., & Zarantonello, L. (2013). Consumer experience and experiential marketing: a critical review. In Review of Marketing Research (Vol. 10, pp. 25– 61). Bingley: Emerald Group Publishing. doi:10.1108/S1548- 6435(2013)0000010006 Schultz, D., Malthouse, E. C., & Pick, D. (2012). From CM to CRM to CN2: A Research Agenda for the Marketing Communications Transition. In M. Eisend, T. Langner, & S. Okazaki (Eds.), Advances in Advertising Research (Vol. III) SE - 32 (Vol. III, pp. 421–432). Wiesbaden: Gabler Verlag. doi:10.1007/978-3-8349-4291- 3_32 Stefanou, C. J., Sarmaniotis, C., & Stafyla, A. (2003). CRM and customer-centric knowledge management: an empirical research. Business Process Management Journal, 9(5), 617–634. doi:10.1108/14637150310496721 Stieglitz, S., Dang-Xuan, L., Bruns, A., & Neuberger, C. (2014). Social Media Analytics. Wirtschaftsinformatik, 56(2), 101–109. doi:10.1007/s11576-014-0407-5 Trainor, K. J. (2012). Relating Social Media Technologies to Performance: A Capabilities-Based Perspective. Journal of Personal Selling and Sales Management, 32(3), 317–331. doi:10.2753/PSS0885-3134320303 Trainor, K. J., Andzulis, J. (Mick), Rapp, A., & Agnihotri, R. (2014). Social media technology usage and customer relationship performance: A capabilities-based examination of social CRM. Journal of Business Research, 67(6), 1201–1208. doi:10.1016/j.jbusres.2013.05.002 3 87 Nicolas Wittkuhn, Tobias Lehmkuhl, Torben Küpper, Reinhard Jung Van Bentum, R., & Stone, M. (2005). Customer relationship management and the impact of corporate culture — A European study. Journal of Database Marketing & Customer Strategy Management, 13(1), 28–54. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=buh&AN=20197691&site =ehost-live Vargo, S. L., Maglio, P. P., & Akaka, M. A. (2008). On value and value co-creation: A service systems and service logic perspective. European Management Journal, 26(3), 145–152. doi:10.1016/j.emj.2008.04.003 Vom Brocke, J., Simons, A., Niehaves, B., Riemer, K., Plattfaut, R., & Cleven, A. (2009). Reconstructing the Giant: On the Importance of Rigour in Documenting the Literature Search Process. In S. Newell, E. A. Whitley, N. Pouloudi, J. Wareham, & L. Mathiassen (Eds.), 17th European Conference on Information Systems (pp. 3226–3238). Retrieved from http://www.ecis2009.it/papers/ecis2009- 0566.pdf Wirtz, B. W., Schilke, O., & Ullrich, S. (2010). Strategic Development of Business Models. Long Range Planning, 43(2-3), 272–290. doi:10.1016/j.lrp.2010.01.005 Woodcock, N., Green, A., & Starkey, M. (2011). Social CRM as a business strategy. Journal of Database Marketing & Customer Strategy Management, 18(1), 50–64. doi:10.1057/dbm.2011.7 Zeng, D., Chen, H., Lusch, R., & Li, S. (2010). Social media analytics and intelligence. Intelligent Systems, IEEE, (November / December), 13–16. Retrieved from http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5678581 Zott, C. (2003). Dynamic capabilities and the emergence of intraindustry differential firm performance: insights from a simulation study. Strategic Management Journal, 24(2), 97–125. doi:10.1002/smj.288 388 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Crowdsourcing in Software Development: A State-of-the-Art Analysis Niklas Leicht University of St.Gallen, Switzerland niklas.leicht@unisg.ch David Durward Kassel University, Germany david.durward@uni-kassel.de Ivo Blohm University of St.Gallen, Switzerland ivo.blohm@unisg.ch Jan Marco Leimeister University of St.Gallen, Switzerland janmarco.leimeister@unisg.ch Abstract As software development cycles become shorter and shorter, while software complexity in- creases and IT budgets stagnate, many companies are looking for new ways of acquiring and sourcing knowledge outside their boundaries. One promising solution to aggregate know-how and manage large distributed teams in software development is crowdsourcing. This paper analyzes the existing body of knowledge regarding crowdsourcing in software development. As a result, we propose a fundamental framework with five dimensions to structure the existing insights of crowdsourcing in the context of software development and to derive a research agenda to guide further research. Keywords: Crowdsourcing, Software, Development, Literature Review 389 Leicht, Durward, Blohm & Leimeister 1 Introduction Faced with an increasingly dynamic environment, shorter product lifecycles, cost pressure, and an increasing complexity due to the rapid development of new software-based business models and a fragmented hardware market, companies are looking for new ways of acquiring and sourcing knowledge from outside the boundaries of their units, functions, or even outside their organization in order to develop software solutions (Jain 2010). On top of the continuous trend towards globalization and its focus on collaborative methods and infrastructure, it fosters the emergence of developing software in large distributed teams and communities (Boehm 2006; Stol and Fitzgerald 2014a). One solution to manage large distributed teams is crowdsourcing. With crowdsourcing, companies can reach out to the masses (Vukovic 2009) and open tasks to what Howe (2006) describes as “an undefined (…) network of people”. The term itself derives from the concept of the outsourcing of a corporate, company-internal task to an independent mass of people, the crowd (Howe 2008). IT industry leaders such as Fujitsu-Siemens (Füller et al. 2011), IBM (Bjelland and Wood 2008), or SAP (Blohm et al. 2011; Leimeister et al. 2009) already leveraged the “wisdom of the crowds” (Surowiecki 2005) for improving innovation management. Similarly, Lakhani et al. (2013) exhibit the tremendous potential of crowdsourcing in the domain of software development. They report on a programming contest in which about 75% of the submitted algorithms to solve an immunogenomic problem outperformed the industry standard while the total cost of the contest equaled 6000$. Extreme solutions were up to a thousand times faster than the industry standard. Software testing is another field of application in software development in which crowdsourcing is gaining importance. The World Quality Report (2014), the benchmark for software testing practices, indicates that more than half of the asked organizations either already employed crowdsourcing in their software testing process or planned to do so in 2014. However, research on crowdsourcing is still in its inception. So far, crowdsourcing research has predominantly focused on (1) conceptualizing the phenomenon and comparing and designing, coding, testing, and documenting software. We intend to tackle this issue by reviewing existing crowdsourcing literature with a structured and systematic literature review following Webster & Watson (2002) and Vom Brocke et al. (2009). Based on this review, we propose a framework that summarizes existing research on crowdsourced software development. Following this research goal, our paper contributes to crowdsourcing literature by providing a basis for future theory development while elaborating various avenues for future research. The remainder of this paper is structured as follows: Section two covers the literature review. Within this section, we first define the review scope and conceptualize the topic. Following that, we describe the literature search approach and introduce the literature framework. In section three, we present our findings in order to derive and discuss the research agenda which is presented in section four. Finally, we point out limitations and conclude the paper by summarizing the results of the literature review. 390 Crowdsourcing in Software Development: A State-of-the-Art Analysis contrasting it to related phenomena such as collective intelligence (Malone et al. 2010), human computation, or open innovation (Gassmann et al. 2010), (2) classifying socio-technical crowdsourcing systems with taxonomies and categorizations to identify the basic characteristics (Geiger et al. 2011; Rouse 2010), and (3) applying crowdsourcing in different domains such as innovation development or marketing (Brabham 2008; Burger-Helmchen and Penin 2010; Kittur et al. 2008; Zhao and Zhu 2012). The thereby generated insights provide first references for the management and organization of crowdsourcing initiatives. Although there are already numerous research projects examining crowdsourced software development, e.g., Lakhani et al. (2013), Nag et al. (2012), and Liu et al. (2012) on the application of crowdsourced software development or Murray-Rust et al. (2014) and Wu et al. (2013a; 2013b) on system and process design, there are much less as well as no structured insights on research of crowdsourced software development. Lacking are theories and approaches to gain a deeper understanding and to systematically use crowdsourcing in Literature Review This literature review is based on Vom Brocke et al.’s (2009) framework for reviewing scholarly literature and comprises five steps: (1) defining the review scope, (2) conceptualizing the topic, (3) searching for literature, (4) analyzing and synthesizing the literature, and (5) deriving a research agenda. 1.1 Definition of the Review Scope The first step of a rigorous literature review is the definition of the review scope for which we follow the taxonomy of Cooper (1988). The paper focuses on research outcomes and the applications of crowdsourced software development (1). The goal of the literature review is to build an integrative (2) overview of the existing body of knowledge to present the state of the art (4) as it addresses specialized scholars (5). Table 1 depicts the literature review scope. Characteristics Categories Research 1 Focus Research Methods Theories Applications Outcomes 2 Goal Integration Criticism Central Issues 3 Organization Historical Conceptual Methodological 4 Perspective Neutral Representation Espousal of Position 5 Audience Specialized Scholars General Scholars Practitioners General Public Exhaustive & 6 Coverage Exhaustive Representative Central/pivotal Selective Table 1: Definition of the Review Scope 1.2 Conceptualization of the Topic A rigor literature review has to “provide a working definition of key variables” (Webster and Watson 2002). This work focuses on crowdsourcing and software development. 391 Leicht, Durward, Blohm & Leimeister 1.2.1 Crowdsourcing Crowdsourcing describes a new form of outsourcing tasks, or more accurately, value creation activities and functions. The term itself is a neologism that combines crowd and outsourcing (Rouse 2010), introduced by Howe (2008), who defines crowdsourcing as “the act of taking a job traditionally per-formed by a designated agent (usually an employee) and outsourcing it to an undefined, generally large group of people in the form of an open call”. Whereas outsourcing describes the outplacement of specific corporate tasks to a designated third-party contractor or a certain institution, in crowdsourcing the tasks are allocated to an undefined mass of anonymous individuals, who are in turn rewarded for their effort of performing the tasks (Zogaj et al. 2014). In a crowdsourcing model, a firm or some type of institution first selects specific internal tasks it intends to crowdsource and subsequently broadcasts the underlying tasks online, i.e., via a crowdsourcing platform. In a second step, individuals (e.g., users registered on a crowdsourcing platform) self-select to work on the task solutions – either individually or in a collaborative manner – and subsequently submit the elaborated solutions via the crowdsourcing platform (Zogaj et al. 2014). The submissions are then assessed and – in case of successful completion – remunerated by the initiating organization. Hence, in a crowdsourcing model, at least two types of actors are engaged: the initiating organization that crowdsources specific tasks as well as the individuals from the crowd who perform these tasks. We denote the first entity as the crowdsourcer [“system owner” (Doan et al. 2011); “designated agent” (Howe 2006)]. The latter, namely the undefined contractors from the crowd, we label as crowdworkers since they perform the work (i.e., jobs or – more specifically – the tasks) that is outsourced by crowdsourcers. In most crowdsourcing initiatives, there is also a third type of agent: the crowdsourcing intermediary (also referred to as “crowdsourcing marketplace”; see e.g., Vukovic (2009) and Ipeirotis (2010). Crowdsourcing intermediaries mediate the process between the crowdsourcer and the crowdworkers by providing a platform for interaction between the parties. However, in some rare cases, the crowdsourcer establishes and hosts its own crowdsourcing platform such that an intermediary is not necessary. 1.2.2 Crowdsourced Software Development In an early definition, Robillard (1999) describes software development as the processing of knowledge in a very focused way as well as a progressive crystallization of knowledge into a language that can be read and executed by a computer. This language creation is increasingly taking place in a steady, irreversible trend toward the globalization of business, in particular in software-intensive high-technology businesses. Hence, software has become an essential component of almost any value chain, and success in business increasingly depends on using software as a competitive weapon (Herbsleb and Moitra 2001). In the era of cloud computing, mobile computing, collaboration, and big data, software development and its requirements are significantly changing. Organizations as well as the users of software are calling for an improved ease of use, shorter development cycles, and a better integration by lower overhead operations (Huhns et al. 2013). This leads to more flexible and effective ways to build software 392 Crowdsourcing in Software Development: A State-of-the-Art Analysis solutions such as crowdsourcing software development. This approach uses the online crowd to outsource (sub-) tasks including requirements, design, coding, testing, evolution, and documentation. Crowdsourcing software development represents a paradigm shift from conventional industrial software development to a crowdsourcing-based peer-production software development and can be seen as next-generation outsourcing or offshoring (Huhns et al. 2013). 1.3 Literature Search In order to identify relevant articles and to assure a rigorous, comprehensive, and traceable literature search, a systematic literature review was conducted (Vom Brocke et al. 2009). First, a journal search was executed, followed by a database search with keywords. Finally, a forward and backward search of citation indexes was conducted. The journal search is the first step as major contributions are likely to be found in leading journals (Webster and Watson 2002) as well as in proceedings of highly ranked conferences (Rowley and Slack 2004). For the journal search, leading journals from Information Systems (IS) and Software Engineering (SE) were considered. For information systems, these included: Information Systems Research (ISR), MIS Quarterly (MISQ), Journal of Information Systems (JIS), and the Journal of Management Information Systems (JMIS). For software engineering, the highest ranked journals according to the ISI Web of Science were chosen, i.e., IEEE Transactions on Software Engineering, Communications of the ACM, IEEE Software, and IEEE Computer. The selection of relevant conferences included the International Conference on Information Systems (ICIS), the European Conference on Information Systems (ECIS), and the American Conference on Information Systems (AMCIS) as well as the Hawaii International Conference on System Sciences (HICCS). For Software Engineering, the International Conference on Software Engineering (ICSE), Foundations of Software Engineering (FSE), International Test Conference (ITC), and Conference on Object-Oriented Programming Systems, Languages, and Applications (OOPSLA) were considered. Furthermore, the following databases were queried: EBSCOhost, Web of Science, ProQuest, ScienceDirect, as well as IEEE Xplore database, since the topic is also at the interface to software engineering. Core of a literature search is the keyword search. According to the above defined key variables, the keyword search was conducted in afore mentioned databases with the following search strings: (1) “crowdsourcing” AND “software development”, (2) “crowdsourcing” AND “software”, and (3) “crowdsourcing” AND “software engineering”, as well as (4) “crowd” AND “software”, (5) “crowd” AND “software development”, and (6) “crowd” AND “software engineering”. Additionally, the keyword search contained the following search strings in order to increase the coverage: (7) “crowdsourcing” AND “software testing” and (8) “crowdtesting”. The literature search closed with a forward and backward search (Levy and Ellis 2006). Table 2 depicts the detailed result auf the literature search. 393 Leicht, Durward, Blohm & Leimeister EBSCO Web of ProQuest Science Direct IEEE Xplore TOTAL Host(BSP) Science Search String Revi Revie Revie Revie Revie Revie Hits ew- Hits Hits Hits Hits Hits wed w-ed w-ed w-ed w-ed ed “crowdsourcing” AND 5 1 28 3 20 3 5 0 29 11 87 18 “software development” “crowdsourcing” AND 19 1 63 5 65 3 17 1 117 10 281 20 “software” “crowdsourcing” AND 3 0 3 1 10 1 2 0 44 7 62 9 “software engineering” “crowd” AND 21 0 183 3 42 2 25 2 431 5 702 12 “software" “crowd” AND “software 7 1 29 1 38 4 3 2 79 8 156 16 development” “crowd” AND “software 2 0 7 0 28 1 1 0 130 6 168 7 engineering” “crowdsourcing” AND 2 0 4 3 10 3 4 2 16 6 36 14 “software testing” “crowdtesting” 0 0 1 1 1 0 0 0 1 1 3 2 TOTAL 59 3 318 17 214 17 57 7 847 54 1495 98 Table 2: Results of the Literature Search per Database 1.4 Literature Analysis and Synthesis The literature review identified a total of 27 relevant papers. Considering the publication dates, it is no surprise that crowdsourcing in software development is at an early stage of scientific research, since crowdsourcing itself is still an emerging research topic. Only one paper was published before 2012 (Kazman and Chen 2009). More than 85% of all identified relevant papers were published in 2013 or later. Figure 1 depicts the publications per year. Another indication of the early stage of this field of research is that not a single paper was published in one of the major and leading journals. The articles were rather published in smaller and specialized journals or at conferences. Overall, more than two thirds of the relevant papers are from the field of software engineering. 394 Crowdsourcing in Software Development: A State-of-the-Art Analysis 16 15 12 8 8 4 3 1 0 <2012 2012 2013 2014 Figure 1: Publications per Year In order to synthesize the literature, appropriate categories need to be developed. This paper tackles this issue by developing categories based on existing literature on crowdsourcing in general. Based on Zhao and Zhu’s (2012) research roadmap, a key role-based perspective (Vukovic 2009; Zogaj et al. 2014), and applications of crowdsourcing in a software development context, the following categories were developed: (1) organization perspective, (2) intermediary perspective, (3) system perspective, (4) user perspective, and (5) application and evaluation. (1) Organization perspective In the archetypical crowdsourcing process (Vukovic 2009; Zogaj et al. 2014), organizations appear as the requester of a crowdsourcing task (crowdsourcer). This category sums up papers dealing with the organizational implementation, its according challenges, as well as the development of necessary capabilities to harness crowdsourcing in an enterprise environment. (2) Intermediary perspective The intermediary manages the crowdsourcing process and thereby its customers, crowd, and technology (Zogaj et al. 2014). This category sums up papers addressing process and design requirements, an according evaluation, as well as other managerial challenges the intermediary faces in crowdsourced software development tasks. (3) System perspective Crowdsourcing systems are socio-technical systems to enable and support the crowdsourcing process (Zhao and Zhu 2012). This category sums up papers dealing with the requirements or the design of crowdsourcing platforms for software development. Since software development tasks are way more complex than simple tasks that are frequently crowdsourced, it might take other design principles to develop a system tailored for software development tasks. (4) User perspective The participants of crowdsourcing initiatives (crowdworkers) are without a doubt an essential part and therefore need to be treated as a partner. By means of crowdsourcing, participants can expand their working experiences or even turn their hobbies into something beneficial (Zhao and Zhu 2012). This category sums up papers dealing with user motivation, payoff, and other user-centered aspects. 395 Leicht, Durward, Blohm & Leimeister (5) Application and evaluation The last category sums up papers which apply crowdsourcing in different contexts to evaluate its performance and/or highlight application possibilities for crowdsourcing in different software development contexts and stages. 2 Findings Overall, it can be stated that existing research in the field of crowdsourcing software development mainly focuses on crowdsourcing systems and applications. Almost 60% of the investigated literature dealt with a particular IT system and its design (system perspective). About two fifths of the research dealt with the application of crowdsourcing in software development. Only one paper addresses the user perspective in crowdsourced software development. Table 3 depicts the detailed results. Paper Organization Intermediary System User Application Amini et al. (2012) x x Chen and Luo (2014) x Dolstra et al. (2013) x x Hossfeld et al. (2014) x x Hu and Wu (2014) x x Jayakanthan and Sundararajan (2013) x Kazman and Hong-Mei Chen (2009) x Lakhani et al. (2013) x LaToza et al. (2013) x Li et al. (2013) x Liu et al. (2012) x Mäntylä and Itkonen (2013) x Mao et al. (2013) x x Murray-Rust et al. (2014) x Musson et al. (2013) x Nag et al. (2012) x x Pastore et al. (2013) x Peng et al. (2014) x Ponzanelli et al. (2013) x x Stol and Fitzgerald (2014a) x Stol and Fitzgerald (2014b) x Stol and Fitzgerald (2014c) x Tajedin and Nevo (2013) x Tung and Tseng (2013) x Wu et al. (2013a) x Wu et al. (2013b) x Zogaj et al. (2013) x x TOTAL (n=27) 3 (11.1%) 3 (11.1%) 16 (59.3%) 1 (3.7%) 12 (44.4%) Table 3: Literature Synthesis 396 Crowdsourcing in Software Development: A State-of-the-Art Analysis (1) Organization perspective So far, most notably Stol and Fitzgerald (2014a; 2014b; 2014c) examined crowdsourced software development from an organization’s point of view. Their contribution to the understanding of crowdsourcing software development is twofold. First, they point out potential benefits and thus deliver a first explanation of why companies tend to use crowdsourcing in this area. The benefits combine traditional outsourcing benefits such as cost reduction, a faster time-to-market, and higher quality (Dibbern et al. 2004) with benefits of crowdsourcing such as creativity, increased openness, and diverse solutions (Afuah and Tucci 2012; Leimeister 2010). Second, they develop a framework of key concerns regarding the application of crowdsourcing for enterprises in software development. According key concerns are (1) task decomposition, (2) coordination and communication, (3) planning & scheduling, (4) quality assurance, (5) knowledge & IP, and (6) motivation & remuneration (Stol and Fitzgerald 2014a). (2) Intermediary perspective Zogaj et al. (2014) deliver a profound overview of the challenges of an intermediary from a managerial point of view. In their case study, they explicitly address the challenges in managing a crowd, the crowdsourcing process, as well as the crowdsourcing platform. Besides managing the process itself, the pivotal challenge for intermediaries constitutes building virtual teams and fostering collaboration among the crowdworkers (Peng et al. 2014). Furthermore, Mao et al. (2013) address the pricing of programming competitions, finding that the main antecedents of project pricing are whether the task is a component update or a new component, the size and the amount of illustrations in the specification document, as well as the overall size of the project and the posted reward amount. (3) System perspective Current literature mostly focuses on a system perspective. That means the development and derivation of specific development models, design principles for platforms or processes to enable crowdsourcing in diverse fields of application in software development. Therefore, the majorities of the papers have a technical perspective. Kazman & Chen (2009) and LaToza et al. (2013) propose a specific software development model tailored for crowdsourcing. Moreover, the human-machine interaction process is crucial for successful crowdsourcing campaigns. Murray-Rust et al. (2014) elaborate two collaboration models for community-based development of software and provide a conceptual model for combining process models with crowdsourced teams. Other research deals with success factors of crowdsourcing projects (Li et al. 2013; Tajedin and Nevo 2013) or process design for specific applications and purposes (Amini et al. 2012; Pastore et al. 2013; Tung and Tseng 2013). Furthermore, Wu et al. (2013a; 2013b) analyze software crowdsourcing processes by examining their key characteristics. They propose a novel evaluation framework for software crowdsourcing processes. Hossfeld et al. (2014) present key issues in the field of QoE-Testing (quality of experience) as they apply a QoE-Test and provide design guidelines for crowdtesting in this field of application. 397 Leicht, Durward, Blohm & Leimeister (4) User perspective As defined in the synthesis, this category clusters papers investigating the motivation and behavior of crowdworkers. Hu & Wu (2014) apply a game-theoretic approach to better understand the competitive behavior of crowdworkers in software development challenges. (5) Application and evaluation There are multiple examples for the application and evaluation addressing multiple parts of software development stages and functions. This research reaches from algorithm development (Lakhani et al. 2013), to embedded software for space robotics (Nag et al. 2012), or software testing (Mäntylä and Itkonen 2013). Further, Chen and Luo (2014) apply crowdsourced software testing in an educational context. As part of their studies, students had to test several web and mobile applications. Liu et al. (2012) compare traditional laboratory- based usability testing with crowdsourced usability testing, indicating that the acquisition of testers through crowdsourcing is much easier at significantly lower costs . Contrariwise, the received feedback per participant was less informative. Crowdsourcing also seems to be a promising approach to test graphical user interfaces or to evaluate mobile applications (Amini et al. 2012; Dolstra et al. 2013). In the domain of documenting software code, Ponzanelli et al. (2013) research the case of “Stack Overflow”, the world’s largest language-independent collaboratively edited question and answer site for programmers. They propose a new interaction interface for increasing the productivity of software documentation. The power of crowdsourcing has also been used to monitor software performance. One major advantage is the “real world setting” in which different network environments and bandwidths are accessible, which are not to be covered in laboratory tests (Musson et al. 2013). Jayakanthan & Sundararajan (2012) introduce a prototype for a corporate crowdsourcing solution at TCS, one of the largest IT consulting and software development companies worldwide. The crowdsourcing system will unify three modes of crowdsourcing as the crowdsourcer can select whether to choose a single expert among crowdworkers performing the task (e.g., when special knowledge is required or the task is critical), recruit a group of crowdworkers (e.g., for software testing), or create a competition to choose the best solution among the submissions (e.g., for coding). 3 Discussion 3.1 Research Agenda Although various scholars have examined crowdsourced software development projects, our literature review reveals that research in this regard is still at an early stage. Existing literature most notably focuses on the design of crowdsourcing platforms from a system perspective providing only little generalization and approaches to gain a broad perspective. At first sight, there is obviously more research needed on the user perspective since we found only one paper which addresses this stream within the literature. But it is important to note 398 Crowdsourcing in Software Development: A State-of-the-Art Analysis that a wide body of research regarding the incentives and motivation of crowdworkers (Muhdi and Boutellier 2011; Pilz and Gewald 2013), qualification, as well as the impact on work conditions and labor rights (Brabham 2012) already exists in broader crowdsourcing literature. However, most of this research deals with simplistic “micro tasks” or highly creative tasks such as idea generation. Thus, future research has to validate that the findings are also applicable to software development – a more complicated knowledge task. Crowdsourced software development has been applied and evaluated in a number of studies. So far, there is no in- depth knowledge on the basic conditions of crowdsourcing projects to leverage this potential. Software development is a very complex process with diverse stages and tasks with very different requirements, complexity, modularity, and structures – things which have all been found to determine the effectivity of crowdsourcing (Afuah and Tucci 2012; Stol and Fitzgerald 2014a). Future research should address this aspect by examining which tasks can be crowdsourced and how crowdsourcing projects should be structured in terms of task decomposition with respect to the knowledge intensity and a high degree of complexity in software development. A third possible stream is based on these insights and addresses the organization perspective. The literature review has shown first activity regarding this topic with Stol & Fitzgerald (2014a) identifying six “key concerns” in crowdsourcing software development. To eliminate these concerns, it is not sufficient to solely understand at which stage in the development process they can crowdsource tasks and how to structure this work. Just like outsourcing, crowdsourcing is a whole new form of organizing work and therefore requires a different process model, as well as governance structures and control mechanisms in software development projects. To systematically enable organizations to conduct crowdsourcing projects within their existing process and framework, we propose to develop a reference model for crowdsourcing projects which addresses the key concerns and guides organizations through the crowdsourcing process. From an intermediary perspective, Zogaj et al. (2014) discuss the challenges an intermediary faces in the crowdsourcing process. As these intermediaries appear as interface between the crowdsourcer and the crowdworkers, two research fields unfold. The first overlaps with the user perspective as it is crucial for the intermediary to promote a motivated and active crowd. Further research should investigate the question of how to motivate the users, especially for less entertaining tasks such as software documentation. On the other side, the intermediary is a vital part in the crowdsourcing process from an organization’s point of view. Further research should investigate the intermediary’s role and support in crowdsourcing software development. In this regard, investigating crowdsourcing intermediaries as two-sided markets may be of particular interest. 3.2 Limitations Reflecting the paper, two limitations are worth mentioning. First, the results only rely on scientific literature and thus lack insights from practice. Second, this research only focuses on core crowdsourcing literature as the search strings only contained “crowd” or “crowdsourcing”. However, there are other streams of research that might be suitable to 399 Leicht, Durward, Blohm & Leimeister address some of the key issues. For instance, the concept of open source software development is a more mature research field and overlaps with the concept of crowdsourcing. Further research should target this research stream in order to integrate and enable a deeper understanding of how collaborative work in software development can be organized and processed. 4 Conclusion In summary, the research on crowdsourcing in software development is still limited, despite its potential and gaining importance in organizations. In this paper, we reviewed the existing body of literature regarding crowdsourced software development. In so doing, the contribution of this paper is twofold. First we provide an initial framework summarizing the key aspects of crowdsourcing and thus contribute to an enhanced development of a theoretical understanding of crowdsourcing. Second, our literature review points out gaps in the literature that could be addressed in future research. References Afuah, A., and Tucci, C.L. 2012. "Crowdsourcing as a Solution to Distant Search," Academy of Management Review (37:3), pp. 355-375. Amini, S., Lin, J., Hong, J., Lindqvist, J., and Zhang, J. 2012. "Towards Scalable Evaluation of Mobile Applications through Crowdsourcing and Automation," CMU-CyLab-12-006, Carnegie Mellon University). Bjelland, O.M., and Wood, R.C. 2008. "An inside View of Ibm's' Innovation Jam'," MIT Sloan management review (50:1), pp. 32-40. Blohm, I., Bretschneider, U., Leimeister, J.M., and Krcmar, H. 2011. "Does Collaboration among Participants Lead to Better Ideas in It-Based Idea Competitions? An Empirical Investigation," International Journal of Networking and Virtual Organisations (9:2), pp. 106-122. Boehm, B. 2006. "A View of 20th and 21st Century Software Engineering," Proceedings of the 28th international conference on Software engineering: ACM, pp. 12-29. Brabham, D.C. 2008. "Crowdsourcing as a Model of Problem Solving," Convergence: The International Journal of Research into New Media Technologies (14:1), pp. 75-90. Brabham, D.C. 2012. "The Myth of Amateur Crowds: A Critical Discourse Analysis of Crowdsourcing Coverage," Information, Communication & Society (15:3), pp. 394-410. Buenen, M., and Makarand, T. 2014. "World Quality Report 2014-2015." Burger-Helmchen, T., and Penin, J. 2010. "The Limits of Crowdsourcing Inventive Activities,"). Chen, Z., and Luo, B. 2014. "Quasi-Crowdsourcing Testing for Educational Projects," Companion Proceedings of the 36th International Conference on Software Engineering: ACM, pp. 272-275. Cooper, H.M. 1988. "Organizing Knowledge Syntheses: A Taxonomy of Literature Reviews," Knowledge in Society (1:1), pp. 104-126. Dibbern, J., Goles, T., Hirschheim, R., and Jayatilaka, B. 2004. "Information Systems Outsourcing: A Survey and Analysis of the Literature," ACM SIGMIS Database (35:4), pp. 6-102. 400 Crowdsourcing in Software Development: A State-of-the-Art Analysis Doan, A., Ramakrishnan, R., and Halevy, A.Y. 2011. "Crowdsourcing Systems on the World- Wide Web," Communications of the ACM (54:4), pp. 86-96. Dolstra, E., Vliegendhart, R., and Pouwelse, J. 2013. "Crowdsourcing Gui Tests," Software Testing, Verification and Validation (ICST), 2013 IEEE Sixth International Conference on: IEEE, pp. 332-341. Füller, J., Hutter, K., and Faullant, R. 2011. "Why Co‐Creation Experience Matters? Creative Experience and Its Impact on the Quantity and Quality of Creative Contributions," R&D Management (41:3), pp. 259-273. Gassmann, O., Enkel, E., and Chesbrough, H. 2010. "The Future of Open Innovation," R&D Management (40:3), pp. 213-221. Geiger, D., Seedorf, S., Schulze, T., Nickerson, R.C., and Schader, M. 2011. "Managing the Crowd: Towards a Taxonomy of Crowdsourcing Processes,"). Herbsleb, J.D., and Moitra, D. 2001. "Global Software Development," Software, IEEE (18:2), pp. 16-20. Hoßfeld, T., Keimel, C., Hirth, M., Gardlo, B., Habigt, J., Diepold, K., and Tran-Gia, P. 2014. "Best Practices for Qoe Crowdtesting: Qoe Assessment with Crowdsourcing," Multimedia, IEEE Transactions on (16:2), pp. 541-558. Howe, J. 2006. "The Rise of Crowdsourcing," Wired magazine (14:6), pp. 1-4. Howe, J. 2008. Crowdsourcing: How the Power of the Crowd Is Driving the Future of Business. Random House. Hu, Z., and Wu, W. 2014. "A Game Theoretic Model of Software Crowdsourcing," Service Oriented System Engineering (SOSE), 2014 IEEE 8th International Symposium on: IEEE, pp. 446-453. Huhns, M.N., Li, W., and Tsai, W.-T. 2013. "Cloud-Based Software Crowdsourcing," Citeseer. Ipeirotis, P.G. 2010. "Analyzing the Amazon Mechanical Turk Marketplace," XRDS: Crossroads, The ACM Magazine for Students (17:2), pp. 16-21. Jain, R. 2010. "Investigation of Governance Mechanisms for Crowdsourcing Initiatives,"). Jayakanthan, R., and Sundararajan, D. 2012. "Enterprise Crowdsourcing Solution for Software Development in an Outsourcing Organization," in Current Trends in Web Engineering. Springer, pp. 177-180. Kazman, R., and Chen, H.-M. 2009. "The Metropolis Model a New Logic for Development of Crowdsourced Systems," Communications of the ACM (52:7), pp. 76-84. Kittur, A., Chi, E.H., and Suh, B. 2008. "Crowdsourcing User Studies with Mechanical Turk," CHI '08 Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (453-456). Lakhani, K.R., Boudreau, K.J., Loh, P.-R., Backstrom, L., Baldwin, C., Lonstein, E., Lydon, M., MacCormack, A., Arnaout, R.A., and Guinan, E.C. 2013. "Prize-Based Contests Can Provide Solutions to Computational Biology Problems," Nature biotechnology (31:2), pp. 108-111. LaToza, T.D., Towne, W.B., Van Der Hoek, A., and Herbsleb, J.D. 2013. "Crowd Development," Cooperative and Human Aspects of Software Engineering (CHASE), 2013 6th International Workshop on: IEEE, pp. 85-88. Leimeister, J.M. 2010. "Collective Intelligence," Business & Information Systems Engineering (2:4), pp. 245-248. Leimeister, J.M., Huber, M., Bretschneider, U., and Krcmar, H. 2009. "Leveraging Crowdsourcing: Activation-Supporting Components for It-Based Ideas Competition," Journal of management information systems (26:1), pp. 197-224. Levy, Y., and Ellis, T.J. 2006. "A Systems Approach to Conduct an Effective Literature Review in Support of Information Systems Research," Informing Science: International Journal of an Emerging Transdiscipline (9:1), pp. 181-212. 401 Leicht, Durward, Blohm & Leimeister Li, K., Xiao, J., Wang, Y., and Wang, Q. 2013. "Analysis of the Key Factors for Software Quality in Crowdsourcing Development: An Empirical Study on Topcoder. Com," Computer Software and Applications Conference (COMPSAC), 2013 IEEE 37th Annual: IEEE, pp. 812-817. Liu, D., Bias, R.G., Lease, M., and Kuipers, R. 2012. "Crowdsourcing for Usability Testing," Proceedings of the American Society for Information Science and Technology (49:1), pp. 1-10. Malone, T.W., Laubacher, R., and Dellarocas, C. 2010. "The Collective Intelligence Genome," IEEE Engineering Management Review (38:3), p. 38. Mäntylä, M.V., and Itkonen, J. 2013. "More Testers–the Effect of Crowd Size and Time Restriction in Software Testing," Information and Software Technology (55:6), pp. 986- 1003. Mao, K., Yang, Y., Li, M., and Harman, M. 2013. "Pricing Crowdsourcing-Based Software Development Tasks," Proceedings of the 2013 International Conference on Software Engineering: IEEE Press, pp. 1205-1208. Muhdi, L., and Boutellier, R. 2011. "Motivational Factors Affecting Participation and Contribution of Members in Two Different Swiss Innovation Communities," International Journal of Innovation Management (15:03), pp. 543-562. Murray-Rust, D., Scekic, O., Truong, H.-L., Robertson, D., and Dustdar, S. 2014. "A Collaboration Model for Community-Based Software Development with Social Machines," Collaborative Computing: Networking, Applications and Worksharing (CollaborateCom), 2014 International Conference on: IEEE, pp. 84-93. Musson, R., Richards, J., Fisher, D., Bird, C., Bussone, B., and Ganguly, S. 2013. "Leveraging the Crowd: How 48,000 Users Helped Improve Lync Performance," Software, IEEE (30:4), pp. 38-45. Nag, S., Heffan, I., Saenz-Otero, A., and Lydon, M. 2012. "Spheres Zero Robotics Software Development: Lessons on Crowdsourcing and Collaborative Competition," Aerospace Conference, 2012 IEEE: IEEE, pp. 1-17. Pastore, F., Mariani, L., and Fraser, G. 2013. "Crowdoracles: Can the Crowd Solve the Oracle Problem," International Conference on Software Testing, Verification and Validation (ICST). Peng, X., Babar, M.A., and Ebert, C. 2014. "Collaborative Software Development Platforms for Crowdsourcing," IEEE software (31:2), pp. 30-36. Pilz, D., and Gewald, H. 2013. "Does Money Matter? Motivational Factors for Participation in Paid-and Non-Profit-Crowdsourcing Communities,"). Ponzanelli, L., Bacchelli, A., and Lanza, M. 2013. "Leveraging Crowd Knowledge for Software Comprehension and Development," Proceedings of the 2013 17th European Conference on Software Maintenance and Reengineering: IEEE Computer Society, pp. 57-66. Robillard, P.N. 1999. "The Role of Knowledge in Software Development," Communications of the ACM (42:1), pp. 87-92. Rouse, A.C. 2010. "A Preliminary Taxonomy of Crowdsourcing," Australian Conference on Information Systems (ACIS), 1-3 Dec 2010, Brisbane. Rowley, J., and Slack, F. 2004. "Conducting a Literature Review," Management Research News (27:6), pp. 31-39. Stol, K.-J., and Fitzgerald, B. 2014a. "Two's Company, Three's a Crowd: A Case Study of Crowdsourcing Software Development," ICSE, pp. 187-198. 402 Crowdsourcing in Software Development: A State-of-the-Art Analysis Stol, K.-J., and Fitzgerald, B. 2014b. "Researching Crowdsourcing Software Development: Perspectives and Concerns," Proceedings of the 1st International Workshop on CrowdSourcing in Software Engineering: ACM, pp. 7-10. Stol, K.-J., and Fitzgerald, B. 2014c. "Research Protocol for a Case Study of Crowdsourcing Software Development." Surowiecki, J. 2005. The Wisdom of Crowds. Random House LLC. Tajedin, H., and Nevo, D. 2013. "Determinants of Success in Crowdsourcing Software Development," Proceedings of the 2013 annual conference on Computers and people research: ACM, pp. 173-178. Tung, Y.-H., and Tseng, S.-S. 2013. "A Novel Approach to Collaborative Testing in a Crowdsourcing Environment," Journal of Systems and Software (86:8), pp. 2143-2153. Vom Brocke, J., Simons, A., Niehaves, B., Riemer, K., Plattfaut, R., and Cleven, A. 2009. "Reconstructing the Giant: On the Importance of Rigour in Documenting the Literature Search Process," ECIS, pp. 2206-2217. Vukovic, M. 2009. "Crowdsourcing for Enterprises," Services-I, 2009 World Conference on: IEEE, pp. 686-692. Webster, J., and Watson, R.T. 2002. "Analyzing the Past to Prepare for the Future: Writing a Literature Review," Management Information Systems Quarterly (26:2), p. 3. Wu, W., Tsai, W.-T., and Li, W. 2013a. "An Evaluation Framework for Software Crowdsourcing," Frontiers of Computer Science (7:5), pp. 694-709. Wu, W., Tsai, W.T., and Li, W. 2013b. "Creative Software Crowdsourcing: From Components and Algorithm Development to Project Concept Formations," International Journal of Creative Computing (1:1), pp. 57-91. Zhao, Y., and Zhu, Q. 2012. "Evaluation on Crowdsourcing Research: Current Status and Future Direction," Information Systems Frontiers), pp. 1-18. Zogaj, S., Bretschneider, U., and Leimeister, J.M. 2014. "Managing Crowdsourced Software Testing: A Case Study Based Insight on the Challenges of a Crowdsourcing Intermediary," Journal of Business Economics (84:3), pp. 375-405. 403 THIS PAGE IS INTENTIONALY LEFT BLANK 404 THIS PAGE IS INTENTIONALY LEFT BLANK 405 THIS PAGE IS INTENTIONALY LEFT BLANK 406 THIS PAGE IS INTENTIONALY LEFT BLANK 407 THIS PAGE IS INTENTIONALY LEFT BLANK 408 THIS PAGE IS INTENTIONALY LEFT BLANK 409 THIS PAGE IS INTENTIONALY LEFT BLANK 410 THIS PAGE IS INTENTIONALY LEFT BLANK 411 THIS PAGE IS INTENTIONALY LEFT BLANK 412 THIS PAGE IS INTENTIONALY LEFT BLANK 413 THIS PAGE IS INTENTIONALY LEFT BLANK 414 THIS PAGE IS INTENTIONALY LEFT BLANK 415 THIS PAGE IS INTENTIONALY LEFT BLANK 416 THIS PAGE IS INTENTIONALY LEFT BLANK 417 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Social CRM Performance Model: An Empirical Evaluation Torben Küpper University of St.Gallen, Switzerland torben.kuepper@unisg.ch Tobias Lehmkuhl University of St.Gallen, Switzerland tobias.lehmkuhl@unisg.ch Nicolas Wittkuhn University of St.Gallen, Switzerland nicolas.wittkuhn@student.unisg.ch Alexander Wieneke University of St.Gallen, Switzerland alexander.wieneke@unisg.ch Reinhard Jung University of St.Gallen, Switzerland reinhard.jung@unisg.ch Abstract This paper presents an empirical investigation of a Social CRM performance model within an organizational perspective. A constraining factor regarding the implementation of Social CRM and the achievement of its objectives is the lack of an appropriate performance model. Little research has been conducted on a corresponding holistic approach and on the development of formative performance constructs. To address this gap, the article develops and empirically evaluates a Social CRM performance model, including an infrastructure-, process-, customer- and organizational performance construct. The data is analyzed using a structural equation model with a surveying sample of 126 marketing, communication and IT decision makers. The results show that infrastructure performance has an indirect, process performance a direct and customer performance no association with organizational performance. The Social CRM performance model generates deeper insights into a company’s performance interrelationship and provides a control system, in order to assess Social CRM activities and enhance organizational performance. 418 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung Keywords: Social CRM, Social CRM Performance, Social CRM Performance Model, Empirical Performance Model 1 Introduction Social Customer Relationship Management (Social CRM) deals with the integration of Web 2.0 and Social Media into CRM (Lehmkuhl and Jung 2013). Social CRM is a rising phenomenon, leading to a new scientific paradigm (Askool and Nakata 2011). It is defined as “[…] a philosophy and a business strategy, supported by a technology platform, business rules, processes and social characteristics, designed to engage the customer in a collaborative conversation in order to provide mutually beneficial value in a trusted and transparent business environment” (Greenberg 2010). Gartner has identified Social CRM as one of the top innovation-triggered themes in the next five to seven years (Alvarez 2013). Given that Social CRM is defined as a business strategy, its implementation requires holistic “transformational efforts among all organizational parts” (Lehmkuhl and Jung 2013). Particularly the implementation of Social CRM has the potential to provide mutually beneficial value for a company and its customers. Today, companies transform their business by applying new strategies, conducting organizational change, and purchasing new Social CRM technologies to achieve competitive business benefits (Trainor et al. 2014). Yet, companies implement Social CRM rather warily due to the lack of a holistic performance model, which allows companies to assess Social CRM activities and enhance organizational performance (e.g., increase brand awareness +10%). A literature review by Küpper et al. (2014) focuses on the current state of knowledge for Social CRM performance measures and reveals the lack of clearly defined and robust constructs and corresponding formative indicators. Previous work covers CRM performance measurement models, aiming at developing a balanced score card (Grabner-Kraeuter et al. 2007; Jain, Jain, and Dhar 2003; Kim and Kim 2009; Kim, Suh, and Hwang 2003; Llamas-Alonso et al. 2009; Sedera and Wang 2009; Wang, Sedera, and Tan 2009). Other research approaches test the interrelated association of different performance constructs empirically within the context of CRM (e.g., Jayachandran et al. 2013; Coltman et al. 2011; Reinartz et al. 2004; Roh et al. 2005; Keramati et al. 2010). The current articles to Social CRM focus on the conceptualization of Social CRM performance measures (Küpper et al. 2015; Trainor 2012) or evaluate individual Social CRM performance constructs (e.g., Trainor et al. 2014). Given the novelty of the topic and lack of research, no article investigates a holistic Social CRM performance model, i.e., including different dimensions (e.g., infrastructure, processes). Therefore, the objective of the article is to develop and evaluate a Social CRM performance model within an organizational perspective. The corresponding research question (RQ) is as follows: RQ: Which performance constructs for Social CRM have a significant influence on organizational performance? To achieve the stated objective, the article develops and evaluates a structural model, deriving five hypotheses from current literature. Accordingly, data from a survey sample of 126 marketing, communication and IT decision makers are analyzed through 419 Social CRM Performance Model: An Empirical Evaluation a structural equation model, as proposed by Hair et al. (2013), so as to answer the RQ. The result shows that two of three constructs influence organizational performance. The Social CRM performance model constitutes a scientific contribution as well as practical implication. The practical implication is given by providing a control system, in order to assess Social CRM activities and enhance organizational performance. The rigorous methodology enables researchers to adopt and apply the model as well as the new constructs and indicators for their research. The remainder of the paper is structured as follows. Section 2 presents the theoretical framing, including the conceptual background and the derived hypotheses of the article. Next, a methodology is given. Section 4 highlights the results of the Social CRM performance model, regarding the support as well as not support hypotheses. Section 5, presents the discussion and highlights scientific contributions and practical implications. The final section presents the limitations and outlines further research approaches. 2 Theoretical Framing 2.1 Conceptual Background To the best of our knowledge, this article contributes the first holistic empirically evaluated performance model for Social CRM. Due to the definition of Social CRM, the obvious related context is on CRM. Related performance measurement models shall be adopted to develop a conceptual Social CRM performance model. An overview of performance measurement models in CRM literature is presented in Table 1. Authors Levela Typb Scope Relationshipc Background Ind. Org. Con. Emp. Part. Holist. N.-cas. Cas. CRM SCRM Kim and Kim (2009) x x x x x Kim et al. (2003) x x x x x Öztayşi, Sezgin, et al. (2011) x x x x x Öztayşi, Kaya, et al. (2011) x x x x x Kimiloglu and Zarali (2009) x x x x x Llamas-Alonso et al. (2009) x x x x x Zinnbauer and Eberl (2005) x x x x x Shafia et al. (2011) x x x x x Lin et al. (2006) x x x x x Grabner-Kraeuter et al. (2007) x x x x x Jain et al. (2003) x x x x x Wang et al. (2009) x x x x x Sedera & Wang (2009) x x x x x Sum 1 12 8 5 4 9 12 1 13 0 This article x x x x x Ind. = Individual Level; Org. = Organizational Level; Con. = Conceptual; Emp. = Empirical; Part. = Partial; Holist. = Holistic; N.-cas. Rel. = Non-causal Relationships; Cas. Rel. = Causal Relationship; a Level of Analysis; b Type of validated model; c Development of relationships between the mentioned dimensions Table 1: Overview of performance measurement models in literature Kim and Kim's (2009) performance measurement model is adopted for the current research based on three reasons, covering scientific and practical aspects. First, after having conducted a rigorous and in-depth literature review on different performance models and performance measures for Social CRM, the model by Kim and Kim appears most holistic and well balanced. This impression is further support by the fact that it is published within a high-ranked journal and widely used, providing a high degree of 420 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung external validity1. Second, the authors derived conceptually causal interrelationships between its dimensions (cf. Table 1), which are a valuable approach to develop a performance model (e.g., focusing on a quantitative evaluation with a structural equation model). Lastly, the model has been well received by practitioners: within two focus groups, representatives from companies have classified Social CRM-specific objectives into the different constructs of the performance measurement model, showing its high feasibility and comprehensiveness as a management tool. In a final step, these practitioners also have created exemplified metrics for each performance measure, using these metrics for application in the corresponding departments of their companies, again stressing the usefulness of the model for application in real-life. The corresponding performance measurement model adopts a company perspective and includes four dimensions (i.e., constructs), namely (1) infrastructure performance, (2) process performance, (3) customer performance, and (4) organizational performance. The previous literature review (Küpper et al. 2014), based on a systematic research process (vom Brocke et al. 2009), was conducted to derive performance measures and to classify them within the constructs of the performance measurement model, as recommended by Kim and Kim (2009). Additionally, 15 semi-structured interviews identifies 25 Social CRM performance measures (Küpper et al. 2015). After another evaluation (e.g., discussing the results), two measures are removed and eight sub- dimensions are built to separate the performance measures in detail (i.e., each of the four constructs captures two sub-dimensions). To sum up, Table 2 presents the four adopted constructs, the eight derived sub-dimensions and the 23 performance measures in the context of Social CRM. Constructs (dimensions) Sub-dimensions Performance Measures ID Employee Commitment IN1 Cultural Performance Infrastructure Cultural Readiness IN2 Performance Online Brand Communities IN3 IT Performance IT-Readiness IN4 Customer Orientation PR1 Company-wide Performance Social Selling PR2 Multi-Channel and Ubiquitous Interaction PR3 Process Customer Insights PR4 Performance Market and Customer Segmentation PR5 Department-specific Customer Co-Creation PR6 Performance Customer Interaction PR7 Target-Oriented Customer Events PR8 Peer-to-Peer-Communication CU1 Indirect Customer Performance Customer-Based Relationship Performance CU2 Customer Customer Loyalty CU3 Performance Personalized Product and Services CU4 Direct Customer Performance Customer Convenience CU5 New Product Performance OR1 Monetization Performance Customer Lifetime Value OR2 Organizational Financial Benefits OR3 Performance Business Optimization OR4 Intangible Performance Brand Awareness OR5 Competitive Advantage OR6 Table 2: Dimensions of Social CRM performance 1 It is the most cited article for the abovementioned CRM performance measurement models, based on Google Scholar in October 2014. 421 Social CRM Performance Model: An Empirical Evaluation 2.2 Hypotheses Development and Conceptual Model A current analysis of the academic literature yields a total of 101 articles. The focus of the analysis is on performance models (CRM background) with an empirical investigation, identifying significant effects. After analyzing (reading) title, abstract and introduction and eliminating duplets, 29 relevant articles are identified. The analysis of the relevant articles, containing the four constructs (including the 23 measures), reveals five hypotheses, which yield a conceptual Social CRM performance model. Figure 1 presents an overview of all investigated direct, significant interrelationships of the conceptual Social CRM performance model. Figure 1: Conceptual Social CRM performance model (references are listed in the appendix) 2.2.1 Infrastructure Performance The infrastructure performance construct describes activities and/or results of infrastructural aspects (Neely, Gregory, and Platts 1995), which includes an IT dimension (e.g., IT-Readiness) and a cultural dimension (e.g., employee commitment). Due to cultural integration and the implementation of, e.g., an IT-infrastructure, employees are able to communicate in a more customer-oriented way and the company is able to monitor their customers, in order to generate new customer insights. The reviewed literature especially reveals that infrastructure performance has an association with process performance. This conclusion is supported by Peltier et al. (2013), Kim (2008), and Keramati et al. (2010), which found positive significant relationships between a cultural dimension and process performance within the context of CRM. Positive and significant relationship for the IT perspective to process performance within CRM, is supported by the contributions of Chuang and Lin (2013), Ernst et al. (2011), Lee et al. (2010), Wang and Feng (2012). Thus, the first hypothesis is as follows: H1: Infrastructure performance has a positive association with process performance within the context of Social CRM. 422 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung Additionally, the literature also supports an association of infrastructure performance with customer performance. Especially, IT enables organizations to interact more effectively and efficiently with customers (Trainor et al. 2014). The results of Ahearne et al. (2007), Jayachandran et al. (2005), and Ahearne et al. (2005) support a positive and significant relationship with customer performance within the context of CRM. Thus, the second hypothesis is stated as: H2: Infrastructure performance has a positive association with customer performance within the context of Social CRM. 2.2.2 Process Performance The construct describes aspects that relate to company-wide as well as department- specific processes and activities of Social CRM (i.e., activities using resources that are executed to achieve a business goal to create value). Within CRM literature the construct is also named CRM process capabilities, covering the abovementioned aspects in the corresponding topic. Due to target-oriented customer events, new customer insights, better customer interactions with the company and across customers etc., process performance provides a more efficient customer performance as well as enhances the organizational performance. Particularly, the literature supports a positive and significant association of process performance with customer performance within the CRM context (Chen et al. 2009; Liu, Zhou, and Chen 2006; Padmavathy, Balaji, and Sivakumar 2012; Roh, Ahn, and Han 2005). Thus, the third hypothesis is stated as: H3: Process performance has a positive association with customer performance within the context of Social CRM Concerning the association with organizational performance, the literature also reveals positive and significant relationships. Especially, the results within a CRM context from Chen et al. (2004), Dutu and Hălmăjan (2011), Ernst et al. (2011), Harrigan et al. (2010), and Reinartz et al. (2004), provide strong support for the next hypothesis: H4: Process performance has a positive association with organizational performance within the context of Social CRM. 2.2.3 Customer Performance The construct describes the effects of Social CRM on the customers (customer perception) and the aspects of Social CRM, which are perceived by customers. Additionally, the construct includes direct aspects (i.e., the company has to operate actively) as well as indirect aspects (i.e., management activities of a company, e.g., the peer-to-peer communication), in order to achieve the desired organizational performance. Especially, the results from Chen et al. (2009), Harrigan et al. (2010), Liu et al. (2006), Thongpapanl and Ashraf (2011), Zablah et al. (2012) supports the last hypothesis: H5: Customer performance has a positive association with organizational performance within the context of Social CRM. 2.2.4 Organizational Performance The construct describes the dimension of the company’s success and business results. Particularly, the constructs includes monetization aspects (e.g., financial benefits, customer lifetime value etc.) as well as intangible aspects (e.g., brand awareness, 423 Social CRM Performance Model: An Empirical Evaluation competitive advantage etc.), capturing a holistic approach (Kaplan and Haenlein 2010), in order to establish a long-term and profitable customer relationship. 3 Methodology 3.1 Instrument Development The process of developing instruments (i.e., indicators) is depicted in Figure 2 (cf. Walther et al., 2013). It is conducted in a three stage approach (I. item creation, II. scale development and III. indicator testing), including six sub-stages in total, as proposed by Moore and Benbasat (1991). The first sub-stage “Conceptualization Content Specification” focuses on a literature review, in order to identify context-specific constructs (dimensions), corresponding sub-dimensions and indicators (i.e., performance measures, see Table 2). Second (“Item Generation”), based on the results, indicators are deduced to operationalize the previous constructs. Third, a Q-sorting procedure assesses the “Access Content Validity” with the calculation of an inter-rater reliability index (or related indexes, e.g., Cronbach’s Alpha). Within the next two sub- stages (“Pretest and Refinement” and “Field Test”), the questionnaire is tested, in order to obtain some initial feedback, for instance on problematic areas (definitions, wording, length of the questionnaire etc.). Especially for the unique characteristics of formative indicators and the corresponding constructs, the last sub-stage “Evaluation of Formative Measurement Model and Re-Specification” is based on the process of formative measurements from Cenfetelli and Bassellier (2009). The applied confirmatory factor analysis is designed according to Diamantopoulos and Winklhofer (2001), and focuses on a statistical evaluation of formative indicators and corresponding constructs. Figure 2: Process of developing instruments 3.2 Data Collection A pre-test is distributed online to PhD students and four selected practitioners in the corresponding Social CRM context. To ensure a high degree of validity and increase the quality of the data two screen-out questions are implemented. Participants that answered any of these questions with “no” have been excluded from the online-survey. The final survey is distributed over several Social Media channels (e.g., Xing, LinkedIn, Twitter etc.), focusing on marketing, communication, and IT decision makers. The indicators are measured using a 7-point Likert scale from the agreement-level “strongly disagree” (1) to “strongly agree” (7). In total, a dataset of 126 answers was captured and serves as the basis for the analysis. Some statistics of the data are presented in Table 3. 424 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung Industry % # of Employees % Position in Company % Manufacturing & Utility 30% < 10 15% Executives 30% Others 18% 10 – 49 17% Team Manager 20% Information & Communication 16% 50 – 499 28% Specialized Manager 18% Finance & Insurance 15 % 500 – 999 10% Department Manager 15% Public Administration & Logistics 11% 1000 – 5000 17% Division Manager 14% Health Industry 10% > 5000 13% Others 3% Table 3: Descriptive sample statistic 3.3 Data Analysis The prerequisite step to analyze the structural model is the evaluation of the measurement model, which is calculated using the statistical software SmartPLS and SPSS (e.g., calculating the variance inflation factors). The five hypotheses are tested with SmartPLS. In particular, the coefficients of the corresponding associations are estimated by conducting a structural equation model with a partial least square method (Hair et al. 2013). 4 Results The estimators from the partial least square method are reported, as recommended by Hair et al. (2013), in a two-step approach (Chin 2010). First, the measurement model is calculated. The reflective measurement model is reported as provided by Söllner et al. (2012) and investigate the higher order constructs. The development process of formatively measured indicators and corresponding constructs follows the first four steps recommended by Cenfetelli and Bassellier (2009), applying a confirmatory factor analysis (Diamantopoulos and Winklhofer 2001). Second, the coefficients of the structural model are calculated (Hair et al. 2013) and two quality criteria are presented (i.e., f2, GoF) (Gefen et al., 2011; Wetzels et al., 2009). Both estimations are calculated with a parameter setting using 120 cases and 5000 samples. 4.1 Measurement Model Reflective indicators AVE Com. R. Load. p-value Infrastructure performance 0.896 0.945 IN1_R* In general, sufficient resources are available and cultural 0.944 < 0.01 aspects within the company are established. IN2_R* All in all, resources are available and cultural aspects 0.949 < 0.01 disseminated throughout the company. Process performance 0.916 0.956 PR1_R* In general, the processes and activities in the company 0.957 < 0.01 are improved through Social CRM. PR2_R* All in all, the improvement of business processes and 0.957 < 0.01 activities is substantial. Customer performance 0.923 0.960 CU1_R* Generally, Social CRM activities improve a positive 0.959 < 0.01 customer perception. CU2_R* All in all, customer perceptions are enhanced 0.962 < 0.01 substantially due to Social CRM activities. Organizational performance 0.921 0.959 OR1_R* Generally, Social CRM activities increase business 0.957 < 0.01 results. OR2_R* All in all, the profitability of the Social CRM activities 0.963 < 0.01 enhancing results is high. AVE = Average Variance Extracted; Com. R. = Composite Reliability; Load. = Loadings; *p-value < 0.05 Table 4: Test statistics for the reflective measurement model 425 Social CRM Performance Model: An Empirical Evaluation The reflective measurement model is assessed by estimating (1) convergent validity (i.e., AVE and factor loadings), (2) internal consistency (i.e., composite reliability) and (3) discriminant validity (Hair et al. 2013). Table 4 provides an overview of the test statistics. The indicators show (1) a satisfactory convergent validity as all reflective loadings are clearly above the threshold of 0.5 and significant (Hulland 1999). Additionally, the average variance extracted (AVE) of al reflective constructs is clearly above 0.5 (Fornell and Larcker 1981). (2) Composite reliability also present adequate results of all constructs being above the threshold of 0.7 (Nunnally and Bernstein 1994). The (3) discriminant validity shows a robust result (Hair, Ringle, and Sarstedt 2011), due to the fact that all square roots of each AVE are higher than the corresponding latent variable correlation (Table 5). To conclude, the reflective measurement model is validated for the higher order constructs. (I) (II) (III) (IV) Infrastructure Performance (I) 0.946 Customer Performance (II) 0.430 0.961 Process Performance (III) 0.535 0.758 0.977 Organizational Performance (IV) 0.487 0.680 0.784 0.980 Table 5: Discriminant validity After the fulfillment of the quality criteria for the reflective measurement model, the focus is on evaluating the formative measurement model, concerning the steps: 1. multicollinearity testing, 2. the effect of the number of indicators and non-significant weights, 3. co-occurrence of negative and positive indicators weights, and 4. absolute versus relative indicator contributions (Cenfetelli and Bassellier 2009). Table 6 provides an overview of the test statistics. For the first step (multicollinearity testing), the variance inflation factors (VIFs) are calculated using SPSS. All VIFs are below the maximum threshold of 5.0, recommended by Hair et al. (2011) and Walther et al. (2013). The results reveal that multicollinearity is not an issue in this article. Steps two to four are based on calculated values and test statistics using SmartPLS. The second step (the effect of the number of indicators and non-significant weights) deals with the problem that a large number of indicators cause non-significant weights. The results show that the indicators PR4, PR7 and OR5 are not significant (i.e., illustrated by a high p-value), which has to be considered in the following steps. Cenfetelli and Bassellier (2009) also state that this should not be misinterpreted concerning any irrelevance of the indicators. The only interpretation of this issue is that some indicators have a lower influence than others. In order to gain a deeper understanding, this article continues with step three (co-occurrence of negative and positive indicators weights). No indicator has negative weights; therefore this is not an issue in the article. Step four (absolute versus relative indicator contributions) needs to be conducted by reporting the respective loadings. The loadings indicate that an “indicator could have only a small formative impact on the construct (shown by a low weight), but it still could be an important part of the construct (shown by a high loading)” (Söllner et al. 2012). Concerning the issues with PR4, PR7 and OR5, which show non-significant or low weights, but very high loadings (i.e., higher than 0.7), no further improvements (e.g., dropping indicators) have to be performed (Cenfetelli and Bassellier 2009; Hair, Ringle, and Sarstedt 2011; Hair et al. 2013). 426 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung Formative Indicators P.C. VIF Weights p-value Load Within the context of Social CRM, the company … Infrastructure Performance Cultural Performance 0.443 < 0.01 IN1* integrates Social CRM into the company culture. 1.000 0.303 < 0.01 0.686 IN2* considers cultural aspects. 1.000 0.822 < 0.01 0.963 IT Performance 0.469 < 0.01 IN3* provides an online brand community to interact with 1.000 0.399 < 0.01 0.784 customers e.g., about service or product-related content. IN4* has established a good infrastructure (e.g., IT 1.000 0.731 < 0.01 0.941 resources). Process Performance Company-wide Performance 0.531 < 0.01 PR1* improves organizational processes and activities so 2.059 0.339 < 0.01 0.875 that they are more customer-oriented. PR2* supports sales activities by other users. 2.051 0.43 < 0.01 0.923 PR3* improves ubiquitous communication between the 1.747 0.349 < 0.01 0.878 customers and the company. Department-specific Performance 0.345 < 0.01 PR4 improves the level of knowledge about a customer 2.296 0.138 0.095 0.845 through new customer insights. PR5* enables a more efficient and effective segmentation 2.277 0.376 0.015 0.907 (e.g., market and customer segmentation). PR6* improves the involvement of customers as co- 2.937 0.27 0.012 0.872 creators (e.g., in the innovation process). PR7 enhances the effectiveness of company-initiated 4.609 0.129 0.149 0.887 interactions with customers. PR8* improves the efficient and effective arrangement of 3.122 0.231 0.033 0.836 target-oriented customer events. Customer Performances Indirect Customer Performance 0.480 < 0.01 CU1* enhances and simplifies the exchange of information 1.641 0.281 < 0.01 0.808 between consumers. CU2* enhances the perceived relationship quality of 2.370 0.390 < 0.01 0.910 customers with the company. CU3* increases customer interest in company products, 1.646 0.452 < 0.01 0.925 services and/or company activities. Direct Customer Performance 0.200 0.077 CU4* improves personalized and customer-oriented 1.000 0.326 < 0.01 0.787 products and services. CU5* improves customer access to a variety of support 1.000 0.770 < 0.01 0.965 options for interacting with the company. Organizational Performance Monetization Performance 0.354 < 0.01 OR1* increases the success of newly introduced or 1.867 0.302 < 0.01 0.843 developed products and services. OR2* increases customer value over the relationship 2.354 0.314 < 0.01 0.897 lifespan. OR3* increases the company’s profit and/or decreases 1.757 0.496 < 0.01 0.933 costs. Intangible Performance 0.392 < 0.01 OR4* increases the efficiency and effectiveness of 1.999 0.584 < 0.01 0.914 business activities (e.g., increases the efficiency of supply chain management). OR5 increases brand awareness and brand recognition 1.627 0.036 0.270 0.733 (e.g., by means of customer recommendations). OR6* secures a competitive advantage. 1.537 0.497 < 0.01 0.885 P.C. = Path Coefficient between 1st- and 2nd-order construct;.VIF = Variance Inflation Factor; Load. = Loadings; * p-value < 0.05 Table 6: Test statistics for the formative measurement model 427 Social CRM Performance Model: An Empirical Evaluation To investigate all relationships of the measurement model, the interrelationship between the first- and second-order constructs have to be considered. Due to the fact of having eight first-order constructs (cultural-, IT performance etc.), resulting in four second- order constructs (infrastructure performance etc.), the path coefficients have to be investigated. Seven out of eight interrelationships reveal highly significant path coefficients (i.e., p-value < 0.01). Based on the high, but still significant, p-value of “Direct Customer Performance” (i.e., p-value < 0.10), no further improvements have to be performed. To conclude, the measurement model is well-suited and validated within the Social CRM context. 4.2 Structural Model Having established the appropriateness of the measures, the structural model is tested with the outlined parameter setting. Three path coefficients (H1, H3, H4) show significant structural relationships (p-value lower than 0.05). In contrast, the derived hypotheses (H2, H5) reveal non-significant structural relationships (Figure 3). Figure 3: Result of the evaluated Social CRM performance model In addition, two quality criteria are presented (i.e., f2, GoF) (Gefen et al., 2011; Wetzels et al., 2009). The f2 criteria highlight possible omission of structural relationships. All calculated values are below the threshold of 0.02 (Wetzels et al. 2009). Therefore, it can be stated that no important structural relationships are omitted. The Goodness of Fit (GoF) criteria is “defined as the geometric mean of the average communality and average R2 (for endogenous constructs)” (Wetzels et al. 2009). The calculated value of 0.849 is above the threshold of 0.36 and indicate a well global performance of the structural model (Tenenhaus et al. 2005). 5 Discussion and Implications The article makes several important contributions by presenting an empirically evaluated performance model for Social CRM. The four adopted constructs (infrastructure performance, process performance, customer performance and organizational performance) are well-suited for the Social CRM context. As outlined in the hypotheses development section, the first hypothesis can be supported, starting that 428 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung IT and cultural aspects enable a company to implement effective and efficient Social CRM processes (Chuang and Lin 2013; Ernst et al. 2011; Lee et al. 2010; Wang and Feng 2012). According to Chen et al. (2009) and Liu et al. (2006), hypothesis three can be supported. The knowledge of, e.g., customer insights enables a better customer interaction, provides offerings of individual products and services etc. Additionally, the support of hypothesis four is not really surprising (Chen et al. 2004; Dutu and Hălmăjan 2011; Ernst et al. 2011; Harrigan et al. 2010; Reinartz, Krafft, and Hoyer 2004). In particular, process performance has a highly significant association with organizational performance. To conclude, it can be stated that the internal performance aspects (i.e., infrastructure, process and organizational) are well-suited for the Social CRM context. However, the two additional results show no support (hypotheses two and five). Compared to the previous statement, customer performance neither has an association with organizational performance nor serves as a mediator for infrastructure performance. One possibility is the maturity level of already implemented Social CRM activities. Companies are on an early stage of this process. As interviews with practitioners show, companies are starting to implement Social CRM in a testable and internal setting, i.e., by creating a Social CRM campaign. Therefore, the internal performance aspects are significant influence factors. Companies are still neglecting the effect of a good communicated added value for their customers, which lead to the non- significant influence factor as well as mediator for the organizational performance. The study has various implications for the scientific community. Firstly, the resulting measurement model facilitates the use of new indicators and corresponding constructs for measuring Social CRM performance. Secondly, the rigorous nature of the study enables researchers to adopt and apply the measurement model for their own research. Finally, the holistic approach, including different dimensions of performance, generates deeper insights into Social CRM performance within a company and guides future research activities. Three practical implications in particular can be stated. First, the model facilitates a control system for current Social CRM activities, e.g., an appraisal of social campaigns, considering various aspects of effective or ineffective campaigns. Second, it enables the justification of current and future Social CRM initiatives and engagements in a company, e.g., spending money on new investments in Social CRM processes, like increasing the total number of customer touch-points, which have a strong influence on the organizational performance. Finally, companies can detect clearly defined strength and weaknesses of their Social CRM activities. To conclude, the Social CRM performance model generates deeper insights into company’s performance interrelationships and provides a control system, in order to assess Social CRM activities and enhance organizational performance. 6 Limitations and further Research Three potential limitations constrain the results of this research. Firstly, despite the highly significant values of the measurement model (i.e., the statistical test values), there may be missing formative indicators, which should be included in the model. Secondly, due to the fact that the study is the first evaluated performance model for Social CRM, conducting a transferability test is not possible (Cenfetelli and Bassellier 2009). Finally, the study does not control the maturity level of the companies, which could influence the results. 429 Social CRM Performance Model: An Empirical Evaluation One promising approach for further research is an extension of the Social CRM performance model based on the resource-based view. An investigation of resources (e.g., Social CRM technology use) and an empirical investigation of capabilities (e.g., processes) can be tested statistically. For example, the impact of Social CRM capabilities on performance (Rapp et al., 2010), or the impact of Social CRM technology use on performance (Zablah et al., 2012). To conclude, the rigorous and systematically derived results presented in the article form a sound basis for further research projects. References Ahearne, Michael, Douglas E. Hughes, and Niels Schillewaert (2007), “Why sales reps should welcome information technology: Measuring the impact of CRM-based IT on sales effectiveness,” International Journal of Research in Marketing, 24 (4), 336–349. ———, R. Jelinek, and Adam Rapp (2005), “Moving beyond the direct effort of SFA adoption on salesperson performance: Training and support as key moderating factors,” Industrial Marketing Management, 34 (4), 379–388. Akroush, Mamoun N., Samer E. Dahiyat, Hesham S. Gharaibeh, and Bayan N. Abu- Lail (2011), “Customer relationship management implementation: An investigation of a scale’s generalizability and its relationship with business performance in a developing country context,” International Journal of Commerce and Management, 21 (2), 158–191. Alvarez, Gene (2013), “Hype Cycle for E-Commerce 2013,” Gartner, Inc. Askool, Sanaa and Keiichi Nakata (2011), “A conceptual model for acceptance of social CRM systems based on a scoping study,” AI & SOCIETY, 26 (3), 205–220. Becker, Jan U., Goetz Greve, and Sönke Albers (2009), “The impact of technological and organizational implementation of CRM on customer acquisition, maintenance, and retention,” International Journal of Research in Marketing, 26 (3), 207–215. Vom Brocke, J., A. Simons, B. Niehaves, K. Riemer, R. Plattfaut, and A. Cleven (2009), “Reconstructing the Giant: On the Importance of Rigour in Documenting the Literature Search Process,” in Proceedings of the 17th European Conference on Information Systems, Verona, Italy, 3226–3238. Cenfetelli, Ronald T. and Geneviève Bassellier (2009), “Interpretation of Formative Measurement in Information Systems Research,” MIS Quarterly, 33 (4), 689–707. Chen, Jashen, Russell K. H. Ching, Eldon Y. Li, and Yiling Liao (2004), “An Exploratory Study of the Effects of CRM Practices on CRM Effectiveness and Business Performance,” in The Fourth International Conference on Electronic Business, 249–254. Chen, Ja-Shen, H. J. Rebecca Yen, Eldon Y. Li, and Russell K. H. Ching (2009), “Measuring CRM effectiveness: Construct development, validation and application of a process-oriented model,” Total Quality Management & Business Excellence, 20 (3), 283–299. 430 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung Chin, Wynne W. (2010), “How to Write Up and Report PLS Analyses,” in Handbook of Partial Least Squares, V. Esposito Vinzi, W. W. Chin, J. Henseler, and H. Wang, eds., Berlin, Heidelberg: Springer-Verlag, 655–690. Chuang, Shu-Hui and Hong-Nan Lin (2013), “The roles of infrastructure capability and customer orientation in enhancing customer-information quality in CRM systems: Empirical evidence from Taiwan,” International Journal of Information Management, 33 (2), 271–281. Coltman, Tim (2007), “Can Superior CRM Capabilities Improve Performance in Banking Can Superior CRM Capabilities Improve Performance in Banking,” Journal of Financial Services Marketing, 12 (2), 102–114. ———, Timothy M. Devinney, and David F. Midgley (2011), “Customer relationship management and firm performance,” Journal of Information Technology, 26 (3), 205–219. Diamantopoulos, Adamantios and Heidi M. Winklhofer (2001), “Index Construction with Formative Indicators: An Alternative to Scale Development,” Journal of Marketing Research, 38 (2), 269–277. Dutu, Cristian and Horańiu Hălmăjan (2011), “The Effect of Organizational Readiness on CRM and Business Performance,” International Journal of Computers, 5 (1), 106–114. Ernst, Holger, Wayne D. Hoyer, Manfred Krafft, and Katrin Krieger (2011), “Customer relationship management and company performance—the mediating role of new product performance,” Journal of the Academy of Marketing Science, 39 (2), 290– 306. Fornell, Claes and David F. Larcker (1981), “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error,” Journal of Marketing Research, 18 (1), 39–50. Gefen, David, Edward E. Rigdon, and Detmar Straub (2011), “An Update and Extension to SEM Guidelines for Administrative and Social Science Research,” MIS Quarterly, 35 (2), 3–14. Grabner-Kraeuter, Sonja, Gernot Moedritscher, Martin Waiguny, and Werner Mussnig (2007), “Performance Monitoring of CRM Initiatives,” in Proceedings of the 40th Annual Hawaii International Conference on System Sciences, Waikoloa, Hawaii, 150–160. Greenberg, Paul (2010), “The impact of CRM 2.0 on customer insight,” Journal of Business & Industrial Marketing, 25 (6), 410–419. Hair, Joseph F., G. Tomas M. Hult, Christian M. Ringle, and Marko Sarstedt (2013), A Primer on Partial Least Squares Strucutral Equation Modeling (PLS-SEM), Thousand Oaks: SAGE Publications, Inc. ———, Christian M. Ringle, and Marko Sarstedt (2011), “PLS-SEM: Indeed a Silver Bullet,” Journal of Marketing Theory and Practice, 19 (2), 139–152. Harrigan, Paul, Andreas Schroeder, Israr Qureshi, Yulin Fang, Patrick Ibbotson, Elaine Ramsey, and Darren Meister (2010), “Internet Technologies, ECRM Capabilities, 431 Social CRM Performance Model: An Empirical Evaluation and Performance Benefits for SMEs: An Exploratory Study,” International Journal of Electronic Commerce, 15 (2), 7–45. Hulland, J. (1999), “Use of partial least squares (PLS) in strategic management research: a review of four recent studies,” Strategic Management Journal, 20 (2), 195–204. Jain, Rajnish, Sangeeta Jain, and Upinder Dhar (2003), “Measuring Customer Relationship Management,” Journal of Services Research, 2 (2), 97–109. Jayachandran, Satish, Subhash Sharma, Peter Kaufman, and Pushkala Raman (2005), “The Role of Relational Information Processes and Technology Use in Customer Relationship Management,” Journal of Marketing, 69 (4), 177–192. Kalyar, M. N., N. Rafi, and M. Azeem (2013), “Factors Affecting Company Performance and New Produkt Performance,” Journal of Sustainable Development Studies, 2 (1), 127–151. Kaplan, Andreas M. and Michael Haenlein (2010), “Users of the world, unite! The challenges and opportunities of Social Media,” Business Horizons, 53 (1), 59–68. Keramati, Abbas, Hamed Mehrabi, and Navid Mojir (2010), “A process-oriented perspective on customer relationship management and organizational performance: An empirical investigation,” Industrial Marketing Management, 39 (7), 1170– 1185. Kim, Byeong Yong (2008), “Mediated Effects of Customer Orientation on Customer Relationship Management Performance,” International Journal of Hospitality & Tourism Administration, 9 (2), 192–218. Kim, Hyung-Su and Young-Gul Kim (2009), “A CRM performance measurement framework: Its development process and application,” Industrial Marketing Management, 38 (4), 477–489. Kim, Jonghyeok, Euiho Suh, and Hyunseok Hwang (2003), “A Model for Evaluating the Effectiveness of CRM Using the Balanced Scorecard,” Journal of Interactive Marketing, 17 (2), 5–19. Kimiloglu, Hande and Hülya Zarali (2009), “What signifies success in e-CRM?,” Marketing Intelligence & Planning, 27 (2), 246–267. Küpper, Torben, Reinhard Jung, Tobias Lehmkuhl, Sebastian Walther, and Alexander Wieneke (2014), “Performance Measures for Social CRM: A Literature Review,” in Proceedings of the 27th Bled eConference, Bled, Slovenia, 125–139. ———, Alexander Wieneke, Tobias Lehmkuhl, Reinhard Jung, Sebastian Walther, and Torsten Eymann (2015), “Measuring Social CRM Performance: A Preliminary Measurement Model,” in Proceedings of the 12th International Conference on Wirtschaftsinformatik, Osnabück, Deutschland, 887–901. Lee, Chao-Hsiung, Shaio Yan Huang, F. Barry Barnes, and Li Kao (2010), “Business performance and customer relationship management: The effect of IT, organisational contingency and business process on Taiwanese manufacturers,” Total Quality Management & Business Excellence, 21 (1), 43–65. 432 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung Lehmkuhl, Tobias and Reinhard Jung (2013), “Towards Social CRM - Scoping the Concept and Guiding Research,” in Proceedings of the 26th Bled eConference, Bled, Slovenia, 190–205. Lin, Chad, Koong Lin, Yu-An Huang, and Wen-Liang Kuo (2006), “Evaluation of Electronic Customer Relationship Management: The Critical Success Factors,” The Business Review, Cambridge. Liu, Yan, Chang-Feng Zhou, and Ying-Wu Chen (2006), “Determinants of E-CRM in Influencing Customer Satisfaction,” in PRICAI 2006: Trends in Artificial Intelligence, Q. Yang and G. Webb, eds., Berlin, Heidelberg: Springer-Verlag, 767–776. Llamas-Alonso, María Rosa, Ana Isabel Jiménez-Zarco, María Pilar Martínez-Ruiz, and John Dawson (2009), “Designing a Predictive Performance Measurement and Control System to Maximize Customer Relationship Management Success,” Journal of Marketing Channels, 16 (1), 1–41. Moore, Gary C. and Izak Benbasat (1991), “Development of an Instrument to Measure the Perceptions of Adopting an Information Technology Innovation,” Information Systems Research, 2 (3), 192–222. Mumuni, Alhassan G. and Kelley O’Reilly (2014), “Examining the Impact of Customer Relationship Management on Deconstructed Measures of Firm Performance,” Journal of Relationship Marketing, 13 (2), 89–107. Neely, Andy, Mike Gregory, and Ken Platts (1995), “Performance measurement system design - A literature review and research agenda,” International Journal of Operations & Production Management, 15 (4), 80–116. Nunnally, J. C. and I. H. Bernstein (1994), Psychometric Theory, New York: McGraw- Hill. Öztayşi, Başar, Tolga Kaya, and Cengiz Kahraman (2011), “Performance comparison based on customer relationship management using analytic network process,” Expert Systems with Applications, 38 (8), 9788–9798. ———, Selime Sezgin, and Ahmet Fahri Özok (2011), “A measurement tool for customer relationship management processes,” Industrial Management & Data Systems, 111 (6), 943–960. Padmavathy, C., M. S. Balaji, and V. J. Sivakumar (2012), “Measuring effectiveness of customer relationship management in Indian retail banks,” International Journal of Bank Marketing, 30 (4), 246–266. Peltier, James W., Debra Zahay, and Donald R. Lehmann (2013), “Organizational Learning and CRM Success: A Model for Linking Organizational Practices, Customer Data Quality, and Performance,” Journal of Interactive Marketing, 27 (1), 1–13. Rapp, Adam, Kevin J. Trainor, and Raj Agnihotri (2010), “Performance implications of customer-linking capabilities: Examining the complementary role of customer orientation and CRM technology,” Journal of Business Research, 63 (11), 1229– 1236. 433 Social CRM Performance Model: An Empirical Evaluation Reinartz, Werner, Manfred Krafft, and Wayne D. Hoyer (2004), “The Customer Relationship Management Process: Its Measurement and Impact on Performance,” Journal of Marketing Research, 41 (3), 293–305. Roh, Tae H., Cheol K. Ahn, and Ingoo Han (2005), “The priority factor model for customer relationship management system success,” Expert Systems with Applications, 28 (4), 641–654. Sedera, Darshana and Wenjuan Wang (2009), “Towards a CRM and SCM Benefits Measurement Model,” in Proceedings of the International Conference on Information Systems, Phoenix, USA, 1–12. Shafia, Mohammad Ali, Mohammad Mahdavi Mazdeh, Mahboobeh Vahedi, and Mehrdokht Pournader (2011), “Applying fuzzy balanced scorecard for evaluating the CRM performance,” Industrial Management & Data Systems, 111 (7), 1105– 1135. Söllner, Matthias, Axel Hoffmann, Holger Hoffmann, Arno Wacker, and Jan Marco Leimeister (2012), “Understanding the Formation of Trust in IT Artefacts,” in Proceedings of the 33rd International Conference on Information Systems, Orlando, USA, 1–18. Tenenhaus, M., V. E. Vinzi, Y.-M. Chatelin, and C. Lauro (2005), “PLS path modeling,” Computational Statistics and Data Analysis, 48 (1), 159–205. Thongpapanl, Narongsak and Abdul Rehamn Ashraf (2011), “Enhancing online performance through Website content and personalization,” Journal of Computer Information Systems, 52 (1), 3–14. Trainor, Kevin J. (2012), “Relating Social Media Technologies to Performance: A Capabilities-Based Perspective,” Journal of Personal Selling & Sales Management, 32 (3), 317–331. ———, James Andzulis, Adam Rapp, and Raj Agnihotri (2014), “Social media technology usage and customer relationship performance: A capabilities-based examination of social CRM,” Journal of Business Research, 67 (6), 1201–1208. Walther, Sebastian, Darshana Sedera, Saonee Sarker, and Torsten Eymann (2013), “Evaluating Operational Cloud Enterprise Systems Success: An Organizational Perspective,” in Proceedings of the 21st European Conference on Information Systems, Utrecht, Netherlands, 1–12. Wang, Wenjuan, Darshana Sedera, and Felix Tan (2009), “Measuring CRM and SCM Benefits: A Preliminary Measurement Model,” in Proceedings of the 13th Pacific Asia Conference on Information Systems, Hyderabad, India, 1–12. Wang, Yonggui and Hui Feng (2012), “Customer relationship management capabilities - Measurement, antecedents and consequences,” Management Decision, 50 (1), 115–129. Wetzels, Martin, Gaby Odekerken-Schröder, and Claudia van Oppen (2009), “Using PLS Path Modeling for Assessing Hierarichical Construct Models: Guidelines and Empirical Illustration,” MIS Quarterly, 33 (1), 177–195. 434 Torben Küpper, Tobias Lehmkuhl, Nicolas Wittkuhn, Alexander Wieneke, Reinhard Jung Wu, Fang, Vijay Mahajan, and Sridhar Balasubramanian (2003), “An Analysis of E- Business Adoption and its Impact on Business Performance,” Journal of the Academy of Marketing Science, 31 (4), 425–447. Zablah, Alex R., Danny N. Bellenger, Detmar W. Straub, and Wesley J. Johnston (2012), “Performance Implications of CRM Technology Use: A Multilevel Field Study of Business Customers and Their Providers in the Telecommunications Industry,” Information Systems Research, 23 (2), 418–435. Zinnbauer, Markus A and Markus Eberl (2005), “Measuring Customer Relationship Management Performance: A Consumer-Centric Approach,” Journal of Marketing Channels, 12 (3), 79–104. Appendix 1. Becker et al. 7. Wang and 13. Coltman (2007) 19. Chuang and Lin 25. Coltman et al. (2009) Feng (2012) (2013) (2011) 2. Chen et al. 8. Zablah et al. 14. Kalyar et al. 20. Thongpapanl and 26. Trainor et al. (2009) (2012) (2013) Ashraf (2011) (2014) 3. Kim (2008) 9. Ernst et al. 15. Harrigan et al. 21. Wu et al. (2003) 27. Jayachandran et (2011) (2010) al. (2005) 4. Padmavathy et 10. Chen et al. 16. Dutu and 22. Akroush et al. 28. Ahearne et al. al. (2012) (2004) Hălmăjan (2011) (2011) (2007) 5. Reinartz et al. 11. Rapp et al. 17. Peltier et al. 23. Lee et al. (2010) 29. Ahearne et al. (2004) (2010) (2013) (2005) 6. Roh et al. 12. Liu et al. 18. Keramati et al. 24. Mumuni and (2005) (2006) (2010) O’Reilly (2014) 435 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia The Effects of Brand Engagement in Social Media on Share of Wallet Heikki Karjaluoto University of Jyväskylä, Finland heikki.karjaluoto@jyu.fi Juha Munnukka University of Jyväskylä, Finland juha.munnukka@jyu.fi Severi Tiensuu Dagmar Oy, Finland severi.tiensuu@dagmar.fi Abstract Customer engagement and share-of-wallet (SOW) are relatively new in the marketing literature, and academic research has only limitedly examined these concepts. This study presents five motivational drivers of customer brand engagement in social media and examines the nature of the relationship between these drivers and engagement. The moderation effect of consumer innovativeness on the relationship between engagement and SOW is also examined. Results suggest that community exerts the strongest positive effect on customer brand engagement and that such engagement positively influences SOW. The findings also indicate that consumer innovativeness strengthens the relationship between engagement and SOW. The findings also show that frequency of visits on the brand community site predict higher SOW. This study contributes to the understanding of customer brand engagement by describing how online brand community engagement and its antecedents drive SOW. Keywords: Customer Brand Engagement, Share of Wallet, Brand, Social Media 1 Introduction Companies have fast incorporated social media into their marketing and brand building activities (Kaplan & Haenlein, 2010). For example, in recent years, several companies have created brand communities on social media, such as Facebook, which currently has more than 1.2 billion active users on a monthly basis (Facebook Annual Report, 2013). Therefore, rise of social media and technological development have provided 436 Heikki Karjaluoto, Juha Munnukka and Severi Tiensuu companies with new tools that have led to new practices of contacting and engaging with customers. This has made customer engagement increasingly important strategy for companies’ customer relationship management (Libai, 2011; Sashi, 2012; Kumar et al., 2010). This development has also been reflected to academic research, which fast develops theories and accumulates empirical evidence of customer brand engagement (e.g., Bowden, 2009; Sashi, 2012; Libai, 2011). For example Bijmolt et al. (2010) indicate that consumer brand engagement has been one of the emerging measures for maximizing business value. Consumers’ share of spending on the company’s offerings (i.e. share of wallet) has been suggested as a focal measure of business value and behavioural loyalty in consumer marketing context (e.g. Keiningham et al., 2005; Zeithaml, 2000). Especially for retailers who continuously search for new and more effective practices of extracting a higher share of total grocery expenditures from their customers share of wallet (SOW) is of high importance (Meyer-Waarden, 2006). In recent years social media has been recognized as a highly potential channel for effectively contacting and engaging with consumers. However, brand engagement is still relatively new concept to the marketing literature and its drivers as well as consequences on consumer buying behaviour limited. More empirical research is needed especially in the context of online communities (e.g., Cheung et al., 2011; Jahn & Kunz, 2012). Therefore, our study aims at shedding light on the drivers of customer engagement in online brand communities and its impact on customers spending on the companies’ products. Prior literature proposes engagement to arise from motivational drivers (Brodie et al., 2011; van Doorn et al., 2010; Hollebeek, 2011; Ouwersloot & Odekerken‐ Schröder, 2008). McQuail’s (1983) classifies motivations into four main components: social interaction, need for information, entertainment, and developing personal identity. Thereafter, economic benefits have also been presented as a driver of engagement (Gwinner et al., 1998; Muntinga et al., 2011). Previous studies also show several consequences of customer engagement, such as higher brand satisfaction, trust, commitment, emotional connection/attachment, empowerment, consumer value, and loyalty (e.g., Bowden, 2009; Brodie et al., 2011; van Doorn et al., 2010). However, research lacks empirical evidence of how engagement affects consumers’ spending between different brands. A good example of engagement on social media is Coca-Cola, which has successfully capitalized on social media in brand management. They actively participate in the social media brand community to inspire optimism and happiness and to build the Coca-Cola brand (The Coca-Cola Company, 2014). Coca-Cola has nearly 80 million fans and more than 640 000 people talking about the company and its products on Facebook. They aim at building personal relationships with millions of people accruing their brand as well as business value. Consumer innovativeness has been recognized as a focal construct of consumer behaviour especially in the new product adoption context (Hirschman, 1980; Midgley & Dowling, 1978). Cotte and Wood (2004) define consumer innovativeness as a tendency to willingly embrace change, try new things, and buy new products more often and more rapidly than others. In the current work, consumer innovativeness is understood as a consumer’s personality trait that influences the effect strength of the consumer’s 437 The Effects of Brand Engagement in Social Media on Share of Wallet engagement in an online brand community on share of spending on the brand’s products. Previous research is limited in showing evidence how consumer innovativeness affects the effectiveness of specific marketing strategies for influencing consumer buying behaviour. Based on this discussion this study strives to contribute to the identified limitations in the current knowledge by constructing and testing a conceptual model of customer brand engagement in social media context. This study examines behavioural and experiential motives that affect customer brand engagement in a social media context and the effect of engagement on SOW. We combine engagement and SOW theories to develop a framework for the associations between the aforementioned concepts. This research aligns with the suggestions of Brodie et al. (2011), Gummerus et al. (2012), and Jahn and Kunz (2012) calling for more empirical studies on customer engagement to identify different types of brand communities and similarities in engagement behaviours. Also the Marketing Science Institute (MSI) addressed customer engagement as a key research priority. This research contributes to our knowledge by first showing the key drivers of customer engagement in online brand communities, and second, how brand community engagement affects the brand’s share of the consumers’ wallet (SOW). Third, we examine the effect of consumers’ innovativeness on the proposed model. Rest of the paper is structured as follows. First, we briefly describe the research framework and develop hypotheses on how motivational factors, brand engagement, share of wallet and perceived innovativeness are connected to each other. Then we describe the methods and measures applied to test the research model. Finally, we present the analyses results and discuss the findings from both theoretical and managerial aspects. 2 Sources of Brand Engagement and Influence on Share-of-Wallet 2.1 Research Model and Hypotheses The conceptual model of this study and six hypotheses that are derived from a prior literature are presented in Figure 1. It examines the effect of five types of motives on customer engagement in social media context and how brand engagement and perceived customer innovativeness affect SOW. The model is controlled for gender, age and frequency of visits to the social media forums (Facebook and Twitter) of the brand. 438 Heikki Karjaluoto, Juha Munnukka and Severi Tiensuu H1 (+) Community Controls: Gender, Age, Frequency of visits H2 (+) Information H3 (+) H6 (+) Customer Brand Share of Enjoyment Engagement Wallet H4 (+) H7 (+) Identity H5 (+) Perceived Economic Innovativeness Figure 1: Research model and hypotheses (dashed lines represent moderating effects) Brodie et al. (2011) suggest customer engagement as a strategic imperative for establishing and sustaining a competitive advantage and as a valuable predictor of future business performance. It is claimed to improve profitability (Voyles, 2007) as well as promoting customers’ WOM behaviour, such as increasing customers’ tendency to provide referrals and recommendations on specific products, services, and/or brands to others (Brodie et al., 2011). In online context, virtual brand communities constitute an important platform for customer engagement behaviour (Brodie et al., 2011; Dholakia et al., 2004; Kane et al., 2009; McAlexander et al., 2002). Therefore, customer brand engagement in social media is defined here as “an interactive and integrative participation in the fan-page community” (Jahn & Kunz, 2012, p.349). Engagement stems from several motivational drivers (Brodie et al., 2011; Calder & Malthouse, 2008; Hollebeek, 2011; van Doorn et al., 2010). Five main components are addressed here, which are relevant in social media context: community motivations (c.f. social interaction motivations), information motivations, entertainment motivations, personal identity motivations and economic motivations (Heinonen, 2011; McQuail, 1983; Mersey et al., 2012). Jahn and Kunz (2012) reveal that community value is among the strongest drivers of brand fan page use. Need for information is another key motive for participating in online brand communities (Brodie et al., 2013; De Valck et al., 2009). In addition, entertainment is an important motivation for consuming user-generated content (Muntinga et al., 2011). It provides experiential value for customers from using online services such as social media (Gummerus et al., 2012; Men & Tsai, 2013). Similarly, impression management and identity expression have been identified as motivators of social network sites access (Boyd, 2008) where users can express themselves by adjusting their profiles, linking to particular friends, displaying their “likes” and “dislikes,” and joining groups (Tufekci, 2008). Finally, 439 The Effects of Brand Engagement in Social Media on Share of Wallet economic benefits provide impetus for joining brand communities. For example economic incentives such as discounts and time savings or opportunity to participate in raffles and competitions are important motivational drivers for consumers to engage in online brand communities (Gwinner et al., 1998). Against this backdrop four hypotheses are constructed that these five motivations drive consumers’ brand engagement in social media: H1: Community experience is positively associated with customer brand engagement. H2: Information experience is positively associated with customer brand engagement. H3: Enjoyment experience is positively associated with customer brand engagement. H4: Identify-related experience is positively associated with customer brand engagement. H5: Economic-related experience is positively associated with customer brand engagement. Share of wallet is understood as the percentage of the volume of total business transactions between a firm and a customer within a year (Keiningham et al., 2003). For example in retail banking, it is “the stated percentage of total assets held at the bank being rated by the customer” (Keiningham et al., 2007, p. 365). According to Perkins-Munn et al. (2005), a firm’s efforts to manage customers’ spending patterns tend to represent greater opportunities than does simply trying to maximize customer retention rates. In fact, rather than concentrating on customer retention rates a more effective way to increase a company’s profitability is to concentrate on serving existing customers (Reinartz & Kumar, 2000) and increasing the company’s share of wallet in their expenditures (see Zeithaml, 2000). For example Vivek et al. (2012) show that engaging consumers with the company leads to positive outcomes, such as increased SOW. Consumers’ share of spending is an important measure of behavioural loyalty (e.g. De Wulf et al., 2001; Keiningham et al., 2005), which provides essential information to retailers on how and on what grounds customers allocate their purchases across different brands and stores (Meyer-Waarden, 2006). This enables retailers to formulate strategies to motivate their customers to allot a higher share of their expenditure to the retailer's products. Therefore, SOW has been suggested as a more reliable measure of loyalty than other loyalty measures (Jones & Sasser, 1995; Zeithaml, 2000). Although engagement has been linked with satisfaction, commitment and loyalty (e.g., Bowden, 2009; Brodie et al., 2011; van Doorn et al., 2010), only Vivek et al. (2012) has specifically investigated the associations between consumer brand engagement and SOW. As this preliminary evidence indicates a positive association between these constructs and as a strong support exist for the positive relationship between customer engagement and loyalty (Algesheimer et al., 2005; Hollebeek, 2011; Matzler et al., 2008), we expect that a consumer’s higher engagement in an online brand community leads to higher brand’s share of the consumer’s wallet. Therefore, next hypothesis postulates following: H6: Customer brand engagement has a positive effect on SOW. Perceived innovativeness refers to a tendency to embrace change, try new things, and buy new products more often and more rapidly than others (Cotte & Wood, 2004). It is strongly related to the adoption and purchase of products, especially new products. 440 Heikki Karjaluoto, Juha Munnukka and Severi Tiensuu Steenkamp et al. (1999) state innovative consumers change consumption patterns and previous product choices rather than remain with old ones. Joseph and Vyas (1984) further suggest that individuals’ innovativeness affects the use of new information and ability to recognize ideas from others. Innovative individuals are also found as more responsive than less innovative individuals to communication and information (e.g. in brand communities) that has relevance to them. In addition, prior literature suggests that innovativeness is context or product specific (Citrin et al., 2000; Goldsmith & Hofacker, 1991). Thus, an individual may not be innovative in general terms but might still be innovative in a specific context, such as in the case of household appliances or the use of new communication channels. The present study is conducted in the household appliances and online brand community context. In the household appliances context new technological innovations form the basis of brands’ competitive power and consumers’ buying decisions are strongly affected by brands’ technological capabilities. Customers’ brand engagement in social media drives brand loyalty and is suggested to result in improvements in the company’s competitive position (Brodie et al., 2011) as well as profitability (Voyles, 2007). The prior evidence proposes that a customer’s innovativeness affects his/her communication behaviour and enhances the ability to evaluate and apply new information for example in buying decisions. Therefore, customer’s innovativeness is expected to moderate the relationship between customer brand engagement and SOW. The more innovative individual is, the more strongly he/she engages to the brand’s online community and the more strongly community engagement is reflected in the brand’s share of the consumer’s wallet. Thus, the final hypothesis states following, H7: Perceived innovativeness moderates the relationship between customer brand engagement and SOW. 3 Methodology We tested the hypotheses with data obtained from Facebook fans and Twitter followers of a global consumer electronics company. Within a two-week response time, 818 completed questionnaires were returned. The effective response rate was 57%. We used established scales anchored from 1 “strongly disagree” to 5 “strongly agree” to measure the study constructs. Community (four items) and enjoyment (three items) scales were adapted from Calder et al. (2009), Mersey et al. (2012) and Calder and Malthouse (2008). Identity was measured with three items and information with three items adapted from Mersey et al. (2012). Two items were used to measure economic benefits taken from Hennig-Thurau et al. (2004). Customer brand engagement (seven items) scale was adapted from Jahn and Kunz (2012), Gummerus et al. (2012) and Muntinga et al. (2011). SOW was measured with two items from De Wulf et al. (2001). Finally, in measuring customer perceived innovativeness, three items adapted from Lu et al. (2005) were used. The data was first subjected to exploratory factor analysis and thereafter the hypotheses were tested with partial least squares structural equation modelling software SmartPLS 3.0 (Ringle, Wende, & Becker, 2014). All the study constructs are reflective. 441 The Effects of Brand Engagement in Social Media on Share of Wallet Common method bias was minimized already in the data collection stage by mixing the items in the questionnaire and keeping the respondents’ identities confidential. In the analysis phase, we ran a PLS model with a method factor. The results suggest that average variance explained by the indicators (0.704) was considerably higher than the average method-based variance (0.016). Given the magnitude of method variance, common method bias is unlikely to be of serious concern in this study. 4 Results Most of the respondents were male 547 (67%). The major age group falls between 26 and 35 years (25%). The next largest groups are those aged 36–45 (19.9%) and 18– 25 (18.9%). Most of the respondents visit the fan page 1–3 times per week (30%) or 2–3 times per month (24%). This composition aligns with the profile of the visitors to the case company’s Facebook fan page, where the female population accounts for approximately 40% of the community’s population. The confirmatory factor analysis was acceptable as the factor loadings were high (>0.75) and significant, composite reliabilities for the scales were larger than 0.840, AVE values exceeded the cut-off criteria 0.50, and discriminant validity is achieved as the square root of Ave exceeded the value of correlation between the factors (see Table 1). AVE (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) COMb (1) .645 .803 INFc (2) .719 .618 .848 ENJd (3) .699 .616 .690 .836 IDEe (4) .683 .654 .580 .683 .826 ECOf (5) .727 .388 .326 .372 .444 .852 CBE g (6) .687 .765 .583 .631 .686 .539 .829 PIh (7) .777 .150 .206 .309 .174 .158 .221 .881 SOWi (8) .868 .248 .246 .297 .283 .190 .358 .283 .931 FVj (9) n/a .352 .327 .381 .341 .241 .451 .284 .361 n/a Gender (10) n/a .114 .143 .028 .038 .080 .039 -.276 -.054 -.102 n/a Age (11) n/a -.017 .011 -.071 .011 -.042 -.016 -.239 -.073 -.065 .142 n/a Mean - 2.99 3.44 3.33 2.67 3.29 2.75 4.08 4.17k 3.28 n/a n/a s.d. - 1.10 1.00 0.96 1.05 1.17 1.14 0.99 2.55 1.24 n/a n/a CRa .879 .884 .874 .866 .840 .939 .912 .929 n/a n/a n/a Table 1: Discriminant validity Notes: a CR = Composite reliability; b COM –Community; c INF – Information; d ENJ – Enjoyment; e IDE – Identity; f ECO –Economic; g CBE – Customer brand engagement; h PI – Perceived innovativeness; i SOW – Share of wallet; j FV – Frequency of visits; k SOW item scale transformed from 0-100 to 0-10. 442 Heikki Karjaluoto, Juha Munnukka and Severi Tiensuu n/a = Not applicable. Construct measured through a single indicator; composite reliability and AVE cannot be computed The model’s predictive relevance was medium-high as the model explains more than 50% of the R2 of customer brand engagement (R2 = 0.695). The R2 for SOW was 0.206. The Q2 values were larger than 0.15 for SOW and larger than 0.35 for customer brand engagement. Figure 2 shows the results of the hypotheses testing. Controls: Community 0.466*** Gender (0.008, ns1, Age (-0.015, ns), Frequency of visits (0.210***) 0.042 (ns) Information 0.105*** 0.224*** Customer Brand Share of Enjoyment Engagement Wallet 0.186*** 0.105*** (2) Identity 0.173*** 0.223*** Perceived Economic Innovativeness Figure 2: Hypotheses testing (path coefficients) Notes: *** p < 0.01, 1 ns = Not significant, 2 Moderating effect As shown in Figure 2, of the proposed five motivational factors, four exhibit positive relationships with brand engagement, thus confirming H1 and H3-H5. The effects of community is the strongest (β = 0.466, p < 0.01), followed by the effects of economic motives (β = 0.223, p < 0.01) and identity motives (β = 0.186, p < 0.01). No relationship between information motives and engagement was found, thus rejecting H2. Moreover, customer brand engagement (H6) is positively associated with SOW (β = 0.224, p < 0.01), confirming H6. Of the control variables, frequency of visits (β = 0.210, p < 0.01) is positively associated with SOW whereas the effects of gender (β = 0.008, ns) and age (β = -0.015 ns) on SOW were not significant. The results of the moderating effects indicate that perceived innovativeness (H7) exerts a positive effect on the relationship between customer brand engagement and SOW, such that when perceived innovativeness is high, the link between customer brand engagement and SOW is strengthened. Without the moderating effect, the relationship between customer brand engagement and SOW is 0.224; with the significant moderating effect (0.105), this relationship is 0.329. The moderator therefore significantly strengthens the relationship so that the more strongly a 443 The Effects of Brand Engagement in Social Media on Share of Wallet customer perceives himself/herself as innovative; the stronger the relationship between brand engagement and SOW. Thus, H7 is accepted. We finally also examined the indirect effects of the five motivational drivers on SOW through brand engagement. The results reveal that community motives has the largest indirect effect on SOW (β = 0.105, p < 0.01). In sum, the results suggest that 1) community benefits is the strongest motivator of customer brand engagement in the social media context; 2) customer brand engagement is positively associated with SOW; and 3) customers’ innovativeness moderates the positive brand engagement-SOW relationship. 5 Conclusion Customer brand engagement is growing in importance in companies’ customer relationship and brand management activities along with the growth of social media. However, theories and conceptual models still need more empirical testing. Research is especially needed on the drivers of customer brand engagement in social media, how it affects consumers’ buying behaviour, and how consumers’ personality traits such as innovativeness affect these relationships. One of the key measures of behavioural loyalty is share of wallet (SOW). However, prior research is limited in examining the effect of customer brand engagement in social media on share of wallet (SOW) (see Brodie et al., 2011; Vivek et al., 2012). This study contributes to the customer brand engagement literature with three important findings. First, we identify four motivational drivers that positively influence consumers’ brand engagement in social media. The results indicate that the consumers’ who follow a brand in social media and receive benefits related to community, enjoyment, identity and economics are more intensively engaged with the brand than those receiving less benefits. Interestingly, information motives were not found to be related to engagement. Our findings confirm the existence of four motivational drivers of brand engagement in social media (Jahn & Kunz, 2012; Muntinga et al., 2011; Ouwersloot & Odekerken-Schröder, 2008) and add to the literature by identifying the community experience as the key driver of customer brand engagement in social media and finding no support for the effects of information motives on engagement. The latter is a unique finding and might be a special feature of Facebook brand communities that are built around the other four motives identified. Second, we make an important contribution to literature by investigating the relationship between brand engagement in social media and share of wallet. The relationship of customers’ engagement with a company’s Facebook site and the brand’s share of the customers’ spending has not been previously studied. Our results show that customer engagement is positively associated with SOW (c.f. Vivek et al., 2012). In other words, the percentage of the expenditure that engaged customers allocate to a brand is larger than those allocated by customers who are unengaged with the brand in social media. Finally, we add to current knowledge by showing that customers’ context-specific innovativeness positively affects the brand engagement and SOW relationship in the social media context. Thus, the higher the perceived innovativeness, the stronger is the positive relationship between brand engagement and SOW (c.f. 444 Heikki Karjaluoto, Juha Munnukka and Severi Tiensuu Citrin et al., 2000; Cotte & Wood, 2004; Joseph & Vyas, 1984). Therefore, among consumers with higher innovativeness brand engagement in social media is a stronger driver of SOW than among those with lower innovativeness. Three managerial implications arise from the findings. First, our results show four (out of five) motivational factors that drive engagement with brand in social media. Of these motives, community motives turned out to be the most important. Thus, we recommend managers to develop social media sites that foster especially we-intentions and belongingness (c.f. De Valck et al., 2009; Saho, 2009). In addition, we encourage managers to offer economic benefits on Facebook communities. Second, as the results confirm the positive link between brand engagement in social media and SOW, our results encourage brands to invest in fostering engagement in social media brand sites. Third, the results indicate that managers should implement strategies for social media in the light of the users’ perceived innovativeness and frequency of visits as they positively relate to SOW. A company should invest in creating up-to-date information and innovative activities to those customers that are identified as innovative and high- frequent visitors of the brand’s social media site. This would be the most effective way of driving increases in the brand’s share of the customers’ wallet. Finally, the study is concerned of limitations that offer opportunities for future studies. The sample can be biased towards more motivated users as participation was voluntary. Thus, in generalizing the results caution has to be made. Future studies should strive for data that includes also the respondents with less motived users of the brand. Although we minimized common method bias in the survey design, its effect can only be ruled out with longitudinal study design. Last, this study was concerned of only brands sold one household appliances store in Finland, which limits the generalization of these results to other types of brands or outside of Finland. Future research should be conducted in study a cross-country setting concentrating on different types of brands. References Bowman, D., Farley, J.U., & Schmittlein, D.C. (2000). Cross-national empirical generalization in business services buying behavior. Journal of International Business Studies, 31(4), 667–686. Boyd, D. (2008). Why youth (heart) social network sites: the role of networked publics in teenage social life. In Buckingham, D. (Ed.), MacArthur Foundation Series on Digital Learning – Youth, Identity, and Digital Media Volume (pp. 119–142). Cambridge, MA: MIT Press. Brodie, R.J., Hollebeek, L.D, Juric, B., & Ilic, A. (2011). Conceptual domain, fundamental propositions and implications for research. Journal of Service Research, 14(3), 252–271. Brodie, R.J., Ilic, A, Juric, B., & Hollebeek, L.D. (2013). Consumer engagement in a virtual brand community: An exploratory analysis. Journal of Business Research, 66(1), 105–114. Calder, B.J., Malthouse, E.C., & Schaedel, U. (2009). An experiental study of the relationship between online engagement and advertising effectiveness. Journal of Interactive Marketing, 23(4), 321–331. 445 The Effects of Brand Engagement in Social Media on Share of Wallet Calder, B.J., & Malthouse, E.C. (2008). Media engagement and advertising effectiveness. Hoboken, New Jersey: John Wiley & Sons. Citrin, C.A., Sprott, D.E., Silverman, S.N., & Stem Jr, D.E. (2000). Adoption of Internet shopping: the role of consumer innovativeness. Industrial Management & Data Systems, 100(7), 294–300. Cotte, J., & Wood, S.L. (2004). Families and innovative consumer behavior: a triadic analysis of sibling and parental influence. Journal of Consumer Research, 31(1), 78–86. De Valck, K., van Bruggen, G.H., & Wierenga, B. (2009). Virtual communities: marketing perspective. Decision Support Systems, 47(3), 185–203. De Wulf, K., Odekerken-Schröder, G., & Iacobucci, D. (2001). Investments in consumer relationships: a cross-country and cross-industry exploration. Journal of Marketing, 50(4), 33–50. Dholakia, U.M., Bagozzi, R.P., & Pearo, L.K. (2004). A social influence model of consumer participation in network- and small-group-based virtual communities. International Journal of Research in Marketing, 21(3), 241–263. Facebook (2013). Annual report 2013. Facebook Inc. Retrieved 19.12.2014, from http://investor.fb.com/annuals.cfm. Goldsmith, R.E., & Hofacker, C.F. (1991). Measuring consumer innovativeness. Journal of the Academy of Marketing Science, 19(3), 209–221. Gummerus, J., Lijander, V., Weman, E., & Pihlström, M. (2012). Customer engagement in a Facebook brand community. Management Research Review, 35(9), 857–877. Gwinner, K.P., Gremler, D.D., & Bitner, M.J. (1998). Relational benefits in service industries: the customer’s perspective. Journal of the Academy of Marketing Science, 26(2), 101–114. Heinonen, K. (2011). Consumer activity in social media: Managerial approaches to consumers’ social media behavior. Journal of Consumer Behavior, 10(6), 356– 364. Hennig-Thurau, T., Gwinner, K.P., Walsh, G., & Gremler, D.D. (2004). Electronic word- of-mouth via consumer-opinion platforms: what motivates consumers to articulate themselves on the internet? Journal of Interactive Marketing, 18(1), 38–52. Hollebeek, L. (2011). Exploring customer brand engagement: definition and themes. Journal of Strategic Marketing, 19(7), 555–573. Jahn, B., & Kunz, W. (2012). How to transform consumers into fans of your brand. Journal of Service Management, 23(3), 322–361. Joseph, B., & Vyas, S.J. (1984). Concurrent validity of a measure of innovative cognitive style. Journal of the Academy of Marketing Science, 12(1/2), 159–175. Kane, G.C., Fichman, R.G., Gallaugher, J., & Glaser, J. (2009). Community relations 2.0. Harvard Business Review, 87(11), 45–50. Keiningham, T.L., Aksoy, L., Perkins-Munn, T., & Vavra, T.G. (2005). The brand– customer connections. Marketing Management, 14(4), 33–37. Lu, J. Yao, J.E., & Yu, C.S. (2005). Personal innovativeness, social influences and adoption of wireless Internet services via mobile technology. The Journal of Strategic Information Systems, 14(3), 245–268. 446 Heikki Karjaluoto, Juha Munnukka and Severi Tiensuu McAlexander, J., Schouten, J., & Koenig, H. (2002). Building brand community. Journal of Marketing, 66(1), 38–54. McQuail, D. (1983). Mass communication theory: An introduction. London: Sage Publications. Men, L.R., & Tsai, W-H. S. (2013). Beyond liking or following: Understanding public engagement on social networking sites in China. Public Relations Review, 39(1), 13–22. Mersey, R.D., Malthouse, E.C., & Calder, B.J. (2012). Focusing on the reader: Engagement trumps satisfaction. Journalism & Mass Communication Quarterly, 89(4), 695–709. Muntinga, D.G., Moorman, M., & Smit, E. G. (2011). Introducing COBRAs: Exploring motivations for brand-related social media use. International Journal of Advertising, 30(1), 13–46. Ouwersloot, H., & Odekerken-Schröder, G. (2008). Who's who in brand communities – and why? European Journal of Marketing, 42(5/6), 571–585. Park, N., Kee, K.F., & Valenzuela, S. (2009). Being immersed in social networking environment: Facebook groups, uses and gratifications, and social outcomes. Cyber Psychology & Behavior, 12(6), 729–733. Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y., & Podsakoff, N.P. (2003). Common method biases in behavioral research: a critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879–903. Ringle, C. M., Wende, S., & Becker, J-M. (2014). SmartPLS 3. Hamburg: SmartPLS. Retrieved from http://www.smartpls.com Shao, G. (2009). Understanding the appeal of user-generated media: A uses and gratifications perspective. Internet Research, 19(1), 7–25. Steenkamp, J.E.M., Hofstede, F., &Wedel, F. (1999). A cross-national investigation into the individual and national cultural antecedents of consumer innovativeness. Journal of Marketing, 63(2), 55–69. Tufekci, Z. (2008). Grooming, gossip, Facebook and Myspace. Information, Communication & Society, 11, 544–564. van Doorn, J., Lemon, K.N., Mittal, V., Nass, S., Pick, D., Pirner, P., & Verhoef, P.C. (2010). Customer engagement behavior: Theoretical foundations and research directions. Journal of Service Research, 13(3), 253–266. Vivek, S.D., Beatty, S.E., & Morgan, R.M. (2012). Customer engagement: Exploring customer relationships beyond purchase. Journal of Marketing Theory and Practice, 20(2), 127–145. Zeithaml, V.A. (2000). Service quality, profitability, and the economic worth of customers: What we know and what we need to learn. Journal of the Academy of Marketing Science, 28(1), 67–85. 447 The Effects of Brand Engagement in Social Media on Share of Wallet Appendix List of survey items Community I am as interested in input from other users as I am in the content generated by company I like the company’s FB-site because of what I get from other users Company’s FB-site gets its visitors to converse or comment I have become interested in things, which I otherwise would not have, because of other users on the site Information I get good tips from the content The content shows me how people live The content helps me learn what to do or how to do it Enjoyment I find following content enjoyable Following content helps me improve my mood The content entertains me Identity Following content makes me a more interesting person Contributing to this content makes me feel like I belong in a group I want other people to know that I am reading this content Economic I write comments and/or like posts on virtual platform because of the incentives I can receive I write comments and/or like posts on virtual platform because I can receive a reward for the writing and liking Brand engagement I am an engaged member of this fan-page community I am an active member of this fan-page community I am a participating member of this fan-page community I engage in conversations and comment in company's FB-site I often like (like-function in FB) contents from company’s FB-site I use to contribute in conversations in company’s FB-site I often share company’s contents in FB Personal innovativeness If I heard about a new domestic appliance technology, I would look for ways of experimenting with it Among my peers, I am usually the first to explore new domestic appliance technologies I like to experiment with new domestic appliance technologies Share of wallet What percentage of your total expenditures for domestic appliance technologies do you spend for company’s products? Of the 10 times you select to buy domestic appliance technologies, how many times do you select company? 448 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Quasi-Empirical Scenario Analysis and Its Application to Big Data Quality Roger Clarke Xamax Consultancy Pty Ltd, Australia Visiting Professor, Research School of Computer Science, ANU, Canberra Visiting Professor, Faculty of Law, UNSW, Sydney Roger.Clarke@xamax.com.au Abstract Big data and big data analytics have been the subject of a great deal of positive discussion, not only in traditionally upbeat popular management magazines but also in nominally scientific and therefore professionally sceptical academic journals. Research was undertaken to assess the impact on the quality of inferences drawn from big data of the quality of the underlying data and the quality of the processes applied to it. Empirical study is difficult, however, because big data is emergent, and hence the phenomena are unstable, and likely to vary considerably across different settings. A research technique was accordingly sought that enabled theoretical treatment to be complemented by consideration of real-world data. The paper introduces Quasi- Empirical Scenario Analysis, which involves plausible story-lines, each commencing with a real-world situation and postulating lines of plot-development. This enables interactions among factors to be analysed, potential outcomes to be identified, and hypotheses to be generated. The set of seven scenarios that was developed investigated the nature and impacts of shortfalls in data and decision quality in a range of settings. In all cases, doubts arise about the reliability of inferences that arise from big data analytics. This in turn causes concern about the impacts of big data analysis on return on investment and on public policy outcomes. The research method was found to offer promise in the challenging contexts of technologies in the process of rapid change. Keywords: big data analytics, data quality, decision quality, scenario analysis 449 Roger Clarke 1 Introduction As sensor technologies have matured, and as individuals have been encouraged to contribute data into organisations' databases, more transactions than ever before have been captured. Meanwhile, improvements in data-storage technologies have resulted in the cost of evaluating, selecting and destroying old data being now considerably higher than that of simply letting it accumulate. The glut of stored data has greatly increased the opportunities for data to be inter-related, and analysed. The moderate enthusiasm engendered by 'data warehousing' and 'data mining' in the 1990s has been replaced by unbridled euphoria about 'big data' and 'data analytics'. What could possibly go wrong? Professional and management periodicals on big data topics have a very strong focus on opportunities, and so do most academic papers in the area to date. Far too little attention has been paid to the threats that arise from re-purposing data, consolidating data from multiple sources, applying analytical tools to the resulting collections, drawing inferences, and acting on them. The research reported on in this paper was conceived as a way to test the key claims made in the 'big data' literature, together with some of its implicit assumptions. The paper commences by briefly reviewing the emergent theory and current practice of big data. Working definitions are provided that distinguish the processes whereby the data becomes available from the processes whereby inferences are drawn from it. A short summary is then provided of theory relating to the central concepts of data quality and decision quality. Formal empirical research techniques are difficult to apply to emergent phenomena. A research technique is described which is intended to provide a quasi-empirical base to complement the theoretical analysis. A set of scenarios is outlined, and the theoretical material is then used as a lens to gain insights into the nature of big data and big data analytics, and the risks that they entail. 2 Big Data and Big Data Analytics During the 1980s, organisations had multiple data collections that were largely independent of one another. In order to enable the manipulation of data structures and content without disrupting underlying operational systems, copies of data were extracted from two or more collections and stored separately, in what was referred to as a 'data warehouse' – "a copy of transaction data specifically structured for query and analysis" (Jacobs 2010. See also Inmon 1992 and Kimball 1996). The processing of the contents of 'data warehouses' was dubbed 'data mining' (Fayyad et al. 1996, Ratner 2003, Ngai et al. 2009, Hall et al. 2009). During the same era, government agencies that could gain access to data-sets from multiple sources practised an additional technique called data matching. This involves comparing machine-readable records from different data-sets that contain data that appears to relate to the same real-world entities, in order to detect cases of interest. In most data matching programs, the category of entities that is targeted is human beings (Clarke 1994b, 1995a). A further technique that has long been applied in both the public and private sectors is profiling. This is "a technique whereby a set of characteristics of a particular class of person is inferred from past experience, and data-holdings are then searched for 450 Quasi-Empirical Scenario Analysis individuals with a close fit to that set of characteristics" (Clarke 1993). Although profiles can be ad hoc, they have been increasingly "supported by analyses of existing data-holdings within and beyond the organisation, whereby individuals who are known to belong to that class are identified, their recorded characteristics examined, and common features isolated". During the last few years, there has been a resurgence in enthusiasm for such techniques. The range and intensity of data capture has greatly increased in the intervening years. In addition, the economics of storage and destruction have shifted, such that retaining data is now cheaper than deleting it. This has led a wide variety of media commentators, consultants and excitable academics making enthusiastic pronouncements about revolutions, break-throughs and sparkling new opportunities. One of the most-quoted authorities on big data is Mayer-Schonberger & Cukier (2013), which had accumulated over 700 citations in its first 21 months after publication. According to those authors, the cornerstone of big data thinking is that 'datafication' – the expression of phenomena "in a quantified format so it can be tabulated and analyzed" (p.78). This they say undermines the kinds of analyses conducted in the past: "With enough data, the numbers speak for themselves. Petabytes allow us to say: Correlation is enough ... [W]hen you are stuffed silly with data, you can tap that instead [of experience, expertise and knowledge], and to greater effect. Thus those who can analyze big data may see past the superstitions and conventional thinking not because they're smarter, but because they have the data ... [S]ociety will need to shed some of its obsession for causality in exchange for simple correlations: not knowing why but only what ... Knowing why might be pleasant, but it's unimportant ... [L]et the data speak" (pp. 71, 143, 7, 52, 141). At its most extreme, this argument postulates that understanding is no longer necessary, and that the ready availability of vast quantities of data justifies the abandonment of reason: "If the statistics ... say it is, that's good enough. No semantic or causal analysis is required. ... [M]assive amounts of data and applied mathematics replace every other tool that might be brought to bear. Out with every theory of human behavior, from linguistics to sociology. Forget taxonomy, ontology, and psychology. ... [F]aced with massive data, [the old] approach to science — hypothesize, model, test — is becoming obsolete. ... Petabytes allow us to say: 'Correlation is enough'" (Anderson 2008). Such bombast is common in papers on big data. More sceptical views do exist, however. For example, big data is "a cultural, technological, and scholarly phenomenon that rests on the interplay of [three elements]" boyd & Crawford (2012, p.663). Their first two elements are technical, and are examined below. Their third element, on the other hand, emphasises the importance of "mythology: the widespread belief that large data sets offer a higher form of intelligence and knowledge that can generate insights that were previously impossible, with the aura of truth, objectivity, and accuracy". Where definitions of big data are offered, they tend to use imprecise and upbeat explanations, and to use a single term to encompass both the data and the techniques applied to it. A common expression is 'data that's too big, too fast or too hard for existing tools to process', and many writers refer to the distinguishing features being 'volume, velocity and variety'. Even the definition offered by an international agency is vague: "the capacity to analyse a variety of mostly unstructured data sets from sources as diverse as web logs, social media, mobile communications, sensors and financial 451 Roger Clarke transactions" (OECD 2013, p.12). An aspect that appears in some discussions is the accumulation of measurements of the same phenomenon over time, resulting in a longitudinal dimension within the data collection. This paper distinguishes 'big data' from 'big data analytics', and proposes working definitions, listed in Table 1. 'Big Data' A relatively large data collection that is associated with, or purports to be associated with, or can be interpreted as being associated with, particular entities. The following sub-categories can be usefully distinguished: • A single very large data collection • A consolidation of two or more data collections in such a manner that the association of data with specific entities is sustained or achieved. Three sub-categories can be usefully distinguished: • Merger into a single physical data collection • Interlinkage into a single virtual data collection, by means of: • Ephemeral Links, which exist only during the running of a particular analysis • Stored Links, which support multiple analyses 'Big Data Analytics' The techniques whereby a big data collection is used to draw inferences about an entity, or a category of entities, or relationships among entities Table 1: Working Definitions Earlier research performed by the author distinguished a number of categories of purpose to which big data analytics may be applied. These are extracted in Table 2. Big data analysis may test hypotheses about populations, which may be predictions from theory, existing heuristics, or hunches. Inferences may be drawn about the digital personae, which may be further applied to the population of entities that the data purports to represent, or to segments of that population. Profiles may be constructed for categories of entities that are of specific interest, such as heavy weather incidents, risk- prone card-holders, wellness recipes, or welfare cheats. Further functions that can be performed are concerned not with populations but with individual instances. Outliers of many different kinds can be identified. Inferences can be drawn about individual entities, directly from a particular digital persona, or from apparent inconsistencies within a consolidated digital persona, or through comparison of a digital persona against the population of similar entities, or against a previously- defined profile, or against a profile created through analysis of that particular big data collection. 452 Quasi-Empirical Scenario Analysis Population Focus • Hypothesis Testing This approach evaluates whether a particular proposition is supported by the available data • Population Inferencing This approach draws inferences about the entire population of (id)entities, or about sub-populations • Profile Construction This approach identifies key characteristics of some category of (id)entities. Individual Focus • Outlier Discovery Statistical outliers are commonly disregarded, but this approach regards them instead as valuable needles in large haystacks, because they may herald a 'flex-point' or 'quantum shift' • Inferencing about Individuals This approach draws inferences about individual entities within the population. Table 2: Functions of Big Data Analytics – After Clarke (2014b) The performance of these functions places a great deal of reliance on the analytical tools and the big data collections to which they are applied. The purpose of the research reported in this paper is to assess the impact of the quality of the big data, and of the big data analytics, on the quality of inferences drawn. The following section juxtaposes against big data notions the understanding accumulated within the information systems literature of the characteristics of data quality and decision quality. 3 Quality Theory During its first 50 years, the information systems (IS) discipline has adjusted its focus over time. Underlying many threads of IS theory, however, are characteristics of data that influence the effectiveness and efficiency of decision-making based on it. This section briefly summarises key aspects of data and decision quality. 3.1 Data Quality The computing and information systems literatures contain a body of material relating to data quality factors. See, for example, OECD (1980), Huh et al. (1990), Clarke (1995b, pp. 601-605), Wang & Strong (1996), Rusbridge et al. (2005), English (2006), Piprani & Ernst (2008), and the still-emergent ISO 8000 series of Standards. The primary factors are listed in Table 3. 453 Roger Clarke • Accuracy The degree of correspondence of the data with the real-world phenomenon that it is intended to represent, typically measured by a confidence interval, such as 'accurate to within 1 degree Celsius' • Precision The level of detail at which the data is captured, reflecting the domain on which valid contents for that data-item are defined, such as 'whole numbers of degrees Celsius' • Timeliness, which comprises distinct elements: • Temporal Applicability This reflects, for example, the date and time when the temperature was measured, the period during which an income-figure was earned, and the date after which a qualification or licence was applicable • Up-to-Dateness This reflects the lag between the real-world occurrence and the recording of the corresponding data. This is relevant to volatile items such as the current temperature and the balance owing on a credit account • Currency This reflects when the data-item was captured or was last authenticated, or the period over which an average was computed. This is relevant to volatile data-items, such as average temperature, total rainfall for the last 12 months, age, marital status and fitness for work • Completeness The availability of sufficient contextual information that the data is not liable to be misinterpreted Table 3: The Primary Data Quality Factors Each item of data is a measurement against some kind of scale. Some data arises from measurement against a ratio scale, and is capable of being subjected to analysis by powerful statistical tools. Frequently, however, data arises from measurements against cardinal, and against merely ordinal scales, such as Likert-scale data. Such data is capable of analysis using a smaller collection of statistical tools. However, researchers often assume for convenience that the data was gather against a ratio scale, in order to justify the application of more powerful inferencing techniques. Meanwhile, a great deal of data is collected against nominal scales, including text, sound, images and video. This supports only weak analytical tools, fuzzy matching and fuzzy logic. A further challenge arises where data that has been measured against different kinds of scales is consolidated and analysed. The applicability of analytical tools to mixed-scale data is more of a murky art than a precise science. 454 Quasi-Empirical Scenario Analysis A further consideration when assessing the quality of data is that what is collected, against what scales, and with what trade-offs between data quality and collection costs, all reflect the purpose of collection and the value-judgements of the sponsor. Particularly where data is collected frequently over time, data collection may also involve compression, through sampling, filtering and averaging. Further, where interesting outliers are being sought, compression is likely to ensure that the potentially most relevant data is absent from the collection. To be useful, data needs to be associated with real-world entities, with the data currently held about any particular entity being that entity's 'digital persona' (Clarke 1994, 2014a). The reliability of the association depends on the attributes that are selected as identifiers, and the process of association has error-factors. In some circumstances the link between the digital persona and the underlying entity is unclear (the condition referred to as pseudonymity), and in some cases no link can be achieved (anonymity). In order to protect important interests and comply with relevant laws, the link may be broken (the process of de-identification or anonymisation – UKICO 2012, DHHS 2012). On the other hand, rich data-sets are vulnerable to re-identification procedures. These problems afflict all big data collections, particularly those that are intended to support longitudinal studies. Where the data is sensitive, significant public policy issues arise. Over time, many threats arise to data integrity, including the loss of metadata such as the scale against which data was originally collected, the data's definition at the time of collection, the data's provenance, any supporting evidence for the data's quality, undocumented changes in meaning over time, and loss of contextual information that would have enabled its appropriate interpretation. The big data movement commonly features the use of data for a purpose extraneous to its original purpose – and indeed many big data proponents champion this proposition. The many data quality issues identified above are exacerbated by the loss of context, the lack of clarity about the trade-offs applied at the time of collection, and the greatly increased likelihood of misinterpretation. A further aspect of the big data movement is the step of physically or virtually consolidating data from multiple sources. This depends on linkages among data-items whose semantics and syntactics are different, perhaps substantially, perhaps subtly. The scope for misunderstandings and misinterpretation multiply. Theorists and practitioners perceive deficiencies in the data, such as missing elements, and inconsistencies among seemingly similar data-items gathered from two or more sources. To address these concerns about the analysts' particular perceptions of data quality, they devise data 'scrubbing', 'cleansing' or 'cleaning' processes. Some of these processes use an external, authoritative reference-point, such as a database of recognised location-names and street-addresses. Most, however, lack any external referent, and are merely based on 'logical data quality', i.e. internal consistency within the consolidated data-sets: "rules are learned from data, validated and updated incrementally as more data is gathered and based on the most recent data" (Saha & Srivastava 2014. See also Jagadish et al. 2014). Such process guidance as exists overlooks intrinsic and contextual data quality, and omits controls and audit (e.g. Guo 2013). As a result, the notion of 'cleanliness' primarily relates to the ease with which analytical tools can be applied to the data, rather than to the quality of the data. 455 Roger Clarke 3.2 Decision Quality It is feasible for big data analytics to be applied as a decision system. This may be done formally, by, for example, automatically sending infringement or 'show cause' notices. However, it is also possible for big data analytics to become a decision system not through conscious intent by an organisation, but by default. This can arise where a decision-maker becomes lazy, or is replaced by a less experienced person who is not in as good a position to apply human intelligence as a means of checking the reasonableness of the inferences drawn by software. Where decisions are made by analytics, a number of concerns arise about decision- quality. In Clarke (2014b), three were highlighted: • Relevance The relevance of the particular data to the particular decision needs to be demonstrated • Meaning The meaning of each particular item of data needs to be clear, as does the meaning of each particular value that each data-item adopts. The meaning of each individual item of data is capable of definition at the time it is gathered. However, since the decline of the resource-intensive waterfall method of software development, it is much less common for a data dictionary to be even established, let alone maintained. As a result, data definitions may be unclear, ambiguous and even implicit. The lack of clarity about the original meaning increases the likelihood that the meaning will be subject to various interpretations at any given time, that its meaning(s) will change over time, and hence that different interpretations of the same data-item or its content may be current, and that conflict may exist among mutually inconsistent interpretations • Transparency The decision-mechanism needs to be sufficiently transparent that those depending on it, those affected by it, and those reviewing it, can understand the basis on which the decision was made. Concerns have been expressed about transparency by a range of authors, e.g. Roszak (1986), Dreyfus (1992), boyd & Crawford (2012) Those theoretical aspects lead to practical questions. On what scale, with what accuracy and what precision, was the data collected that was instrumental in leading to the decision, and was the inferencing mechanism that was used really applicable to those categories of data? Did the data mean what the inferencing mechanism implicitly treated it as meaning? To the extent that data was consolidated from multiple sources, were those sources compatible with one another in respect of the data's scale, accuracy, precision, meaning and integrity? In many circumstances to which big data is claimed to be applicable, real-world decisions depend on complex models that feature confounding, intervening and missing variables. Correlations are commonly of a low grade, yet may nonetheless be treated, perhaps implicitly, as though the relationships were causal, and causal in one direction 456 Quasi-Empirical Scenario Analysis rather than the other. Big data proponents blithely dismiss causality and champion correlation instead. Yet mens rea (i.e. intention to cause the outcome) is a fundamental element of most criminal prosecutions, and many decisions in civil jurisdictions also focus on the proximate cause of an event. Big data proponents are swimming against a strong tide of cultural and institutional history. If decisions are being made that have real impacts, is the decision-process, and are the decision-criteria, transparent? And are they auditable? Are they subject to appeal processes and review? And can society have confidence that risks and liabilities are appropriately allocated? Big data analytics may be more commonly used as a form of decision support system, whereby a human decision-maker evaluates the inferences before applying them to any real-world purpose. However, the person may have great difficulty grasping the details of data provenance, data quality, data meaning, data relevance, and the rationale that have given rise to the recommendation. An even more problematic situation arises where the nominal decision-maker is not in a position to appreciate the rationale underlying the 'recommendation' made by the analytical procedure, and hence feels themselves to be incapable of second-guessing the system. With what were once called 'third-generation' development tools, the rationale was evident in the form of an algorithm or procedure, which may have been explicitly documented externally to the software, but was at least extractable by any person who had access to the source-code and who had the capacity to read it. The fourth generation of development tools merely expressed the decision-model in a more generally-applicable manner. The advent of the fifth-generation adopted a different approach, however. Rather than a model of the decision, this involved a model of the problem-domain. It became much more difficult to understand how the model (commonly expressed in logic, rules or frames) applied to particular circumstances. Some tools can provide a form of explanation of the rationale (or at least a list of the rules that were fired), but many cannot. Subsequently, with the sixth generation, an even greater barrier to understanding arose. With a neural network, there is no formal model of a decision or even of a problem- domain. There is just an empirical pile, and a series of inscrutable weightings that have been derived through mathematical processes, and that are then applied to each new instance (Clarke 1991). There have been expressions of concern from many quarters about the delegation of decision-making to software whose behaviour is fundamentally unauditable (e.g. Roszak 1986, Dreyfus 1992, boyd & Crawford 2012). 4 The Research Method An examination of the big data literature, reported in Clarke (2014b), concluded that, to date, there is limited evidence of the body of knowledge about data quality and decision quality being applied by the 'big data' movement, either by practitioners or researchers. The research question addressed by this research was accordingly: What is the impact of the quality of big data, and of big data analytics, on the quality of inferences drawn? 457 Roger Clarke In order to address that question, it is strongly desirable to conduct studies of real-world phenomena. However, conventional empirical research techniques are founded on some key assumptions, most relevantly that: • stable theories exist relevant to the research domain; • hypotheses about the research domain can be generated from those theories, which are explicit, unambiguous and refutable; and • the relevant aspects of the domain can be observed and measured, in such a manner that the hypotheses, if false, can be shown to be so. The big data research domain has a number of characteristics that present serious challenges to the conduct of empirical research. These are summarised in Table 4. • Unobservable Phenomena: • To some extent the big data domain is speculative and the phenomena do not, or do not yet, exist • To the extent that big data is being practised, the phenomena are difficult for researchers to observe, in particular due to the desire for secrecy of the corporations, government agencies and service- providers that are managing the data and conducting the analyses, variously because of the potential for competitive advantage and the risk that the activity is in breach of confidentiality or data protection laws • Unstable Phenomena: • The practices in relation to the collection and consolidation of data are changing • The analytical techniques being applied to data are also in a state of flux • To the extent that the data collection is longitudinal, the behaviours may be changing, rather than merely varying within a stable distribution or pattern • Highly Context-Dependent Phenomena: • The practices depend on the nature of the data, the expertise of the analysts, and the nature of the organisation by which or on whose behalf the big data is being gathered and the analyses are being performed Table 4: Characteristics of the Big Data Research Domain 458 Quasi-Empirical Scenario Analysis These characteristics present serious theoretical challenges. The selection and application of a body of theory is made difficult because each theory is relevant only to particular entities, relationships and/or environmental circumstances, and the underlying model may map to the reality at one time, but not at another. In addition, inferences drawn from the new context may or may not be relevant to the body of theory that was used to guide the research design. The characteristics also present practical challenges, because there are difficulties in reliably defining populations, population segments, and sampling frames, and in defining what is to be observed and what the terms mean that are used in interviews, questionnaires and case study reports. Another category of research can be envisaged, which is referred to here as 'quasi- empirical'. This draws on the technique of scenario analysis that has been used in futurology and long-term strategic planning since the 1960s, and is well-described in Wack (1985), Leemhuis (1985), Mobasheri et al. (1989), Schwarz (1991) and van der Heijden (1996). A scenario is a story-line that represents a composite or 'imagined but realistic world'. The prefix 'quasi' is an appropriate qualifier, because it is from the Latin, in use in English since the 15th century, meaning 'resembling', or 'seemingly but not actually'. Scenario analysis involves the preparation of a small set of scenarios that have the potential to provide insights into an emergent new context. The Quasi-Empirical Scenario Analysis (QuESA) technique proposed here involves a set of scenarios within which analytics are applied to big data collections. Each scenario is empirical to the extent that it commences with a setting-description that is drawn from contemporary phenomena. A story is then developed from the elements within the setting, by overlaying further data that is surmised, inferred or postulated, but is plausible or at least tenable, and by applying social and economic process sequences that are commonly observed in comparable settings. The purpose of the QuESA technique is emphatically not to generate assertions of predictive, explanatory or even descriptive standing. The intention is to assist in the emergence of insights, and in the formulation of hypotheses for testing by means of conventional empirical techniques as phenomena emerge, stabilise and become observable. In some respects, the QuESA technique might be compared with focus group research, in that the purpose of both is not to formally test hypotheses, but to gain insights and to bring to the surface potential hypotheses. In the QuESA technique, each scenario is built from a factual base, utilises real or at least realistic story elements, and is presented using narrative logic, i.e. plausible interactions and sequences. A single scenario is too limiting, firstly because it cannot capture sufficient of the richness of the research domain, and secondly because of the risk that the researcher or readers of reports on the research will lapse into predictive thinking rather than recognising the technique's purposes and limitations. Hence multiple scenarios are developed, addressing a requisite diversity of contexts, postulating events, interactions and sequences that are plausible within the particular context associated with the factual base. 459 Roger Clarke 5 Conduct of the Research This section describes the manner in which the QuESA technique was applied to big data and big data analytics, and presents the findings that resulted from the work. 5.1 The Scenarios A range of contexts was identified from the big data literature. In most cases, reports exist of applications of the specific kind referred to in the scenario, although in a few cases the application may be at this stage only aspirational. An endeavour was made to achieve diversity in the settings, in the nature of the data, in the nature of the analytical tools applied to it, and in the type of function being performed. It is infeasible to attempt representativeness in the sample, because the population is as-yet ill-defined, and indeed the dimensions across which the population varies are still under investigation. The text for the seven Scenarios is in Appendix 1. The basis on which each Scenario was developed is explained in Supplementary Materials on the author's web-site. See: http://www.rogerclarke.com/EC/BDSA.html#App2 Some of the Scenarios involve data that identifies or may identify specific individuals. This is the case with (2) Creditworthiness, (4) Foster Parenting, (6) Fraud Detection and (7) Insider Detection. In (3) Ad Targeting, the data relates to online identities that may but may not relate to a particular human being. One instance depends on de-identified patient data – (5) Cancer Treatment; whereas in Scenario (1) Precipitation Events, the data relates to environmental phenomena. In each case, starting with the factual base, various events or trajectories were postulated, which appeared to be tenable for that particular context. A narrative was then developed, and a story-line written. For example, Scenario (5) Cancer Treatment is based on a widely-quoted case in Mayer-Schonberger & Cukier (2013). An impact is postulated on research funding policies, which is a natural corollary of such a discovery. Because of the tight linkage between medical science research and industry, an associated development in the pharmaceutical industry is postulated. The impact on the population of young researchers is a natural demographic consequence. It is commonly the case in empirical research that subsequent work establishes that the correlations that were initially discovered were not as they seemed, and that the inferences drawn from the correlations need substantial qualification; and it is surmised that just such an eventuality will occur in this context as well. 5.2 Findings The inferences that are drawn from Quasi-Empirical Scenario Analysis are by definition not empirically based, but are 'insights' that arise from a plausible story-line, not from observation of the real world or of some more or less carefully-controlled proxy for the real world. The 'findings' reported in this section are accordingly hypotheses. The credibility of the hypotheses is weaker than that which arises from the application of a demonstrably relevant theory to a research domain – which are of course to be preferred if such a theories are available. On the other hand, the credibility of the hypotheses is considerably greater than that of ad hoc propositions and hunches, and greater than ideas generated on the basis of anecdotes alone. 460 Quasi-Empirical Scenario Analysis The first data quality issue discussed above related to the analysis of big data that was gathered against varying data scales. Challenges of these kinds are evident in Scenarios (3) Ad Targeting, (6) Fraud Detection and (7) Insider Detection. For example, data about online identities includes nominal data (interests), binary data although possibly with some data missing (gender), ordinal data (age-group), ratio-scale data (transaction counts and proportions), and longitudinal data. Few analytical techniques that are supported by mathematical statistics theory are available to cope with such mixed-mode data. The relationship between purpose of collection and the investment in original data quality was identified as a factor that may undermine the quality of inferences drawn. Administrative actions in relation to fraud, and particularly prosecutions, are undermined by poor-quality evidence. In some contexts, commercial liability might arise, e.g. Scenario (2) Creditworthiness, while in others a duty of care may be breached, e.g. Scenario (4) Foster Parenting. Loss of quality through inappropriate or inconsistent data compression, and through inappropriate or inconsistent handling of missing data over a timeline, arise in Scenario (1) Precipitation Events. Various uncertainties arise about identity, and hence about whether the data from different sources, and data collected over time, actually relate to the same real-world entity. This is significant in big health data contexts generally, e.g. Scenario (5) Cancer Treatment, and big social data applications, e.g. Scenario (4) Foster Parenting. The issues also loom large in Scenario (3) Ad Targeting, and – given the prevalence of identity fraud – in Scenario (2) Creditworthiness. Public demands for anonymisation/deidentification, and active endeavours to obfuscate and falsify identity, can be reasonably expected to exacerbate these challenges. Retrospective studies across long periods, as in Scenario (1) Precipitation Events, (4) Foster Parenting and (6) Fraud Detection, are confronted by problems arising from multiple sources with inconsistent or unclear syntax and semantics. The vagaries of 'data cleansing' techniques ensure that there will be instances of worsened data quality, resulting in both wrong inferences and spurious matches. This is a problem shared by all big data projects that depend on data consolidated from multiple sources. It afflicts Scenarios as diverse as (1) Precipitation, (5) Cancer Treatment and (7) Insider Detection. Of the decision quality issues, dubious relevance appears as an issue in (2) Creditworthiness, (5) Cancer Treatment, (6) Fraud Detection and (7) Insider Detection. In Scenario (6) Fraud Detection, data meaning is an issue, in that suspects may be unjustifiably identified as a result of inconsistencies arising from attempts to deceive, and from semantic issues even within a single database, let alone within a consolidation of multiple, inherently incompatible databases. Meanwhile, transparency of the decision mechanism results in the misallocation of resources in Scenario (5) Cancer Treatment, and in unfair discrimination in (2) Creditworthiness and (7) Insider Detection. 461 Roger Clarke 6 Conclusions Many applications of big data and big data analytics are not currently able to be studied because the organisations conducting them do not provide ready access. This paper has introduced and applied a new research technique, dubbed QuESA. Its purpose is to enable research to be undertaken in circumstances in which the gathering of the data necessary to support empirical research is not possible, not practicable, or not economic. The scenarios that have been used in this research are not case studies of specific instances of big data at work. They are story-lines, devised in order to encompass a range of issues not all of which are likely to arise in each particular real-life application. Each plot-line builds on factual foundations, then infiltrates into the stories additional elements that are plausible in the particular context. The intention of the scenarios was to test the assumptions underlying the big data value-proposition, not to pretend to be a substitute for deep case studies of actual experience. The big data movement involves data-collections that are of uncertain original quality, and lower current quality, and that have uncertain associations with real-world entities. Data collections are combined, by means of unclear validity, and modified by unaudited means, in order to achieve consolidated digital personae that have uncertain validity, and have uncertain relationships with any particular real-world entity. To this melange, powerful analytical tools are then applied. Theoretical analysis, complemented by the findings from the QuESA study, identifies a wide range of circumstances in which these problems can arise. Big data proponents claim that 'more trumps better'. More specifically: "With big data, the sum is more valuable than its parts, and when we recombine the sums of multiple datasets together, that sum too is worth more ... [W]e no longer need to worry so much about individual data points biasing the overall analysis" (Mayer-Schonberger & Cukier 2013, pp. 108, 40). The over-claiming by those authors extends beyond the general (where, in some circumstances, the much-overstated 'law of large numbers' actually does apply) to the specific: "Big data gives us an especially clear view of the granular" (p.13). As summarised by Leonelli (2014, p.3): "Big Data is viewed, through its mere existence, as countering the risk of bias in data collection and interpretation". These assertions lack adequate support by either theoretical argument or empirical evidence. In the face of the analysis reported on in this paper, they must be regarded as being at best wishful thinking, or self-delusional, or – because they are so often framed as recommendations to executives – reckless or even fraudulent. Given the uncertain quality of data and of decision processes, it appears that many inferences from big data projects may currently be being accorded greater credibility than they actually warrant. If that is the case, then resources will be misallocated. Within corporations, the impact will ultimately be felt in lower return on investment, whereas in public sector contexts there will be negative impacts on public policy outcomes, such as unjustified discrimination against particular population segments. When big data analytics are inappropriately applied to population inferencing and profile-construction, the harm that can arise includes resource misallocation and unjustified discrimination. When, on the other hand, inappropriate inferencing is about specific individuals, the costs are inevitably borne not by the organisation(s) involved but by the individuals, sometimes in the form of inconvenience, but sometimes with 462 Quasi-Empirical Scenario Analysis financial, service-availability, psychological, discrimination, or natural justice dimensions. When profiles generated by big data analytics are applied in order to generate suspects, the result is an obscure and perhaps counter-intuitive "predetermined characterisation or model of infraction" (Marx & Reichman 1984, p.429), based not on 'probable cause', but on a merely 'probabilistic cause' (Bollier 2010, pp.33-34). Not only does this result in unjustified impositions on the individuals concerned, but it also denies them natural justice because the lack of transparency relating to data and decision criteria means that the accusations are mysterious and even undefendable and they cannot get a fair hearing. The computer science and management literatures are remarkably lacking in discussion of the issues and the impacts examined in this paper. As a result, they have yet to mature into a literature on appropriate business processes for acquiring and consolidating big data, and applying big data analytics. The information systems literature could be expected to be more sceptical, and more helpful to decision-makers in business and government. Yet, as at the end of 2014, the entire AIS electronic library contained but one paper whose Abstract included both 'big data' and either 'risk assessment' or 'risk management'. The results of the research reported in this paper, limited though it was to an analytical and quasi-empirical base, make clear that much more care is warranted. 463 Roger Clarke Appendix 1: Big Data Scenarios (1) Precipitation Events Historical rainfall data has been gathered from many sources, across an extended period, and across a range of geographical locations. The collectors, some of them professional but mostly amateurs, used highly diverse collection methods, with little calibration and few controls. The data is consolidated into a single collection. A considerable amount of data manipulation is necessary, including the interpolation of data for empty cells, the arbitrary disaggregation of long-period data into the desirable shorter periods, and, in some cases, arbitrary disaggregation and reaggregation of data (e.g. to reconcile midnight-to-midnight with dusk-to-dusk recording times). Attempts are made to conduct quality audits against such sources as contemporaneous newspaper reports. However, these prove to be too slow and expensive, and are curtailed. Analytical techniques are applied to the data. Conclusions are reached about historical fluctations and long-term trends, with appropriate qualifications expressed. Analysts then ignore the qualifications and apply the data as though it were factual rather than a mix of facts and interpolations. Climate-change sceptics point to the serious inadequacies in the database, and argue that climate-change proponents, in conducting their crusade, have played fast and loose with scientific principles. (2) Creditworthiness A financial services provider combines its transactions database, its records of chargebacks arising from fraudulent transactions, and government statistics regarding the geographical distribution of income and wealth. It draws inferences about the risks that its cardholders create for the company. It uses those inferences in its decision- making about individual customers, including credit-limits and the issue of replacement and upgraded cards. Although not publicised by the company, this gradually becomes widely known, and results in negative media comments and recriminations on social media. Questions are raised about whether it conflicts with 'redlining' provisions in various laws. Discrimination against individuals based on the behaviour of other customers of merchants that they use is argued to be at least immoral, and possibly illegal, but certainly illogical from an individual consumer's perspective. The lender reviews the benefits arising from the technique, the harm done to its reputation, and the trade-off between the two. (3) Ad Targeting A social media services provider accumulates a vast amount of social transaction data, and some economic transaction data, through activity on its own web-sites and those of strategic partners. It applies complex data analytics techniques to this data to infer attributes of individual digital personae. Based on the inferred attributes of online identities and the characteristics of the available materials, the service-provider allocates third-party ads and its own promotional materials to the available space on web-pages. The 'brute force' nature of the data consolidation and analysis means that no account is taken of the incidence of partial identities, conflated identities, and obfuscated and falsified profiles. This results in mis-placement of a significant proportion of ads, to the 464 Quasi-Empirical Scenario Analysis detriment mostly of advertisers, but to some extent also of individual consumers. It is challenging to conduct audits of ad-targeting effectiveness, and hence advertisers remain unaware of the low quality of the data and of the inferences. The nature of the data exploitation achieves a considerably higher level of public consciousness as a result of the increasing incidence of inappropriate content appearing on childrens' screens. (4) Foster Parenting A government agency responsible for social welfare programs consolidates data from foster-care and unemployment benefits databases, and discovers a correlation between having multiple foster parents and later being chronically unemployed. On the basis of this correlation, it draws the inference that the longstanding practice of moving children along a chain of foster-parents should be discontinued. It accordingly issues new policy directives to its case managers. Because such processes lack transparency, and foster-children are young and largely without a voice, the new policy remains 'under the radar' for some time. Massive resistance then builds from social welfare NGOs, as it becomes apparent that children are being forced to stay with foster-parents who they are fundamentally incompatible with, and that accusations of abuse are being downplayed because of the forcefulness of policy directives based on mysterious 'big data analytics'. (5) Cancer Treatment Millions of electronic medical records reveal that cancer sufferers who take a certain combination of aspirin and orange juice see their disease go into remission. Research funding agencies are excited by this development, and transfer resources to 'big health data analytics' and away from traditional systemic research into causes, pathways and treatments of disease. Pharmaceutical companies follow the trend by purchasing homeopathic suppliers and patenting herb genes. The number of doctoral and post-doctoral positions available in medical science drops sharply. After 5 years, enough data has become available for the conclusion to be reached that the health treatments 'recommended' by these methods are ineffectual. A latter-day prophet emerges who decries 'the flight from reason', fashion shifts back to laboratory rather than digital research, and medical researchers slowly regain their previous high standing. The loss of momentum is estimated to have delayed progress by 10-15 years and generated a shortage of trained medical scientists. (6) Fraud Detection A company that has large sums of money flushing through its hands is under pressure from regulators, knows that stock exchanges run real-time fraud detection schemes, and accepts at face value the upbeat claims made by the proponents of big data analytics. It combines fraud-detection heuristics with inferences drawn from its large transaction database, and generates suspects. It assigns its own limited internal investigation resources to these suspects, and refers some of them to law enforcement agencies. The large majority of the cases investigated internally are found to be spurious. Little is heard back from law enforcement agencies. Some of the suspects discover that they are being investigated, and threaten to take their business elsewhere and to initiate 465 Roger Clarke defamation actions. The investigators return to their tried-and-true methods of locating and prioritising suspicious cases. (7) Insider Detection A government agency receives terse instructions from the government to get out ahead of the whistleblower menace, with Macbeth, Brutus, Iago, Judas Iscariot, Manning and Snowden invoked as examples of trusted insiders who turned. The agency increases the intrusiveness and frequency of employee vetting, and lowers the threshold at which positive vetting is undertaken. It applies big data analytics to a consolidated database comprising all internal communications, and all postings to social media gathered by a specialist external services corporation. To increase the pool of available information, it exercises powers to gain access to border movements, credit history, court records, law enforcement agencies' persons-of-interest lists, and financial tracking alerts. The primary effect of these measures is to further reduce employee loyalty to the organisation. To the extent that productivity is measurable, it sags. The false positives arising from data analytics explode, because of the leap in negative sentiments expressed on internal networks and in social media, and in the vituperative langauge the postings contain. The false positives greatly increase the size of the haystack, making the presumed needles even harder to find. The poisonous atmosphere increases the opportunities for a vindictive insider to obfuscate their activities and even to find willing collaborators. Eventually cool heads prevail, by pointing out how few individuals ever actually leak information without authority. The wave of over-reaction slowly subsides, leaving a bruised and demotivated workforce with a bad taste in its mouth. Acknowledgements This paper has benefited from valuable feedback from Kasia Bail, Lyria Bennett Moses, Russell Clarke, David Vaile and Graham Greenleaf, and from comments by reviewers. Aspects were presented to a Symposium on Legal Implications of Big Data, run by Prof. Dan Svantesson of Bond University, in Sydney, on 12 December 2014. References Anderson C. (2008) 'The End of Theory: The Data Deluge Makes the Scientific Method Obsolete' Wired Magazine 16:07, 23 June 2008, at http://archive.wired.com/science/discoveries/magazine/16-07/pb_theory Bollier D. (2010) 'The Promise and Peril of Big Data' The Aspen Institute, 2010, at http://www.ilmresource.com/collateral/analyst-reports/10334-ar-promise-peril-of-big-data.pdf boyd D. & Crawford K. (2012) 'Critical Questions for Big Data' Information, Communication & Society, 15, 5 (June 2012) 662-679, DOI: 10.1080/1369118X.2012.678878, at http://www.tandfonline.com/doi/abs/10.1080/1369118X.2012.678878#.U_0X7kaLA4M Clarke R. (1991) 'A Contingency Approach to the Software Generations' Database 22, 3 (Summer 1991) 23 - 34, PrePrint athttp://www.rogerclarke.com/SOS/SwareGenns.html 466 Quasi-Empirical Scenario Analysis Clarke R. (1993) 'Profiling: A Hidden Challenge to the Regulation of Data Surveillance' Journal of Law and Information Science 4,2 (December 1993), PrePrint at http://www.rogerclarke.com/DV/PaperProfiling.html Clarke R. (1994) 'The Digital Persona and Its Application to Data Surveillance' The Information Society 10,2 (June 1994), at http://www.rogerclarke.com/DV/DigPersona.html Clarke R. (1995a) 'Computer Matching by Government Agencies: The Failure of Cost/Benefit Analysis as a Control Mechanism' Information Infrastructure & Policy 4,1 (March 1995), PrePrint at http://www.rogerclarke.com/DV/MatchCBA.html Clarke R. (1995b) 'A Normative Regulatory Framework for Computer Matching' J. of Computer & Info. L. 13,3 (June 1995), PrePrint at http://www.rogerclarke.com/DV/MatchFrame.html Clarke R. (2014a) 'Promise Unfulfilled: The Digital Persona Concept, Two Decades Later' Information Technology & People 27, 2 (Jun 2014) 182-207, PrePrint at http://www.rogerclarke.com/ID/DP12.html Clarke R. (2014b) 'Quality Factors in Big Data and Big Data Analytics' Working Paper, Xamax Consultancy Pty Ltd, September 2014, at http://www.rogerclarke.com/EC/BDQF.html DHHS (2012) 'Guidance Regarding Methods for De-identification of Protected Health Information in Accordance with the Health Insurance Portability and Accountability Act (HIPAA) Privacy Rule’ Department of Health & Human Services, November 2012, at http://www.hhs.gov/ocr/privacy/hipaa/understanding/coveredentities/De- identification/guidance.html Dreyfus H.L. (1992) 'What Computers Still Can't Do: A Critique of Artificial Reason' MIT Press, 1992 English L.P. (2006) 'To a High IQ! Information Content Quality: Assessing the Quality of the Information Product' IDQ Newsletter 2, 3, July 2006, at http://iaidq.org/publications/doc2/english-2006-07.shtml Fayyad U., Piatetsky-Shapiro G. & Smyth P. (1996) 'From Data Mining to Knowledge Discovery in Databases ' AI Magazine 17, 3 (1996) 37-54, at http://aaaipress.org/ojs/index.php/aimagazine/article/download/1230/1131.. Guo P. (2013) 'Data Science Workflow: Overview and Challenges' ACM Blog, 30 October 2013, at http://cacm.acm.org/blogs/blog-cacm/169199-data-science-workflow-overview-and- challenges/fulltext Hall M., Frank E., Holmes G., Pfahringer B., Reutemann P. & Witten I.H. (2009) 'The WEKA Data Mining Software: An Update' SIGKDD Explorations 11, 1 (2009), at http://www.sigkdd.org/sites/default/files/issues/11-1-2009-07/p2V11n1.pdf van der Heijden K. (1996) 'Scenarios: The Art of Stategic Conversation' Wiley, 1996 Huh Y.U., Keller F.R., Redman T.C. & Watkins A.R. (1990) 'Data Quality' Information and Software Technology 32, 8 (1990) 559-565 Inmon B. (1992) 'Building the Data Warehouse' Wiley, 1992 Jacobs A. (2009) 'The Pathologies of Big Data' Communications of the ACM 52, 8 (August 2009) 36-44 Jagadish H.V., Gehrke J., Labrinidis A., Papakonstantinou Y., Patel J.M., Ramakrishnan R. & Shahabi C. (2014) 'Big data and its technical challenges' Communications of the ACM 57, 7 (July 2014) 86-94 Kimball R. (1996) 'The Data Warehouse Toolkit' Wiley, 1996 Leemhuis J.P. (1985) 'Using scenarios to develop strategies' Long Range Planning 18 (1985) 30-37 467 Roger Clarke Leonelli S. (2014) 'What difference does quantity make? On the epistemology of Big Data in biology ' Big Data & Society (April–June 2014) 1–11, at http://bds.sagepub.com/content/1/1/2053951714534395.full.pdf+html Marx G.T. & Reichman N. (1984) 'Routinising the Discovery of Secrets' Am. Behav. Scientist 27,4 (Mar/Apr 1984) 423-452 Mayer-Schonberger V. & Cukier K. (2013) 'Big Data: A Revolution That Will Transform How We Live, Work and Think' John Murray, 2013 Mobasheri F., Orren L.H. & Sjoshansi F.P. (1989) 'Scenario Planning at Southern California Edison' Interfaces (September/October 1989) 31-44 Ngai E.W.T., Xiu L. & Chau D.C.K. (2009) 'Application of data mining techniques in customer relationship management: A literature review and classification' Expert Systems with Applications, 36, 2 (2009) 2592-2602 OECD (1980) 'Guidelines on the Protection of Privacy and Transborder Flows of Personal Data' OECD, Paris, 1980 OECD (2013) 'Exploring Data-Driven Innovation as a New Source of Growth: Mapping the Policy Issues Raised by "Big Data"' OECD Digital Economy Papers, No. 222, OECD Publishing, at http://dx.doi.org/10.1787/5k47zw3fcp43-en Piprani B. & Ernst D. (2008) 'A Model for Data Quality Assessment' Proc. OTM Workshops (5333) 2008, pp 750-759 Ratner B. (2003) 'Statistical Modeling and Analysis for Database Marketing: Effective Techniques for Mining Big Data' CRC Press, June 2003 Roszak T. (1986) 'The Cult of Information' Pantheon 1986 Rusbridge C., Burnhill P., Seamus R., Buneman P., Giaretta D., Lyon L. & Atkinson M. (2005) 'The Digital Curation Centre: A Vision for Digital Curation' Proc. Conf. From Local to Global: Data Interoperability--Challenges and Technologies, Sardinia, 2005, pp. 1-11, at http://eprints.erpanet.org/archive/00000082/01/DCC_Vision.pdf Saha B. & Srivastava D. (2014) 'Data Quality: The other Face of Big Data', Proc. ICDE Conf., March-April 2014, pp. 1294 - 1297 Schwartz P. (1991) 'The Art of the Long View: Planning for the Future in an Uncertain World' Doubleday, 1991 UKICO (2012) 'Anonymisation: managing data protection risk: code of practice' Information Commissioners Office, November 2012, at http://ico.org.uk/for_organisations/data_protection/topic_guides/~/media/documents/library/Dat a_Protection/Practical_application/anonymisation-codev2.pdf Wack P. (1985) 'Scenarios: Uncharted Waters Ahead' Harv. Bus. Rev. 63, 5 (September- October 1985) 73-89 Wang R.Y. & Strong D.M. (1996) 'Beyond Accuracy: What Data Quality Means to Data Consumers' Journal of Management Information Systems 12, 4 (Spring, 1996) 5-33 468 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia A Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance Heikki Karjaluoto University of Jyväskylä, Finland heikki.karjaluoto@jyu.fi Hanna Mäkinen University of Jyväskylä, Finland hanna.u.makinen@gmail.com Joel Järvinen University of Jyväskylä, Finland joel.jarvinen@jyu.fi Abstract There is growing debate about whether companies’ investments in social media really pay off. This research attempts to contribute to this discussion by investigating the relationship between a firm’s social media activity and two outcomes, namely, company reputation and firm financial performance. Two propositions are developed based on theory and tested with a sample of 59 companies. The findings suggest that a firm’s social media activity is only partially linked with its financial performance and is not linked with corporate reputation. The implications of the research suggest that little is known about the relationships between a firm’s social media activity and corporate reputation and financial outcomes. Keywords: Social media activity, corporate reputation, firm performance 1 Introduction Currently, no company can say it is not affected by social media. Even if a company is not active in social media, communication about its brands still occurs in those channels 469 Heikki Karjaluoto, Hanna Mäkinen, Joel Järvinen (Kietzmann et al., 2011). Social media has caused consumers to be more demanding; therefore, one-way communication from companies to consumers is no longer sufficient (Trainor, 2012). Instead, the importance of communication and conversation with consumers is emphasized (Jones et al., 2009) as consumers want companies to listen to them as well as engage with and respond to them (Kietzmann et al., 2011). Consumers want to participate, interact, and create value by themselves. As a matter of fact, it can be said that power has shifted from companies to consumers (Bunting & Lipski, 2000). For this reason, many companies have increased their social media activity. Furthermore, the financial crisis led companies to seek more cost-effective marketing methods, and social media has become a good channel for this (Kirtiş & Karahan, 2011). Despite that, there are still many companies who do not sufficiently understand social media and so just ignore it (Kietzmann et al., 2011.) Therefore, it is important to determine if it is beneficial for a company to be active in social media and what the advantages are of being active. Literature about the subject is growing, but there is still little scholarly evidence that addresses how the utilization of social media influences company performance. Prior evidence has suggested that social media is important from a public relations (PR) and reputation point of view (Aula, 2010; Firestein, 2006) whereas others have linked it to overall equity value (Luo et al., 2010). Nevertheless, little research has been conducted on how a company’s appearance and activity in social media affect reputation and firm performance. Hence, this study aims to fill this gap in the literature by examining the relationship between social media activity and three variables, namely, corporate reputation, firm performance, and firm size (control variable). This study analyzes the relationships between the constructs but does not propose or discuss the direction of these effects or, in other words, the causal linkages between the variables. 2 Literature Review Social media is defined in various ways, but all the definitions share the idea that it is a way to connect and interact with other people using various communication techniques through online media (Kietzmann et al., 2011; Kirtiş & Karahan, 2011; Ryan & Jones, 2009, p.152). While traditional media is focused on delivering a message (outbound marketing), social media is user-driven (inbound marketing) and includes, for example, building relationships and conversing with an audience (Drury, 2008). Social media consists of various channels and platforms that allow communication, networking, and sharing of content and information (Bowman et al., 2012; Kietzmann et al., 2011). Social media has been classified into six categories based on two key elements: media research (social presence, media richness) and social processes (self-presentation, self- disclosure) (Kaplan & Haenlein, 2010). The greater the social presence/media richness, the greater the social influence of users on the behavior of other users. The greater the self-presentation/self-disclosure, the more willing people are to talk about and reveal aspects of themselves to others. The six categories based on these elements are blogs, social networking sites, virtual social worlds, collaborative projects, content communities, and virtual game worlds (Kaplan & Haenlein, 2010). In this study, we focus on social networking sites (Facebook, LinkedIn), content communities (YouTube), and one form of blogging, namely, microblogging (Twitter) as these are the 470 Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance most widely adopted social media tools among companies. On social networking sites, users can create profiles, share information, including photos, videos, audio files, and blogs, and ask friends to join or connect to their profiles (Kaplan & Haenlein, 2010). Facebook is to date the largest and most popular social media channel (Funk, 2011, 54; Bodnar & Cohen, 2012, p.127) where users find and add friends and contacts to share content through personal profiles (Berthon, 2012). LinkedIn is a business-networking tool that is more focused on networking (Kietzmann, 2011) with other professionals or companies. On the consumer side, LinkedIn is not used for finding customers (Funk, 2011, p.63), but on the business-to-business side, it is used for acquiring customers, gathering market information, and recruiting, for instance (Bodnar & Cohen, 2012, p. 97). Content communities allow people to share content with other users who can comment on it. Content communities include, for example, YouTube for sharing videos and Flickr for sharing photos. Considering the 100 million videos that YouTube serves up each day, it is easy to understand the wide accessibility of content communities (Kaplan & Haenlein, 2010). More recently, YouTube has also been used as a channel for publishing video blogs (Biel & Gatica-Perez, 2013). The most popular microblogging service is Twitter, which allows people to send and read short messages of 140 characters or less from their profile to users who follow them (Berthon, 2012). It is also possible to add links to other pages or send direct messages to other users by including a username in a post (in the form of @username) (Funk, 2011, p. 57). Social media has become a tool for companies to communicate and engage with customers at lower cost and more effectively than traditional channels (Kaplan & Haenlain, 2010). At the present, social media is increasingly being seen as a tool for creating and maintaining customer relationships and, for this reason, has also become an important tool for CRM (Trainor, 2012). Customers can no longer be seen simply as objects for a sale but instead must be considered as decision makers with their own needs and the option to choose what and where they are buying. Just as the entire web has turned into a social web, the customer has also turned into a social customer, standing at the center of the business ecosystem (Greenberg, 2010). These social customers also have social needs and by filling them companies can build long lasting and meaningful relationships with their customers. The idea of social customers significantly affects companies and drives the need for social CRM (Greenberg, 2010), which can be considered a CRM strategy that emphasizes customer relationship communication via new communication technologies. 2.1 Social Media and Corporate Reputation Corporate reputation refers to how stakeholders perceive a company and how the company responds to those perceptions (Williams, 2005). A strong corporate reputation provides a competitive advantage to a company and is very difficult for others to imitate (Hall, 1993). The more a company differentiates itself from its competitors, the stronger its reputation and the greater its reputational capital (Fombrun, 1996, p.392). Hence, reputation can be seen as an intangible and strategic asset of a company (Eberl & Schwaiger, 2005). Earlier studies have shown that company reputation is positively linked with customer loyalty (Keh, 2009), attractiveness of company offerings 471 Heikki Karjaluoto, Hanna Mäkinen, Joel Järvinen (Fombrun & Van Riel, 2003, p. 8), employee commitment and job satisfaction (Alniacik, 2011; Helm, 2011), reduction of transaction (Walsh, 2006) and operating costs ((Fombrun, 1996, p.75), and firm performance (Eberl & Schwaiger, 2005; Carmeli & Tishler, 2005). In traditional media, corporate reputation is seen as the interaction between a company’s communicative actions and stakeholders’ reactions. Accordingly, when a company communicates through their marketing channels, for example, their reputation depends on how stakeholders perceive the message and react to it (Bunting & Lipski, 2000). In the social media era, it is not enough merely to communicate a message to consumers; instead, companies must engage consumers in conversation through social media (Jones et al., 2009). In the social media era, companies have lost the power to control discussion about them, which makes it more difficult to influence reputation (Aula, 2010). Social media has enhanced customers’ need to be active and, at the same time, is what enables companies to address these needs. Customers want to participate, interact, and create value on their own, and social media provides an opportunity for companies to enable that through participation and interaction with customers. This participation leads to greater involvement with and commitment to a company, which has been suggested to increase customer satisfaction (Trainor, 2012), which, again, has been proven in earlier research to lead to better corporate reputation (Carmeli & Tishler, 2005). Therefore, a positive relationship between a firm’s social media activity and its corporate reputation can be proposed: Proposition 1: A firm’s social media activity and corporate reputation have a positive relationship. 2.2 Social Media and Firm Performance Many companies are interested in learning how to benefit financially from social media. To justify investment, it is essential to determine the financial value of social media (Luo et al., 2013; Gilfoil & Jobs, 2012). The use of social media has expanded exponentially among both consumers and corporations, but still only a small amount of money earmarked for marketing is dedicated to social media. One reason for this is the difficulty of measuring the value of investments in social media marketing (Gilfoil & Jobs, 2012). Despite the difficulty of measuring the value of social media, many researchers have attempted to measure its return on investment (ROI). However, the value of a customer to a firm is not only the amount of money they spend, but also the influence they have on other people’s opinions by spreading their thoughts through social media; the actual value of a customer is based on far more than their spending (Fisher, 2009). Measuring and calculating social media ROI begins with measuring costs and then attempting to determine the return on sales. When dealing with social media, however, this is insufficient. Companies should also consider which marketing objectives are met by social media utilization. These can be, for example, brand engagement, providing knowledge about new products, or delivering information to customers. Hence, returns are not always financial (Hoffman & Fodor, 2010) or recognized in the short-term. 472 Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance Nevertheless, several studies have examined the relationship between social media and financial figures (Schniederjans, 2013; Trainor, 2012; Yu et al., 2012). Luo et al. (2013) suggested that social media has strong predictive power of a firm’s future equity value. Their research revealed that through positive blog posts, consumers’ trust and advocacy can be improved, leading to higher firm value. Naturally, negative blog posts can instead harm reputations, leading to weaker firm performance (Luo et al., 2013). Schniederjans (2013) studied the relationship between firm performance and social media from the perspective of impression management and found a partial positive connection between the use of social media and financial performance, depending on the impression management strategy used. Yu et al. (2012) argued that social media has a stronger impact on firm stock performance than conventional media does. Additionally, compared to other media, effects from social media occur much faster. Furthermore, the harmful effects from negative opinions and ideas occur more quickly than the beneficial effects from positive opinions and ideas. Therefore, negative publicity in social media in particular can rapidly affect firm performance (Luo et al., 2013). Social media has fostered customers who are more willing to participate in company activity, which may also increase commitment to the company. It has been suggested that customers’ higher involvement and commitment to a company increase satisfaction and loyalty (Trainor, 2012), which, in earlier studies, was revealed to have a positive impact on firm equity value (Anderson et al., 2004, Luo et al., 2010). Proposition 2: Social media activity and firm performance have a positive relationship. 3 Methodology Three data collection procedures were used in this study. First, self-gathered data on companies’ social media activity were collected from selected companies’ social media channels between March 2014 and April 2014. Social media channels selected for this research were Facebook, Twitter, LinkedIn, and YouTube. Data consisted of number of likes, talks, and activity (Facebook); number of followers, tweets, and activity (Twitter); number of followers and activity (LinkedIn); and number of subscribers, views, and videos (YouTube). The channels were chosen because they are the most popular social media channels currently used by firms. The amount of activity depends on the duration a given social media channel was utilized by a company. Activity was evaluated by the amount of activity such as Facebook posts a company made on their page during 2013 so it could be compared among companies. Detailed classification and measurement of activity is shown in Table 1. 473 Heikki Karjaluoto, Hanna Mäkinen, Joel Järvinen Activity on Facebook Scale No Facebook page 1 Facebook page but no activity 2 Less than 50 posts a year 3 50-100 posts a year 4 100-500 posts a year 5 More than 500 posts a year 6 Amount of likes on posts No Facebook page 1 Less than 10 likes /post 2 10-50 likes /post 3 50-100 likes /post 4 100-1000 likes /post 5 More than 1000 likes /post 6 Activity on Twitter No Twitter account 1 An account but no tweets 2 Less than every other day 3 Max once a day 4 1-2 tweets a day 5 Many tweets a day 6 Activity on LinkedIn No LinkedIn profile 1 Profile but no activity 2 Profile but no regular activity 3 1-5 posts a month 4 5-15 posts a month 5 More than 15 posts a month 6 Table 1: Classification of Social Media Activity Some companies did not have a profile on Facebook for the whole of 2013. For these companies, the number of posts used for analysis was calculated by dividing the number of posts by the number of months the company had a Facebook presence and multiplying the result by 12 for a whole year average. Second, corporate reputation data were obtained from a market research firm. The data are from a 2013 survey that measured the reputation and responsibility of Finnish companies (see Appendix Figure 1). The survey had 9,802 respondents from different age, gender, and regional groups in Finland and examined 59 companies that operate in the country. These same 59 companies were chosen as the sample for this research in order to have comparable data about reputation. The companies operate in seven different industries: food, retail, service, finance, energy, industry, and ICT. The reputation index covers different dimensions. The survey questions evaluated perceptions of five different fields: overall evaluation of reputation, impression about the company, trust, financial success, and the quality of products and services. The 474 Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance questions were divided into three dimensions so that questions dealing with overall evaluation of reputation explained overall reputation; questions dealing with impression of the company and trust explained relationship and emotional attraction; and questions dealing with financial success and quality of the products and services explained competence and rational attraction. Third, data on firm performance and size (used as a control variable) were gathered from secondary sources, i.e., companies’ annual reports. The data included revenue, profits, and the number of personnel in the companies for financial year 2013. If financial data from 2013 were not available, data from 2012 were used. Net profit was used to indicate firm performance and revenue, and number of personnel was used to indicate company size. The relationships between the study constructs were analyzed with two-tailed Pearson’s correlation. 4 Results The largest share was in the food and industry groups with 13 companies (22 %) in each. The second largest group was energy with nine companies (15.3 %). The next largest group was finance with eight companies (13.6 %) followed by service and retail with six companies (10.2 %) each. The smallest group was ICT with four companies (6.8 %). Most companies were B2C (75%). Of the 59 companies, almost all had a LinkedIn page (97%) and YouTube channel (83%). Approximately three out of four had a Facebook presence (76%) and Twitter account (73%). The ICT and service sectors were the most active on social media. All companies in the ICT sector participated in all four types of social media (Facebook, Twitter, LinkedIn, and YouTube). In addition, all companies in the service sector participated in all four types of social media, with the exception of two companies that did not have Twitter accounts. Energy and retail were the most inactive sectors, with 33% of companies having no Facebook, Twitter, or YouTube accounts. In the retail sector, 83% of companies had no Twitter account. A majority of the numeric data were not classified but used directly as ratio variables. On Facebook, a majority of the companies (49%) made 50 to 100 posts during 2013 (Table 2). Activity on Facebook N % No Facebook page 14 23.7 Facebook page but no 0 0 activity Less than 50 posts a year 2 3.4 50-100 posts a year 29 49.2 100-500 posts a year 11 18.6 More than 500 posts a year 3 5.1 Activity on Twitter No Twitter account 18 30.5 An account but no tweets 2 3.4 475 Heikki Karjaluoto, Hanna Mäkinen, Joel Järvinen Less than every other day 6 10.2 Max once a day 13 22.0 1-2 tweets a day 5 8.5 Many tweets a day 15 25.4 Activity on LinkedIn No LinkedIn profile 5 8.5 Profile but no activity 11 18.6 Profile but no regular activity 9 15.3 1-5 posts a month 19 32.2 5-15 posts a month 11 18.6 More than 15 posts a month 4 6.8 Table 2: Firms’ Social Media Activity Every company with a Facebook profile had at least some activity, and only two companies (3.4%) made fewer than 50 posts during the year. One-fourth (25%) of the companies tweeted multiple times per day. On LinkedIn, the largest group, 19 companies (32.2%), made one to five posts per month. Social media activity was also related to whether the company was serving consumers (B2C) or other organizations (B2B). In general, B2C companies were found to be more active on Facebook, Twitter, and YouTube whereas B2B companies were more active on LinkedIn. 4.1 The Relationship Between a Firm’s Social Media Activity and Corporate Reputation Within the overall sample, social media activity and company reputation (based on reputation index values) are not related (at the p < 0.05 level). However, when examining this relationship between industries and in B2B versus B2C companies, significant correlations were found. In the service industry, there is a positive relationship among Facebook activity ( r = 0.843, p < 0.01), number of Twitter followers ( r = 0.976, p < 0.01), tweets ( r = 0.976, p < 0.01), and reputation. In the food industry, reputation correlates with number of LinkedIn followers ( r = 0.976, p < 0.01) and LinkedIn activity ( r = 0.577, p < 0.05). In this industry, there were also correlations between YouTube channel subscribers and reputation ( r = 0.690, p < 0.05) and between YouTube views and reputation ( r = 0.663, p < 0.05). When examining B2B and B2C companies separately, the only correlation found was among B2B companies between reputation and people talking about a company on Facebook ( r = 0.784, p < 0.05). 4.2 The Relationship Between Firm’s Social Media Activity and Firm Performance Within the overall sample, when examining the relationship between variables of Facebook activity (likes, talking about, activity) and financial numbers (net revenue, net profit), the only significant correlations were between the amount of people talking 476 Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance about a company on Facebook and net revenue ( r = 0.313, p < 0.05) and between the amount of people talking about a company on Facebook and a control variable, namely, number of personnel ( r = 0.500, p < 0.01). This indicates that people are talking more about companies that are larger and more profitable on Facebook. No significant relationships were observed between net profit and other values. With respect to Twitter, significant correlations were found between the number of tweets and net profit ( r = 0.384, p < 0.05), between the number of tweets and number of personnel ( r = 0.591, p < 0.01), and between overall activity and net revenue ( r = 0.280, p < 0.05). This indicates that more profitable companies and companies with more personnel made more tweets, and companies with higher net revenue had more yearly activity. On LinkedIn, significant correlations were found between activity and net revenue ( r = 0.311, p < 0.05), between activity and net profit ( r = 0.319, p < 0.05), and between activity and number of personnel ( r = 0.391, p < 0.01). This indicates that more profitable companies and companies with more personnel are more active on LinkedIn. Amount of followers did not correlate with revenue, profit, or number of personnel. On YouTube, there were no significant correlations between any variables (subscribers, amount of videos, views, financial figures). 5 Discussion The aim of the study was to examine the relationship between a company’s activity on social media and corporate reputation and firm performance. The two propositions developed were not supported by the data. The first proposition claimed that activity on social media would have a positive relationship with corporate reputation. This proposition is based on the idea that customers want to participate, interact, and create value by themselves, and social media provides the possibility for companies to participate and interact with customers, leading to higher involvement and commitment and greater bigger customer satisfaction (Trainor, 2012), which again has been proven to lead to better corporate reputation (Carmeli & Tishler, 2005). This leads to the conclusion that social media has an indirect effect on corporate reputation. The second proposition argued that firm’s activity on social media would have a positive relationship with firm performance. This proposition is based on the idea that social media usage can increase customers’ involvement and commitment to a company, leading to increased satisfaction and loyalty (Trainor, 2012), which again have been found to have a positive impact on firm equity value (Anderson et al., 2004, Luo et al., 2010). Hence, an indirect effect of social media participation is better firm performance. However, this study provides some evidence of the partial relationship between social media and firm performance. The relationship was identified in use of Twitter and LinkedIn. Larger companies with more personnel had more tweets overall and were also more active on LinkedIn than smaller companies with fewer personnel. Additionally, companies with larger net revenue were more active on both Twitter and LinkedIn than were companies with smaller net revenue. More profitable companies (with larger net profit) registered more activity on LinkedIn and more tweets than less profitable companies did. On Facebook, people talked more about profitable and larger companies, but a companies’ own activity on Facebook had no relationship to firm 477 Heikki Karjaluoto, Hanna Mäkinen, Joel Järvinen performance or size. On YouTube, there was no relationship between companies’ activity and firm performance and size. The main theoretical contribution of the study is that it demonstrates that the more active companies have better reputations than that are not active in social media. However, a partial link can be found between social media activity and firm performance. In terms of managerial implications, this research presents a good argument that managers should not believe that simply being active in different social media channels is sufficient to enhance corporate reputation or increase financial performance. Even if a company itself is active on social media, reputation and financial performance are not inherently positive as a result. These rather counterintuitive findings are limited by the sampling frame, analysis methods, and firm perspective. Despite companies’ activity on social media, most of the things that happen in social media occur regardless of how active companies are in this arena. Social media has caused power to shift from companies to consumers and, as a result, companies’ ability to control their reputation and what is said about them on the Internet has diminished (Bunting and Lipski, 2000). Nevertheless, companies have not lost all their power as they still control the rules and framework of how the company and its brands participate in social media. Companies, can decide, for example, what is posted, who is posting, and where it is posted (Hoffman & Fodor, 2010). Thus, we encourage researchers to further examine the relationship between firms’ social media activity and corporate reputation/firm performance. References Anderson E.W., Fornell C., & Mazvancheryl, S.K. (2004). Customer satisfaction and shareholder value. Journal of Marketing, 68(4), 172–185. Alniacik, U. (2011). Independent and joint effects of perceived corporate reputation, affective commitment and job satisfaction on turnover intentions. Procedia - Social and Behavioral Sciences, 24, 1177–1189. Aula, P. (2010). Social media, reputation risk and ambient publicity management. Strategy & Leadership, 38(6), 43–49. Berthon, P. R. (2012). Marketing meets Web 2.0, social media, and creative consumers: Implications for international marketing strategy. Business Horizons, 55(3), 261– 271. Biel, J., & Gatica-Perez, D. (2013). The YouTube lens: Crowdsourced personality impressions and audiovisual analysis of vlogs. IEEE Transactions on Multimedia, 15(1), 41–55. Bodnar, K., & Cohen, J. L. (2011). B2B social media book: Become a marketing superstar by generating leads with blogging, LinkedIn, Twitter, Facebook, email, and more. Hoboken, NJ, USA: Wiley. Bowman, N. D., Westerman, D. K., & Claus, C. J. (2012). How demanding is social media: Understanding social media diets as a function of perceived costs and benefits – A rational actor perspective. Computers in Human Behavior, 28, 2298– 2305. 478 Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance Bunting, M., & Lipski, R. (2000). Drowned out? Rethinking corporate reputation management for the Internet. Journal of Communication Management, 5(2), 170- 178. Carmeli, A. A. C., & Tishler, A. (2005). Perceived organizational reputation and organizational performance: An empirical investigation of industrial enterprises. Corporate Reputation Review, 8(1), 13–30. Drury, G. (2008). Opinion piece: Social media: Should marketers engage and how can it be done effectively? Journal of Direct, Data and Digital Marketing Practice, 9(3), 274–277. Eberl, M., & Schwaiger, M. (2005). Corporate reputation: disentangling the effects on financial performance. European Journal of Marketing, 39 (7/8), 838–854. Gilfoil, D. M., & Jobs, C. (2012). Return on investment for social media: A proposed framework for understanding, implementing, and measuring the return. Journal of Business & Economics Research, 10(11), 637–650. Greenberg, P. (2010). The impact of CRM 2.0 on customer insight. Journal of Business & Industrial Marketing, 25(6), 410–419. Firestein, P. J. (2006). Building and protecting corporate reputation. Strategy & Leadership, 34(4), 25–31. Fisher, T. (2009). ROI in social media: A look at the arguments. Journal of Database Marketing & Customer Strategy Management, 16(3), 189–195. Fombrun, C. J. (1996). Reputation: Realizing value from the corporate image. Cambridge, MA: Harvard Business School Press. Fombrun, C. J., & Van Riel, C. B. M. (2003). Fame and fortune: How successful companies build winning reputations. Upper Saddle River, NJ: Pearson Education. Funk, T. (2011). Social media playbook for business: Reaching your online community with Twitter, Facebook, Linkedin and more. Santa Barbara, CA, USA: ABC- CLIO. Hall, R. (1993). A framework linking intangible resources and capabilities to sustainable competitive advantage. Strategic Management Journal, 14(8), 607– 618. Helm, S. (2011). Employees’ awareness of their impact on corporate reputation. Journal of Business Research, 64(7), 657–663. Hoffman, D. L., & Fodor, M. (2010). Can you measure the ROI of your social media marketing? MIT Sloan Management Review, 52(1), 41–49. Jones, B., Temperley, J., & Lima, A. (2009). Corporate reputation in the era of Web 2.0: the case of Primark. Journal of Marketing Management, 25(9/10), 927–939. Kaplan, A., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of social media. Business Horizons, 53(1), 59–68. 479 Heikki Karjaluoto, Hanna Mäkinen, Joel Järvinen Keh, H. T. (2009). Corporate reputation and customer behavioral intentions: The roles of trust, identification and commitment. Industrial Marketing Management, 38(7), 732–742. Kietzman, J. H., Hermkens, K., McCarthy, I. P., & Silvestre, B. S. (2011). Social media? Get serious! Understanding the functional building blocks of social media. Business Horizons, 54(3), 241–251. Kirtiş, A. K., & Karahan, F. (2011). To be or not to be in Social Media arena as the most cost-efficient marketing strategy after the global recession. Procedia-Social and Behavioral Sciences, 24, 260–268. Luo X., Homburg C., Wieseke J. (2010). Customer satisfaction, analyst stock recommendations, and firm value. Journal of Marketing Research, 47(6), 1041– 1058. Ryan, D., & Jones, C. (2009). Understanding digital marketing: Marketing strategies for engaging the digital generation. London, UK: Kogan Page Ltd. Schniederjans, D. (2013). Enhancing financial performance with social media: An impression management perspective. Decision Support Systems, 55(4), 911–918. Trainor, K. J. 2012. Relating Social Media Technologies to Performance: A Capabilities-Based Perspective. Journal of Personal Selling & Sales Management; Vol. 32 Issue 3, 317–331. Walsh, G., E. (2007). Customer-based corporate reputation of a service firm: scale development and validation. Journal of the Academy of Marketing Science, 35(1), 127–143. Williams, R. J. (2005). The impact of corporate strategy on a firm’s reputation. Corporate Reputation Review, 8(3), 187–197. Yu, Y., Duan, W., & Cao, Q. (2013). The impact of social and conventional media on firm equity value: A sentiment analysis approach. Decision Support Systems, 55 (4), 919–926. 480 Firm’s Activity in Social Media and Its Relationship with Corporate Reputation and Firm Performance Appendix Reputation index questions Dimensions Overall evaluation about reputation Overall reputation 1 1 Impression about the company (sympathy) Relationship 2 Emotional attraction Trust 3 (Financial) success Competence 4 Rational attraction The quality of the products and services 5 Figure 1: The dimensions of reputation index 481 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Online Dating Sites: A tool for romance scams or a lucrative e-business model? Mohini Singh Margaret Jackson RMIT University, Australia mohini.singh@rmit.edu.au margaret.jackson@rmit.edu.au Abstract Online dating sites are a new lucrative B2C e-business model, however, these sites are increasingly used by felons and scammers to exploit vulnerable customers. Although the impact of online dating scams on victims is vast, and online dating sites are a growing e-business, research on this topic is almost non-existent. The few studies on online dating sites are generally from Psychology addressing user issues. The aim of this research is to explore online dating sites as an e-business model, types of scams carried out via these sites, and regulations required to protect users of online dating sites. Using document analysis, this research will establish the current business models of online dating sites, the types of online dating scams, the impact of these scams on victims, and will develop a typology of these issues for the protection of users and for reducing this new type of cybercrime. Findings of this research will contribute to knowledge on online dating sites as an e-business model which unfortunately is misused by some users for criminal activities to provide future research directions. Key words: online dating sites, online dating scams, online dating e-business model 1 Introduction Due to increased use of technology, dating has fundamentally changed and moved to a new landscape of finding romance online. The advantages of online dating (privacy, wider access to potential partners, anonymity, 24 hours, 7 days access) are some of the reasons a very large number of people from all parts of the world are resorting to online dating (Rege, 2009). Online dating sites facilitate dating by allowing individuals, couples and groups to set up a personal profile where they can detail personal interests, likes, dislikes, physical attributes, and demographic details of what they are seeking from a partner (Whitty & Carr, 2006 and Couch et al, 2012). There is broad interest as well as niche online dating websites such as those based on sexuality, specific sexual Mohini Singh, Margaret Jackson interests, religion, ethnicity, disabilities and pet ownership (Couch et al, 2014). Although a new concept, online dating is now recognised by many as an appropriate way to meet prospective partners (Ellison et al, 2006; Gibbs et al, 2006). At the same time, while engineering romantic encounters, online dating services have also become a new e-business model for generating revenue (Hancock et al 2007). However, although online dating business is relatively new, fraud and criminal activities associated with it appear to have grown significantly (Buchannan and Whitty, 2013). Although online dating is now a billion dollar business with a very large number of users, this successful industry is plagued with cyber crimes such as romance scams and identity fraud. According to the National Consumers League (2008) and BBC (2007) the average victim of online romance scams lost more than $3,000 in 2007. Similarly in Australia, victims of online dating services lost over 25 million dollars in 2013 (Zielinski 2014). The power of the cyberspace and the lightly regulated online dating industry are allowing fraudsters to take advantage of romance scams. To date, research on online dating fraud is generally on the psychological issues of users (Buchan and Whitty, 2013; Valkenburg and Peter, 2007; Toma et al., 2008). Research on online dating sites as a growing e-business model, with the downside as a growth area for scams, is almost non-existent. Therefore the aim of this research is to explore online dating sites as a growing e- business model, identify the types of scams carried out via online dating sites, the impact of these scams on victims, and to develop a typology for identifying romance scammers to assist in the development of policies to protect users of online dating sites as well this lucrative novel e-business model. The rest of this document entails a review of literature on online dating sites as a lucrative e-business model, how these sites encourage scams, extant regulations, and a research methodology to investigate online dating business and associated scams. 2 Literature Review Online dating sites are now a billion dollar business and the industry’s growth rate is remarkable (Visualeconomics.com, 2011). According to Jupiter Research (2006) online dating sites in Europe were a 228 million Euro business in 2006. These sites offer services of access, communication and matching (Finkel et al, 2012). Membership of a site provides a wider access to potential romantic partners one would not get in offline dating. Mathematical algorithms are used to establish compatibility between potential partners and provide matches based on a range of criteria (Valkenburg & Peter, 2007). The use of online dating services is so wide that Mitchell (2009) and Frost et al (2008) estimated that users spend up to twelve hours a week on online dating activity. Other reasons for the wide use of online dating sites include access to potential dates in privacy, convenience, and anonymity (Rege, 2009), without fear of stigma or rejection (Fiore & Donath, 2004). Interactive online dating enables technology based interactions via live chats, instant messaging, flirtatious emoticons, nudges and winks (Wang & Lu, 2007), and online dating sites find compatible matches instantly using mathematical algorithms ( Mitchell, 2009). In 2008 the online dating industry generated revenues of $957 million, and is anticipated to grow at a rate of 10% each year (Rege, 2009). 2.1 Online Dating Site Business Model Online dating services are amongst the latest and perhaps a controversial Internet (online) business operated by entrepreneurs from different jurisdictions. As with most online businesses, initial start-up costs for online dating services are minimal (Smith, 483 Online Dating Sites: A tool for romance scams or a lucrative e-business model 2004), with low barriers to entry, fierce competition and many users due to the convenience of being able to access potential partners without commitment or face to face meetings (Rosenfield & Thomas, 2012). Revenues for online dating business are derived from membership fees and advertising, a subscription fee paid by users, and through advertising. By generating enough traffic on the site from free public access, they sell advertising space and valuable marketing information gathered by the site to advertisers and researchers http://wiki.media- culture.org.au/index.php/Online_Dating_Business_Models. To remain competitive, online dating sites are becoming more specialised and catering for specific markets (sexual orientation, race, cultural background, relationship status and desired interaction) http://thebusinessofdating.wikidot.com/business-model. For customer relationship management Smith (2004) explained that organisations are applying one stop shop, customer tracking analytics and marketing as well as focussing on each customer’s desires and budget. While there are examples of online dating sites such as RSVP.com.au and Match.com, there are also models that are only matchmaking sites such as eHarmony.com and PerfectMatch.com (Schmitz, 2014). Online dating site entrepreneurs do not require a revolutionary business model, they require a sound business plan with operational budget, advertising, target market, legal issues, investment capital, the design and development of the site (Walters, 2014). The above characteristics of online dating sites describe their model to be a B2C online business, where due to wide-spread use of technology, privacy, access to a broad range of potential relationships, the customer base for this online business is the main focus. However, although online dating sites are a lucrative online business and a growth industry, this business model is relatively new and associated with scams and criminal activities. 2.2 Online Dating Scams Types of scams according to Couch et al, (2012) and Buchannan and Whitty (2013) include criminals contacting their victims with fake profiles created with stolen photographs of attractive people to develop relationships with their victims to defraud them of large sums of money. Hancock et al (2007) reported that 86% of online dating participants felt others misrepresented their physical appearance. Couch et al (2012) explain that scammers ask for money to pay for their parents’ funeral payments, costs of passports and visas and other similar expenses. Other characteristics of scammers explained by Buchannan and Whitty (2013) are that fraudsters claim to be in love from an early stage, then move the relationship away from the online dating site to email or Instant Messenger, over periods of weeks, months or years. This communication between the fraudster and the victim is intense and frequent, with the former asking for small gifts at first, and, once the victim complies, larger sums of money are asked for. To make the scam more plausible, and to increase the amounts of money demanded, third parties such as a doctor sometimes gets involved asking for hospital expenses for the criminals. In some cases, victims have been persuaded to travel to countries from where the scam originated, where they either get kidnapped or fall under the influence of the scammers for a second round of scams (Buchannan and Whitty, 2013). The advance fee scam to transfer large sums of money to the victim’s account continues to be popular (Buchannan and Whitty). Besides user created scams on online dating sites, breach of data by the providers of online dating sites also takes place with members’ 484 Mohini Singh, Margaret Jackson details made available to other online parties (Pilgrim, 2014). Online dating scams vary, however, financial loss for victims of online dating scams is huge and the impact is severe (BBC, 2007). In 2010, 592 victims of online dating crime were identified in the UK, of which 203 individuals lost over 5,000 pounds, although it is estimated that this figure is under reported (National Consumers League, 2008). In Australia, a Sydney woman swindled millions from men via an online dating site (http://www.abc.net.au/new - 31/7/14), while a Western Australia woman paid $102,800 to a Nigerian man she met online and travelled to South Africa where she was murdered (http://www.smh.com.au/world/nigerian-police arrest - 31/7/14). In 2013, Australians lost a total of $25 million to dating and romance scams (ABC News Net and Zelinski, 2014). Other online dating risks are deceitful online profiles, unwanted contact, non- consensual behaviour, violence, rape, drinks being spiked, people turning up at residences after a cursory online communication, and being stalked (Couch et al, 2012). Possibilities of becoming emotionally upset, loss of face and rejection (Fiore and Donath, 2004) are other risks associated with online dating. Although online dating services place high priority on privacy and confidentiality issues to earn the trust of their customers, what these sites do not take responsibility for is the personal protection of users (Smith, 2005). Online dating sites have become such a lucrative business that they have little incentive to conduct background checks on their members although in some US states, dating sites are required to provide common sense safety tips by law (New York Times, 2014). There are organisations specialising in backup checks for a fee which is a deterrent for users (New York Times, 2014). 2.3 Online Dating Regulations Generally, regulation of online dating sites for services to consumers is governed by consumer protection legislation. In Australia, “Best Practice Guidelines for Dating Websites” (ACCC, 2012) provides a guide to users and monitors suspicious cross- border money movements to establish if somebody is being scammed (AAP 2014). In the USA, there is no Federal law dealing with the regulation of online dating sites, and in UK, online dating scams have led to the creation of an online dating association to self-regulate the industry (Gross and Acquisti, 2005; Mitchell, 2009). This lucrative online business is large and growing, however, the loose regulation of the industry has left it open to scams which have serious ramifications for users. Research Question: Are online dating sites a lucrative e-business or a tool for romance scams? 3 Research Methodology Since this topic is still under researched, and is of a nature that direct interviews, surveys or observations will not be feasible, this exploratory study will be accomplished using the document analysis method. Danker and Hunter (2006) explain that document analysis helps address ‘why or how an event occurred and whether such an event could happen again’ (p.74). Since cybercrimes and scams which belong to an underground culture (Rege, 2009) those who operate underground and will not agree to participate in a research project. Also, interviewing law enforcement personnel and dating industry 485 Online Dating Sites: A tool for romance scams or a lucrative e-business model representatives is not easy due to the privacy issues and market credibility of dating sites. Document analysis will be undertaken in the following three phases: Phase One: An extensive review of literature (both academic and web publications) will establish keywords for online dating sites as a business model, the size and makeup of the online dating business in Australia, types of online dating scams (social and technical), impact of online dating scams on victims, and current and proposed international consumer protection laws and policies to identify and prosecute online scammers and identity thieves. Phase Two: With the key words established in phase one, search engines will be used to collect data on reported scams, type of scams, frequency of scams, nature of changes in these scams, impact of these scams on victims (financial loss, death, loss of face, other) and consumer protection laws and policies for online romance and scams. Phase Three: From an analysis of the documents in phase two, a typology will be developed with issues on:  online dating business model, operation and policies for accountability;  online dating scams experienced by Australians;  impact of online dating scams on victims; and  an analysis of the above issues for: o a set of guides for identifying online felons posing as potential sweethearts; and o implications for policy and protection requirements for online dating victims. 4 Data Analysis Data collected in phase two will be analysed using the following three steps: All data collected during phase two of the research will be grouped and coded according to Miles & Huberman’s (1991) qualitative data analysis techniques. All words, phrases and descriptions will be coded (labelled) to reflect the issues of online dating sites, scams and crimes via online dating sites, accountabilities and impact on victims. In order to determine frequencies of occurrences, commonalities of themes and extent of scams, content analysis method according to Mostyn (1985) will be used. Mostyn explains that content analysis helps quantify large volumes of open ended material. This will enable a systematic method of converting qualitative data to numerical data wherever appropriate. To establish the issues for the typology, a cognitive analysis of all of the data analysed in step two will be undertaken using the software Cope. Cognitive analysis according to Collis and Hussey (2013) helps promote reflection and analysis of the problem leading to potential solutions. This cognitive analysis should clearly highlight online dating sites as a business model; perils of online dating and types of scams carried out via these sites; characteristics of personalities described on online sites and reported as frauds; number of deceitful identities established from the data; the extent of scam 486 Mohini Singh, Margaret Jackson impact on victims; protection for victims and other related information. This research is in progress, outcomes of which will be available in the near future. References AAP (2014). Online dating scams to be targeted by money laundering agency and ACCC. The Guardian, 2014, Retrieved 31 August 2014 from http://www.theguardian.com/world/2014/aug/11/online-dating-scams-to-be- targeted-by-money-laundering-agency-and-acc. ABC News (2014). Retrieved 31 August 2014 from http://www.abc.net.au/new ABC News (2014). Retrieved 31 August 2014 from http://www.smh.com.au/world/nigerian-police-arrest-man-over-death-of- australian-woman-in-online-dating-scam-20140204-hvb1q.html ACCC (2012), Best Practice Guidelines for Dating Websites. Retrieved 31 August 2014 from https://www.accc.gov.au/system/files/Best%20Practice%20Guidelines%20for%2 0Dating%20Websites.rtf. BBC NEWS (2007, October 4). UK Police in Nigerian scam haul. Buchanan, T, & Whitty, M. T. (2013). The online dating romance scam: causes and consequences of victimhood. Psychology, Crime and Law. doi:10.1080/1068316X.2013.772180. Collis, J. and Hussey, R., (2013) Business Research, fourth edition, New York, PALGRAVE MACMILLAN. Couch, D., Liamputtong, P. & Pitts, M. (2012). What are the real and perceived risks and dangers of online dating? Perspectives from online daters. Health, Risk & Society, 14(7-8), 697-714, DOI: 10.1080/13698575.2012.720964 Danker, M. L. & Hunter, R. D. (2006). Research Methods for Criminology and Criminal Justice – 2nd edition. MA: Jones and Bartlett Publishers. Ellison, N., Heino, R. & Gibbs, J. (2006). Managing Impressions Online: Self- Presentation Processes in the Online Dating Environment. Journal of Computer- Mediated Communication, 11 (2006), 415–44. Fishbein, M., & Ajzen, I. (1974). Attitudes towards objects as predictors of single. Psychological Review, 81(1), 59-74. Finkel, E. J., Eastwick, P. W., Karney, B. R., Reis, H. T., & Sprecher, S. (2012). Online dating: A critical analysis from the perspective of psychological science. Psychological Science in the Public Interest, 13(1), 3–66. Fiore, A. T. & Donath, J. S. (2004). Online Personals: an overview. paper presented at the Conference on Human Factors in Computing Systems, Vienna, April 24-29, 2009. Frost, J. H., Chance, Z., Norton, M. I. & Anley, D. (2008). People are experience goods: Improving online dating with virtual dates. Journal of Interactive Marketing, 22, 51-61. Gibbs, J., Ellison, N., & Heino, R., (2006). Self-presentation in online personals: the role of anticipated future interaction, self-disclosure, and perceived success in internet dating, Communication Research, 33(2), 152–177. Gross, R. & Acquisti, A. (2005). Information Revelation and Privacy in Online Social Networks, WPES’05, November 7, 2005, Alexandria, Virginia, USA ACM. Hancock, J. T., Toma, C. & Ellison, N. (2007). The truth about Lying in Online Dating Profiles, Online Representation of Self CHI 2007 Proceedings, April 28-May 3, 2007, San Jose, CA, USA , 449–452. 487 Online Dating Sites: A tool for romance scams or a lucrative e-business model Jupiter Research (2006). Online Dating in Europe: Strong Growth Continues Across the Region, Retrieved from http://www.jupiterresearch.com/bin/item.pl/research:concept/83/id=97063 Miles, M. B. and Huberman, A. M. (1994). Qualitative Data Analysis. Thousand Oaks, CA:SAGE Mitchell, R. (2009). Online dating: The technology behind attraction, Computer World UK, Retrieved from http://www.computerworlduk.com/in-depth/it- business/2069/online-dating-the-technology-behind-the-attraction/ Mostyn, B. (1985). The Content Analysis of Qualitative Research Data: A Dynamic Approach’, in Brenner, M., Brown, J. and Canter, D. (eds) The Research Interview, Uses and Approaches, London: Academic Press, 115-46. National Consumers League (2008), Retrieved from http://www.nclnet.org/leadership, 18 August 2014. New York Times.Com (2010), Retrieved from www.nytimes.com,19 December 2010. Rege, A. (2009). What’s Love Got to Do with It? Exploring Online Dating Scams and Identity Fraud 2009. International Journal of Cyber Criminology (IJCC), July - December 2009, 3(2), 494–512. Rosenfield, M. & Thomas, R. (2012). Searching for a Mate: The Rise of the Internet as a Social Intermediary. American Sociological Review, 77(4), 523-547. Schmitz, A. (2014). The online dating market: Theoretical and methodological considerations. Economic Sociology The European Electronic Newsletter, 16 (1) 11-24. Smith, A. D. (2005). Exploring online dating and customer relationship management, Online Information Review, 29(1), 18-33. Sprecher, S. (1989). The importance to males and females of physical attractiveness, earning potential and expressiveness in initial attraction. Sex Roles: A Journal of Research, 21, 591-607. Sydney Morning Herald (2014), The Business of Dating. Retrieved from http://thebusinessofdating.wikidot.com/business-model,18 Toma, C. L., Hancock, J. T. and Nicole, B. E. (2008). Separating Fact from Fiction: An Examination of Deceptive Self- Presentation in Online Dating Profiles, Society for Personality and Social Psychology, 34, pp 1023 – 1036. Valkenburg, P. M. and Peter, J. (2007). Who Visits Online Dating Sites? Exploring Some Characteristics of Online Daters. CYBERPSYCHOLOGY & BEHAVIOR, 10(6), 849-852. Visualeconomics.com (2011). A business of love: Online dating by numbers. Retrieved from http://www.visualeconomics.co/a-business-of-love-online-dating-by- numbers. Walters, L. G. (2014). Online dating law – Legal checklist. Retrieved February 4 2015 from http://www.onlinedatinglaw.com/checklist.htm Whitty, M. & Carr, A. (2006). Cyberspace romance: the psychology of online relationships. New York, Pelgrave Mcmillan. Zielinski, C. (2014). High price of love: new warnings to lonely hearts talking to lovers overseas. Canberra Times, Retrieved 11 August 2014, from http://www.canberratimes.com.au/digital-life/consumer-security/high-price-of- love-new-warnings-to-lonely-hearts-talking-to-lovers-overseas-20140810- 102lzq.html 488 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Assessment Schema for Social CRM Tools: An Empirical Investigation Torben Küpper University of St.Gallen, Switzerland torben.kuepper@unisg.ch Alexander Wieneke University of St.Gallen, Switzerland alexander.wieneke@unisg.ch Nicolas Wittkuhn University of St.Gallen, Switzerland nicolas.wittkuhn@student.unisg.ch Tobias Lehmkuhl University of St.Gallen, Switzerland tobias.lehmkuhl@unisg.ch Reinhard Jung University of St.Gallen, Switzerland reinhard.jung@unisg.ch Abstract This paper presents an assessment schema for Social CRM tools based on an empirical investigation. A constraining factor regarding the implementation of Social CRM tools (e.g., Engagor, Demand Media) is a lack of corresponding comparability of the different features (e.g., analysis of individual data, CRM interface). Little research has been conducted on the assessment of Social CRM tools, and even less have used empirical investigations to develop an assessment schema for surveying the use of corresponding technologies. To address this gap, the study reveals a quantitative investigation of Social CRM technology use as well as develops an assessment schema for Social CRM tools (i.e., including a Monitoring and Capturing, Analysis, Exploitation, Communication, IS integration and Management dimension). The data is analyzed using formative indicators with a sample of 122 marketing, communication and IT decision makers. The results of the analysis serve as weights for the assessment schema. It can be used to develop values for Social CRM tools with regard to their different ‘use’ features and dimensions. 489 Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl, Reinhard Jung Keywords: Social CRM tool, Social CRM tool assessment, assessment schema for Social CRM tools 1 Introduction Social Media enables a new mode of communication and interaction between companies and their customers, which changes the existing approach to customer relationship management (CRM) (Baird and Parasnis 2013; Kumar and Reinartz 2012). Within CRM, companies have only one-directional communication (e.g., by e-mail) and gather information on existing customers. Due to multidirectional communication through Social Media, companies have additional access to public and private information (e.g., profiles, activities, interests etc.) of consumers (e.g., followers of a company’s social media account) as well as customers’ friends (Alt and Reinhold 2012). The integration of Social Media into CRM is a rising phenomenon, leading to a new scientific paradigm (Askool and Nakata 2011) and is referred to as Social Customer Relationship Management (Social CRM) (Lehmkuhl and Jung 2013). It is defined as “[…] a philosophy and a business strategy, supported by a technology platform, business rules, processes and social characteristics, designed to engage the customer in a collaborative conversation in order to provide mutually beneficial value in a trusted and transparent business environment” (Greenberg 2010). Gartner has identified Social CRM as one of the top innovation-triggered themes in the next five to seven years (Alvarez 2013). The exploitation of customer information is “expected to positively contribute to the performance outcomes” (Trainor 2012) and possibly enhance the company’s business success. One viable option for companies to achieve and analyze “the customers content on the companies’ Social Media platforms …” (Küpper 2014) is the implementation of tools. Vendors like Lithium, Jive, Salesforce offer various tools (e.g., Hearsay Social, Radian6, Demand Media, Engagor) for Social CRM. However, research and practice have revealed problems in implementing Social CRM tools successfully. One possible reason is that companies are unable to assess these tools, i.e., they cannot match potential features of different tools to the company-specific requirements, and neither science nor practice are able to provide a useful assessment schema. A literature review in 2014 by Küpper et al. (2014), focuses on the current state of knowledge for Social CRM technology features1. Previous works conceptualize individual features of Social CRM technologies (e.g., Alt and Reinhold, 2012; Reinhold and Alt, 2013; Woodcock et al., 2011) or evaluate the use of Social Media (Trainor et al. 2014). Yet, there is a lack of empirical investigation, because no article measures the use of features of a company’s Social CRM tool (e.g., analysis of individual data, CRM interface) with formative indicators, thus hindering the development of a corresponding assessment schema. Given the novelty of the topic, the objective of the present study is to develop an assessment schema for Social CRM tools. The corresponding research questions (RQs) are as follows: RQ 1: Which features are valuable for the investigation of Social CRM technology use? 1 Social CRM technology is a superordinate term for Social CRM tools. An example: talking about Social CRM technology features means every feature of all Social CRM tools. By talking about Social CRM tool features, the authors mean the features of this individual Social CRM tool. 490 Assessment Schema for Social CRM tools RQ 2: How can a Social CRM tool be assessed? To achieve the stated objective, the study reveals (RQ 1) a quantitative investigation for Social CRM technology use and develops (RQ 2) an assessment schema for Social CRM tools. Accordingly, data from a survey sample of 122 marketing, communication and IT decision makers are analyzed through a confirmatory factor analysis, as in Diamantopoulos and Winklhofer (2001). The result shows that 18 features, classified into six dimensions, including Monitoring and Capturing, Analysis, Exploitation, Communication, IS integration and Management are valuable2 for the investigation of Social CRM technologies use. An application of the developed assessment schema is exemplary used for the tool Engagor. Additionally, a comparison of two tools (Engagor and Demand Media) highlights the practical implications of the study (i.e., illustrated on a dashboard application). The remainder of the paper is structured as follows. Section 2 presents the conceptual background and explains the different features of Social CRM technology. Afterwards, the research design is described. Section 4 contains the findings from the evaluation and highlights the assessment schema. The practical implication (i.e., dashboard application) is illustrated in section 5. Finally, the paper concludes, covers the limitations, and outlines further research approaches. 2 Conceptual Background In order to evaluate the use of Social CRM technologies, the conceptual background focuses on previous evaluation of use constructs. It highlights a definition within the Social CRM context and concludes with a list of 18 Social CRM technology features, which serve as the basis for further investigations. Information technology use and information systems (IS) use are widely and vividly discussed topics in the discipline of IS research. For example, Bhattacherjee (2001) and Bhattacherjee et al. (2008) focus on the construct “information technology continuance intention”. Venkatesh et al. (2003) discuss the “user acceptance of IT” including the construct “use behavior”. Additionally, Venkatesh et al. (2008) focus on the construct “system use” (i.e., measured by duration, frequency, and intensity). According to Petter et al. (2007), all recommended constructs are measured with reflective indicators. Due to the specific research topic (Social CRM) and the formative measurement in this study, the CRM and the Social Media literature additionally need to be considered. Within the CRM as well as Social Media context, information technology use is a central component, and also measured by a single reflective construct. An abstract overview of IS, CRM and Social Media literature regarding the use constructs is presented in Table 1. Only Zablah et al. (2012) develop and evaluate formative indicators and corresponding constructs for CRM technology use, which serve as a theoretical framing for the study. CRM technology is understood as the automation of internal (e.g., among employees like Sales-, Marketing people etc.) and external information processing (e.g., communication with consumers through IT such as e-mail, supported by systems for customer analytics). Therefore, CRM technology is defined as “the degree to which firms use supporting information technology to manage customer relationships” (Reinartz, Krafft, and Hoyer 2004). Due to the lack of a Social CRM 2 “Valuable” means that the results are based on a quantitative evaluation (i.e., showing significant coefficients). 491 Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl, Reinhard Jung technology use definition in the literature, the authors of this study adopt a previous definition for CRM within the Social CRM context. Thus, Social CRM technology use is defined as the degree to which Social CRM technology features are being utilized to support organizational work. Level of References Analysis Typ of Construct Investigation of the “Use” Construct Ind. Org. Refl. Form. IS CRM SM Social CRM Bhattacherjee, 2001 x x x Bhattacherjee et al., 2008 x x x Venkatesh et al., 2003 x x x Venkatesh et al., 2008 x x x Jayachandran et al., 2005 x x x Chang et al., 2010 x x x Zablah et al., 2012 x x x Trainor et al., 2014 x x x Abdul-Muhmin, 2012 x x x Rodriguez et al., 2012 x x x Sum 4 6 9 1 4 4 2 0 This study x x x Ind. = Individual; Org. = Organizational; Refl. = Reflective; Form. = Formative; SM = Social Media Table 1: Overview of the literature According to Zablah et al. (2012), a necessary first step in assessing the degree of a company’s Social CRM technology use is to identify corresponding Social CRM technology features. Therefore, a previous explorative qualitative investigation conceptualizes and validates the current literature and consists of two steps (Wang, Sedera, and Tan 2009). First, a literature review was conducted to identify preliminary Social CRM technology features, based on conceptual arguments. Second, a market study revealed the practitioner perspective through an investigation of current tools from different vendors. The analysis of academic publications highlighted 16 Social CRM technology features. The market study (with a total number of 40 investigated vendors) resulted in (1) the validation of 16 identified Social CRM technology features found in the literature and (2) the identification of two additional features. Thus, a total of 18 Social CRM technology features were identified (Küpper et al. 2014). Subsequently, they were categorized into six dimensions. Table 2 presents the previous findings (the dimensions and features) and illustrating examples. 492 Assessment Schema for Social CRM tools Social CRM Social CRM technology Descriptions technology features Examples ID dimensions (sub-dimensions) Identify content through It describes the real time Real time data system keywords CA1 data observation on social monitoring algorithm Monitoring media (e.g., with in- Capturing aggregate About consumers, and memory technologies) and CA2 data competitors etc. Capturing the collection of different About a single social media data (e.g., Capturing individual consumer, a new CA3 with batch processing). data product release, etc. “Analysis” describes the Analysis of content Recognition of AN1 (real time) consumers questions assessment, segmentation Analysis of aggregate Customer analysis, Analysis and/or analysis of the AN2 data brand feedback etc. monitored and captured Analysis of individual social media data. Personal behavior, etc. AN3 data Forecast consumer Predictive modelling EX1 behavior, new trends “Exploitation” describes Interconnected Social Graphs etc. EX2 consumer network map different activities, which Exploitation Advertising campaigns, are executed especially Sales activities EX3 etc. after the analysis phase. Summary statements Reporting on sales, activities, EX4 reports etc. “IS Integration” describes Integration of existing transmission and CRM interface IN1 CRM systems IS integration functions with Integration other information systems Interface with other IS, Information Systems in the company (e.g., other integration of other IN2 interface IT-tools). tools Communication with a Solving a single “Communication” CO1 single consumer consumer issue, etc. Communi- describes different types of Communication with a Newsletter, etc. CO2 cation external (B2C) and internal group of consumers communication. Communication with Cross-functional CO3 employees communication “Management” describes Community Management of social MA1 management media accounts etc. the support and/or coordination of Allocation of Manage- User permission companywide employees’ access MA2 ment management management functions system rights (e.g., moderation, process Applying engagement Engagement management). features like MA3 management gamification etc. Table 2: Dimensions for Social CRM technology use 3 Methodology 3.1 Research Approach The overall research project is conducted in a three-stage multi-method approach and depicted in Figure 1. The research design aims at developing an assessment schema for Social CRM tools. It comprises (1) an explorative qualitative part (see Section 2), (2) a confirmatory quantitative part, and (3) a practical implication part. Accordingly, the paper focuses on the second and third part of the overall research project. First, 493 Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl, Reinhard Jung indicators of Social CRM technology use are developed. Second, the data collection (through a survey) allows the analysis and the validation of the instruments through a confirmatory factor analysis. Next, the assessment schema is developed based on the results of the data analysis. Finally, the assessment schema is applied within a tool, in order to reveal the practical application of the study. Figure 1: Overview of the research approach 3.2 Instrument Development The process of developing instruments (i.e. indicators) is conducted in a three stage approach (I. item creation, II. scale development and III. indicator testing), including six sub-stages in total, as proposed by Moore and Benbasat (1991), which is depicted in Figure 2 (cf. Walther et al., 2013). The first sub-stage “Conceptualization Content Specification” focuses on a literature review in order to identify context-specific constructs (dimensions) and corresponding sub-dimensions (i.e., features, see Table 2). Second, based on the results, items are deduced to operationalize the previous constructs. Third, a Q-sorting procedure assesses the “Access Content Validity” with the calculation of an inter-rater reliability index (or related indexes, e.g., Cronbach’s Alpha). Within the next two sub-stages (“Pretest and Refinement” and “Field Test”), the questionnaire is tested in order to obtain some initial feedback, for instance on problematic areas. Especially for the unique characteristics of formative indicators and the corresponding constructs, the last sub-stage is based on the first four steps of the formative measurement from Cenfetelli and Bassellier (2009). The applied confirmatory factor analysis is designed according to Diamantopoulos and Winklhofer (2001), and focuses on a statistical evaluation of formative indicators and corresponding constructs. The final survey is distributed over several Social Media channels (e.g., Xing, LinkedIn, Twitter), focusing on marketing, communication, and IT decision makers. The indicators are measured using a 7-point Likert scale from the agreement-level “strongly disagree” (1) to “strongly agree” (7). 494 Assessment Schema for Social CRM tools Figure 2: Process of developing instruments 3.3 Development and Practical Application of the Assessment Schema Based on the quantitative analysis (i.e., the confirmatory factor analysis), the estimated values, for each dimension of the Social CRM technology features, serve as the weights for the assessment schema. The practical application with a tool follows in three steps. First, the tool was downloaded and intensively studied. If the tool covers one of the 18 identified and validated Social CRM technology features, it was coded with 1 otherwise it was stated with 0. Second, each feature was quantified, i.e., coding (1 or 0) multiplied with the value of the path coefficient and the corresponding weight. Finally, the sum is taken into account and serves as the assessment of the corresponding tool. 4 Results 4.1 Instrument Development In total, a dataset of 122 answers was captured and serves as the basis for the analysis. Some statistics of the data are presented in Table 3. Industry Per- # of Per- cent Employees cent Position in Company Per- cent Manufacturing & Utility 31.1% < 10 16.4% Executives 31.1% Others 18.0% 10 – 49 17.2% Team Manager 18.9% Information & Communication 14.8% 50 – 499 28.7% Specialized Manager 17.2% Finance & Insurance 13.9% 500 – 999 9.8% Department Manager 15.5% Public Administration & Logistics 11.5% 1000 – 5000 16.4% Division Manager 14.8% Health Industry 10.7% > 5000 11.5% Others 2.5% Table 3: Descriptive sample statistic In order to develop and evaluate formative indicators and the corresponding constructs for Social CRM technology use, the first four steps from Cenfetelli and Bassellier (2009) are applied, which contains a confirmatory factor analysis, according to Diamantopoulos and Winklhofer (2001), as mentioned above. Using the PLS (partial least square) method to analyze the data, SmartPLS and SPSS are the appropriate tools (Hair et al. 2013). The four steps, as recommended by Cenfetelli and Bassellier (2009), include the investigation of: (1) multicollinearity testing, (2) the effect of the number of indicators and non-significant weights, (3) co-occurrence of negative and positive indicator weights, and (4) absolute versus relative indicator contributions. The appendix provides an overview of the test statistics. For the first step (multicollinearity testing), the variance inflation factors (VIFs) are calculated using SPSS. All VIFs are below the maximum threshold of 5.0, recommended by Hair et al. 495 Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl, Reinhard Jung (2011) and Walther et al. (2013). The results reveal that multicollinearity is not an issue in this study. Steps two to four are based on calculated values and test statistics using SmartPLS3. The second step (the effect of the number of indicators and non-significant weights) deals with the problem that a large number of indicators cause non-significant weights. The results show that indicator MA2 ( Management construct) is not significant, which has to be considered in the following steps. Cenfetelli and Bassellier (2009) also state that this should not be misinterpreted concerning any irrelevance of the indicators. The only interpretation of this issue is that some indicators have a lower influence than others. In order to gain a deeper understanding, this study continues with step three (co-occurrence of negative and positive indicators weights). No indicator has negative weights; therefore this is not an issue in the study. Figure 3: Illustrating formative indicators and the corresponding constructs Step four (absolute versus relative indicator contributions) needs to be conducted by reporting the respective loadings. The loadings indicate that an “indicator could have only a small formative impact on the construct (shown by a low weight), but it still 3 With parameter settings using 110 cases and 3000 samples. 496 Assessment Schema for Social CRM tools could be an important part of the construct (shown by a high loading)” (Söllner et al. 2012). Concerning the issues with MA2, which show non-significant, but very high loadings, no further improvements (i.e., dropping indicator) have to be performed (Cenfetelli and Bassellier 2009; Hair, Ringle, and Sarstedt 2011; Hair et al. 2013). To conclude, all formative indicators and corresponding constructs are suitable for evaluating Social CRM technology use. The corresponding path coefficients for Social CRM technology use are illustrated in Figure 3. To answer RQ 1 ( Which features are valuable for the investigation of Social CRM technology use? ), it can be stated that 18 features are valuable for the investigation of Social CRM technology use and serve the basis for developing the assessment schema for Social CRM tools. 4.2 Development of the Assessment Schema The estimated path coefficients and the weights for each indicator are reliable and robust values for the assessment schema. The assessment schema, which is the answer of RQ 2 ( How can a Social CRM tool be assessed? ), is presented in Table 4. The assessment schema has two different dimensions of values (i.e., value of a construct, and value of the indicator weight), which are calculated as follows4. First, the six constructs have to be compared. Therefore, the value for, e.g., Monitoring and Capturing is calculated with 0.163/(0.163 + 0.191 + 0.242 + 0.119 + 0.166 + 0.220) + 1 = 1.146. Second, the value of the indicator weight is constraint to their corresponding construct, e.g., CA1 = 0.132/(0.132 + 0.458 + 0.508) + 1 = 1.12. The non-significant indicator (MA2) is measured with 1. The “coding” column needs to be filled out for a specific tool (see section 4.3). “Quantification” is the product of the three columns and will be calculated as: CA1 = 1.146 x 1.23 x “coding” column. Dimensions Value of the indicator (constructs) Features Value of the construct weights (features) Coding Quantification CA1 1.12 Monitoring and CA2 1.146 1.42 Capturing CA3 1.46 AN1 1.29 Analysis AN2 1.171 1.41 AN3 1.30 EX1 1.37 EX2 1.24 Exploitation 1.217 EX3 1.21 EX4 1.19 IN1 1.57 IS Integration 1.107 IN2 1.43 CO1 1.23 Communication CO2 1.149 1.27 CO3 1.50 MA1 1.42 Management MA2 1.211 1.00 MA3 1.47 Sum (value of the tool) Table 4: Assessment Schema 4 In general, all values are described in percentage and added with 1. 497 Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl, Reinhard Jung 4.3 Practical Application of the Assessment Schema For the practical application the tool Engagor is investigated for three reasons. First, a download version is available, which enables the researcher to work with the tool. Second, a trainee introduces the researchers, in order to learn all of the corresponding features. Third, two cooperate companies are using Engagor for their current Social CRM activities, which capture detailed insights from practice. Table 5 presents the applied assessment schema for Engagor. Dimensions Value of the indicator (constructs) Features Value of the construct weights (features) Coding Quantification CA1 1.12 1 1.28 Monitoring and CA2 1.146 1.42 1 1.62 Capturing CA3 1.46 0 0.00 AN1 1.29 1 1.51 Analysis AN2 1.171 1.41 1 1.65 AN3 1.30 0 0.00 EX1 1.37 0 0.00 EX2 1.24 0 0.00 Exploitation 1.217 EX3 1.21 0 0.00 EX4 1.19 1 1.44 IN1 1.57 1 1.74 IS Integration 1.107 IN2 1.43 0 0.00 CO1 1.23 1 1.41 Communication CO2 1.149 1.27 1 1.46 CO3 1.50 1 1.72 MA1 1.42 0 0.00 Management MA2 1.211 1.00 1 1.21 MA3 1.47 0 0.00 Sum (value of the tool) 15.06 Table 5: Application of the Assessment Schema. 5 Practical Implication Companies can use the assessment schema to compare different tools for their specific needs. To illustrate the affordance and the practicability with another tool, Demand Media is analyzed with the assessment schema. Demand Media achieves a total value of 14.33, which is very similar to the tool Engagor, as calculated before. However, the values of both tools are distributed differently for the features and dimensions, as shown by the dashboard in Figure 4. With the assessment schema a company is able to calculate the value of several tools and illustrate them on a dashboard. It is possible to optimize the number of relevant tools, which have a high value for more than one dimension. This is highly relevant for practice, which can be explained by three practical implications. First, the illustrative dashboard presents an overview of the best value for money. For example, if a company is looking for a tool with monitoring and capturing features, it can compare the dimensional values of each tool and compare the respective licensing costs (e.g., choosing a tool with a lower total value, but avoiding high licensing costs). Second, the dashboard application illustrates the implemented feature allocation. If a company needs a tool covering all dimensions, it would probably choose Engagor over Demand Media, as this tool does not perform well with regard to the IS integration dimension. Finally, the dashboard application is useful for optimizing a toolset, i.e., combining more than one tool to cover ‘weak spots’. 498 Assessment Schema for Social CRM tools Figure 4: Illustrative dashboard application for tools evaluation 6 Conclusion, Limitations and further Research The study develops an assessment schema for Social CRM tools. The quantitative research approach follows the research procedure of Moore and Benbasat (1991) and particularly the first four steps from Cenfetelli and Bassellier (2009). Accordingly, a sample of n=122 responses is investigated and analyzed, surveying marketing, communication and IT decision makers. In order to answer the RQs, the study makes two major contributions. First, the constructs of Monitoring and Capturing, Analysis, Exploitation, Communication, IS integration and Management are valuable dimensions of Social CRM technology use. Second, the assessment schema for Social CRM tools is robust and a useful management vehicle, representing the practical impact on the research results. Two potential limitations constrain the results of this research. First, despite the highly significant values of the final formative indicators (i.e., the statistical test values), there may be missing indicators, which should be included in the model. Second, the study applied only the first four steps of the formative measurement from Cenfetelli and Bassellier (2009), which could have an effect on the results. One promising approach for further research is the use of the assessment schema in practice, in order to find weaknesses and strength. Two possible improvements are stated: First, the ‘coding-values’ of the assessment schema can be described in detail (e.g., instead of 0 and 1, a five point scale is also possible). Second, it could be interesting to add an additional factor (e.g., a prioritization value, which indicates the company’s current needs). A further scientific research approach could be an investigation of a redundancy analysis for the six constructs, in order to identify higher order constructs and/or evaluate the formative indicators with reflective indicators (i.e., benchmark measuring). Therefore, the rigorously and systematically derived results presented by the study form a basis for further research projects. 499 Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl, Reinhard Jung References Abdul-Muhmin, Alhassan G. (2012), “CRM technology use and implementation benefits in an emerging market,” Journal of Database Marketing & Customer Strategy Management, 19 (2), 82–97. Alt, Rainer and Olaf Reinhold (2012), “Social Customer Relationship Management (Social CRM) - Application and Technology,” Business & Information Systems Engineering, 54 (5), 287–291. Alvarez, Gene (2013), “Hype Cycle for E-Commerce 2013,” Gartner, Inc. Askool, Sanaa and Keiichi Nakata (2011), “A conceptual model for acceptance of social CRM systems based on a scoping study,” AI & SOCIETY, 26 (3), 205–220. Baird, Carolyn Heller and Gautam Parasnis (2013), “From social media to Social CRM - What customers want,” IBM Global Business Services. Bhattacherjee, Anol (2001), “Understanding Information Systems Continuance: An Expectation-Confirmation Model,” MIS Quarterly, 25 (3), 351–370. ———, Johan Perols, and Clive Sanford (2008), “Information Technology Continuance: A Theoretic Extension and Empirical Test,” Journal of Computer Information Systems, 49 (1), 17–26. Cenfetelli, Ronald T. and Geneviève Bassellier (2009), “Interpretation of Formative Measurement in Information Systems Research,” MIS Quarterly, 33 (4), 689–707. Chang, Woojung, Jeong Eun Park, and Seoil Chaiy (2010), “How does CRM technology transform into organizational performance? A mediating role of marketing capability,” Journal of Business Research, 63 (8), 849–855. Diamantopoulos, Adamantios and Heidi M. Winklhofer (2001), “Index Construction with Formative Indicators: An Alternative to Scale Development,” Journal of Marketing Research, 38 (2), 269–277. Greenberg, Paul (2010), “The impact of CRM 2.0 on customer insight,” Journal of Business & Industrial Marketing, 25 (6), 410–419. Hair, Joseph F., G. Tomas M. Hult, Christian M. Ringle, and Marko Sarstedt (2013), A Primer on Partial Least Squares Strucutral Equation Modeling (PLS-SEM), Thousand Oaks: SAGE Publications, Inc. ———, Christian M. Ringle, and Marko Sarstedt (2011), “PLS-SEM: Indeed a Silver Bullet,” Journal of Marketing Theory and Practice, 19 (2), 139–152. Jayachandran, Satish, Subhash Sharma, Peter Kaufman, and Pushkala Raman (2005), “The Role of Relational Information Processes and Technology Use in Customer Relationship Management,” Journal of Marketing, 69 (4), 177–192. Kumar, V. and W. Reinartz (2012), Customer Relationship Management, Berlin, Heidelberg: Springer-Verlag. Küpper, Torben (2014), “Measuring the Success of Social CRM - First Approach and Future Research,” in Proceedings of the 16th International Conference on Enterprise Information Systems, Lisbon, Portugal, 573–582. ———, Tobias Lehmkuhl, Reinhard Jung, and Alexander Wieneke (2014), “Features for Social CRM Technology - An Organizational Perspective,” in Proceedings of the 20th Americas Conference on Information Systems, Savannah, USA, 1–10. 500 Assessment Schema for Social CRM tools Lehmkuhl, Tobias and Reinhard Jung (2013), “Towards Social CRM - Scoping the Concept and Guiding Research,” in Proceedings of the 26th Bled eConference, Bled, Slovenia, 190–205. Moore, Gary C. and Izak Benbasat (1991), “Development of an Instrument to Measure the Perceptions of Adopting an Information Technology Innovation,” Information Systems Research, 2 (3), 192–222. Petter, Stacie, Detmar Straub, and Arun Rai (2007), “Specifying Formative Constructs in Information Systems Research,” MIS Quarterly, 31 (4), 623–656. Reinartz, Werner, Manfred Krafft, and Wayne D. Hoyer (2004), “The Customer Relationship Management Process: Its Measurement and Impact on Performance,” Journal of Marketing Research, 41 (3), 293–305. Reinhold, Olaf and Rainer Alt (2013), “How Companies are Implementing Social Customer Relationship Management: Insights From Two Case Studies,” in Proceedings of the 26th Bled eConference, Bled, Slovenia, 206–221. Rodriguez, Michael, Robert M. Peterson, and Vijaykumar Krishnan (2012), “Social Media’s Influence on Business-to-Business Sales Performance,” Journal of Personal Selling and Sales Management, 32 (3), 365–378. Söllner, Matthias, Axel Hoffmann, Holger Hoffmann, Arno Wacker, and Jan Marco Leimeister (2012), “Understanding the Formation of Trust in IT Artefacts,” in Proceedings of the 33rd International Conference on Information Systems, Orlando, USA, 1–18. Trainor, Kevin J. (2012), “Relating Social Media Technologies to Performance: A Capabilities-Based Perspective,” Journal of Personal Selling & Sales Management, 32 (3), 317–331. ———, James Andzulis, Adam Rapp, and Raj Agnihotri (2014), “Social media technology usage and customer relationship performance: A capabilities-based examination of social CRM,” Journal of Business Research, 67 (6), 1201–1208. Venkatesh, Viswanath, Susan A. Brown, Likoebe M. Maruping, and Hillol Bala (2008), “Predicting Different Conceptualizations of System Use: The Competing Roles of Behavioral Intention, Facilitating Conditions, and Behavioral Expectation,” MIS Quarterly, 32 (3), 483–502. ———, Michael G. Morris, Gordon B. Davis, and Fred D. Davis (2003), “User Acceptance of Information Technology: Toward a Unified View,” MIS Quarterly, 27 (3), 425–478. Walther, Sebastian, Darshana Sedera, Saonee Sarker, and Torsten Eymann (2013), “Evaluating Operational Cloud Enterprise Systems Success: An Organizational Perspective,” in Proceedings of the 21st European Conference on Information Systems, Utrecht, Netherlands, 1–12. Wang, Wenjuan, Darshana Sedera, and Felix Tan (2009), “Measuring CRM and SCM Benefits: A Preliminary Measurement Model,” in Proceedings of the 13th Pacific Asia Conference on Information Systems, Hyderabad, India, 1–12. Woodcock, Neil, Nick Broomfield, Geoff Downer, and M. Starkey (2011), “The evolving data architecture of social customer relationship management,” Journal of Direct, Data and Digital Marketing Practice, 12 (3), 249–266. 501 Torben Küpper, Alexander Wieneke, Nicolas Wittkuhn, Tobias Lehmkuhl, Reinhard Jung Zablah, Alex R., Danny N. Bellenger, Detmar W. Straub, and Wesley J. Johnston (2012), “Performance Implications of CRM Technology Use: A Multilevel Field Study of Business Customers and Their Providers in the Telecommunications Industry,” Information Systems Research, 23 (2), 418–435. Appendix Formative Indicators VIF Weights p- value Load The company utilizes a tool to … Monitoring and Capturing CA1** search different type of content (e.g., posts, tweets, etc.) on 1.846 0.132 0.016 0.735 social media platforms in real time. CA2** collect and store unstructured social media information about 2.385 0.458 < 0.01 0.933 the company, product, etc. on their social media platform(s). CA3** collect and store unstructured information about a single 1.540 0.508 < 0.01 0.936 artifact (e.g., consumer, a single event, etc.) on their social media platform(s). Analysis AN1** analyze and assess different types of content in real time. 2.577 0.317 < 0.01 0.914 AN2** analyze unstructured social media data across various criteria 2.299 0.448 < 0.01 0.937 (e.g., consumer segmentation, etc.) in order to identify general trends, profitable consumers, etc. AN3** analyze unstructured data for a single consumer (e.g., a high 2.300 0.323 < 0.01 0.900 potential influencer) across the one (or more) social media platforms in order to understand their social behavior, motivations, etc. Exploitation EX1** forecast consumer behavior, and trends etc. and enhance the 3.519 0.407 < 0.01 0.912 predictive model. EX2* create a network map of consumers and the relationships 3.207 0.266 < 0.01 0.890 between them. EX3** support product purchase, increase sales, cross- and 2.477 0.227 < 0.01 0.877 upselling (e.g., social advertising campaigns). EX4** prepare summary statements, evaluate user activity and their 4.341 0.207 0.032 0.932 loyalty, and/or prepare management reports. IS Integration IN1** integrate the social media data with an existing CRM system. 1.000 0.602 < 0.01 0.964 IN2** integrate other information systems, sales processes and 1.000 0.450 < 0.01 0.934 existing technologies, and other tools along the project lifecycle (exclude a CRM system). Communication CO1** interact personally, one-to-one communication, with a single 1.937 0.273 0.027 0.800 consume. CO2** communicate with an entire community and/or multiple 1.369 0.320 0.022 0.795 consumers. CO3** communicate with other employees throughout the 1.402 0.592 < 0.01 0.891 organization. Management MA1** manage their social media accounts, communities and forums, 2.377 0.454 < 0.01 0.924 such as moderation, internal process management, etc. MA2 allocate employee access rights. 2.104 0.129 0.103 0.834 MA3** apply different engagement features (e.g., gamification etc.). 2.230 0.507 < 0.01 0.933 VIF = Variance Inflation Factor; Load. = Loadings; ** p-value < 0.05; * p-value < 0.10 502 BACK 28th Bled eConference #eWellBeing June 7 - 105, 2015; Bled, Slovenia Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers Lauri Frank Agora Center, University of Jyväskylä, Finland lauri.frank@jyu.fi Markus Salo Department of CS and IS, University of Jyväskylä, Finland markus.salo@jyu.fi Anssi Toivakka Department of CS and IS, University of Jyväskylä, Finland anssi.toivakka@jyu.fi Abstract Nowadays, a huge number of individuals purchase virtual items in constantly growing service environments: online game communities. Some researchers have studied gamers’ motivations to purchase virtual game items in general, but no one has separated different gamer types regarding their purchasing motivations. Understanding different gamer types is important because gamers may purchase the same virtual game items, such as helmets and weapons, for different individual reasons. Given the importance of the topic and the research gap, we introduce a typology of gamers regarding their motivations to purchase game items by conducting an empirical study on actual first-person shooter (FPS) gamers. As a theoretical contribution, our findings reveal three groups of game-item buyers (aesthetes, adventurers, and performers) and one group of non-buyers (critics). Our results indicate that, even in the context of performance-centric FPS games, hedonic motivations are dominant, particularly for the gamer groups that were most likely to purchase game items in the future. Interestingly, we could not find a group of gamers that emphasized merely functional aspects as purchasing motivations. In line with these findings, we present practical implications for game providers to manage and market their selection of game items in more suitable and efficient ways. Keywords: Typology, Gamer, Game items, Purchasing behavior, Motivation 503 Lauri Frank, Markus Salo, Anssi Toivakka 1 Introduction Millions of people purchase virtual game items—such as helmets and weapons—in constantly growing service environments: online game communities. The trade of virtual goods did not exist some years ago, but already in 2012 the global virtual goods market was worth of 12 billion Euros (Superdata, 2012). The dramatic growth of the virtual in-game purchases has increased the need to study gamers’ motivations for purchasing virtual items (Hamari and Lehdonvirta, 2010). Importantly, Wasko et al. (2011, 650) have pointed out the void of knowledge in understanding how to approach gamers and virtual world users to market virtual items because “avatars are a new form of consumer capable of making purchases of both virtual and real world products and services.” A limited set of researchers have investigated gamers’ and virtual world users’ motivations to purchase virtual items (Guo and Barnes, 2011; Ho and Wu, 2012; Kim et al., 2011; Kim et al., 2012; Lehdonvirta, 2005, 2009; Lehdonvirta et al., 2009; Mäntymäki and Salo, 2011, 2013; Park and Lee 2011; Shang et al., 2012). These prior studies have provided valuable insights about the motivational factors affecting purchase intentions in general, but they have not examined the different gamer types at all. Studying different gamer types in these emerging service environments is essential, as different gamers may purchase exactly the same items for different motivations. For example, one gamer could buy a virtual helmet for his/her character to look cool while another gamer’s motivation may be grounded in the helmet’s protection ability against enemies. By understanding these differences, game providers can manage their selection of virtual items to fit the gamers’ needs as well as market virtual items to gamers in more suitable and efficient ways. According to our best knowledge, there are no existing typologies of gamers regarding their motivations to purchase virtual game items. To address this gap in research, we developed a typology of gamers by empirically investigating actual gamers’ motivations to purchase game items in first-person shooter (FPS) games. We specifically wanted to focus on FPS games because they comprise one of the most popular game genres with a tremendous amount of ongoing trade of virtual in-game purchases. Our research question was thus: What kind of gamer types can be found regarding gamers’ motivations to purchase virtual game items? As a theoretical contribution, our typology revealed four gamer types, with each having specific reasons for purchasing (or not purchasing) virtual game items. This new knowledge assists researchers to take a look beyond the rather generic motivation models and specify which motivational factors are relevant for which gamers. As for practical contributions, providers of online gaming communities and other similar virtual service environments can use our results to manage their virtual item offerings, as well as to market and communicate about their offerings efficiently to gamers and users. 2 Literature Review and Conceptual Model 2.1 Virtual Game Items in FPS Games Virtual game item sales constitute a significant revenue share for numerous computer game providers. Recently, many computer games have transferred to the free-to-play model, according to which the game itself may appear to be free but the incomes derive from premium purchases such as virtual game items. The term virtual game item refers to items that can be bought to empower, personalize, and enrich one’s game character or affect virtual identity and 504 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers status in the gaming community. FPS games—action games where the gamer combats enemies through a first-person perspective—are one of the most popular game genres for computers, with regularly charting titles such as Counter Strike, Team Fortress, Call of Duty, and Battlefield game series. Typical virtual items in FPS games include various types of weapons (e.g., guns and grenades), armors and costumes (e.g., helmets and boots), and vehicles (e.g., cars and aircrafts). Most use purposes of FPS game items relate to functional performance and advancement in the game or hedonic enjoyment and customization. Virtual game items can typically be bought from the game producers for a few Euros, but the prices of FPS items may vary from some cents to hundreds of Euros. 2.2 Previous Typologies of General Game Behavior Although there are no typologies regarding gamers’ motivations to purchase virtual game items, some researchers have categorized gamers according to their actions and behaviors. Bartle (1996) has presented a rather widely known typology for gamers, according to which gamers can be divided into achievers, explorers, socializers, and killers. It is important to note that that these types may intertwine with each other (Yee, 2006). For example, in the FPS games, some gamers are called sociable killers, since they mix individual performance with competition against other gamers and social interaction. Hamari and Tuunanen (2014) have presented a useful synthesis of the previous gamer typologies. According to them, the central concepts regarding in-game behavior include achievement, exploration, immersion, sociability, and domination. Achievement relates with individual-oriented gamers and focuses mainly on in-game goals, performance, and power, while exploration and immersion highlight curiosity, story, fantasy, and even escapism. Sociability reflects community-oriented gamers who appreciate social interaction and collaboration. Dominators, in turn, are considered as aggressive gamers who emphasize power. Additionally, they note that gaming intensity, skills, and demographics can be used to differentiate gamers from each other. The previous studies have provided interesting insights about gamers’ general behavior, but they do not touch upon gamers’ purchase behavior at all. Therefore, we review previous studies related to individuals’ motivations to purchase virtual items as follows. 2.3 Review of Studies on Motivations to Purchase Virtual Items We reviewed studies that have examined gamers’ or virtual world users’ motivations to purchase virtual items. We also chose to include the context of virtual worlds, since they include many similar elements with games and researchers have applied similar theories in explaining purchase motivations in both game and virtual world contexts. Contrary to games, there usually are no clear goals in virtual worlds (Mäntymäki and Salo, 2011; Reeves et al., 2008). We located four studies that included the game context and an additional seven studies that focused purely on the virtual world context in their inspection of individuals’ purchase motivations. The reviewed studies are summarized in Table 1. Many of the studies specify three major motivations behind gamers’ or users’ purchase motivations: functional, hedonic (or enjoyment or emotional), and social aspects. Functional aspects refer to the extrinsic and instrumental value of virtual items in improving performance or achieving certain game-specific goals. For example, armor can enhance a game character’s protection against enemies and, thus, improve the character’s chances to complete certain in- game tasks. Previously introduced specific game-related functional attributes include quality 505 Lauri Frank, Markus Salo, Anssi Toivakka (Ho and Wu, 2012), price (Park and Lee, 2011), performance advantage (Lehdonvirta, 2005, 2009), and character competency (Ho and Wu, 2012; Park and Lee, 2011). Quality refers to the gamer’s appreciation of the excellence of a game item (Ho and Wu, 2012), while price value includes the comparison of the item’s cost-effectiveness and its benefits against the monetary sacrifices (Park and Lee, 2011). Performance advantage is valued because of the item’s contribution to better practical performance in, for example, achieving levels and game points (Lehdonvirta, 2009). Character competency is actually a broader concept including not only practical performance advantage but also the game character’s relative power and authority in the game (Ho and Wu, 2012; Park and Lee, 2011). Therefore, we divided character competency into two dimensions: performance advantage and power advantage. Hedonic aspects involve the intrinsic value of virtual items in generating enjoyment and entertainment. For example, a fancy outfit may promote visual appeal or humor. Based on prior studies in the game context, hedonic attributes include visual appeal or visual aesthetics (Ho and Wu, 2012; Lehdonvirta, 2009), sound effects (Lehdonvirta, 2009), playfulness (Ho and Wu, 2012), story, cultural references, and rarity (Lehdonvirta, 2009). Visual appeal covers the enjoyment of the virtual items’ appearance, while audio appeal means similar pleasure is derived from sounds (Lehdonvirta, 2009). Playfulness, in turn, stimulates curiosity and absorption with the game (Ho and Wu, 2012). Story reflects the background fiction or narrative that may create hedonic enjoyment, and cultural references mirror the joy brought about by the real-world or fictive cultural nuances (Lehdonvirta, 2009). Study Game context Virtual world Main results context Ho and Wu Online role-playing - Purchase intentions are driven by 1) (2012) games and war- functional quality, playfulness, and social strategy games relationship support in online role-playing games, and 2) identification with the character, satisfaction with the game, price utility, and playfulness in online war- strategy games. Lehdonvirta EverQuest, Ultima Habbo Hotel User perceptions on real-money trade can (2005) Online, Project involve three dimensions: achievement, Entropia social, and immersion. Lehdonvirta Several online Several virtual Purchase drivers include three attributes: (2009) games worlds functional (performance, functionality), hedonic (visual appearance and sounds, story, provenance, customizability, culture, branding), and social (rarity). Park and Lee Free-to-play online - Character competency, enjoyment, visual (2011) games authority, monetary value, and character identification affect purchase intentions, while satisfaction with the game does not. Guo and - Second Life Functional motivators (effort, performance, Barnes value), hedonic motivators (enjoyment, (2011) advancement, customization), and habit affect purchase behavior. Kim et al. - Cyworld Aesthetics, playfulness, and social self- (2011) image expression influence purchase intentions. 506 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers Kim et al. - Cyworld, Habbo Desire for online self-presentation (2012) Hotel (including self-efficacy, involvement, and norms) and gender affect purchase intentions. Lehdonvirta - Habbo Hotel Virtual items and physical items can share et al. (2009) the same social meanings when it comes to 1) aesthetics, self-expression, and identity, 2) luxury and social status, and 3) items as vehicles of arbitrary meaning. Mäntymäki - Habbo Hotel Purchase intentions are driven by the and Salo presence of other relevant users and use (2011) continuance intentions. Mäntymäki - Habbo Hotel Purchase intentions are driven by network and Salo size, enjoyment, usefulness, availability, (2011) and ease of use. Shang et al. - iPart Social and emotional motivations affect (2012) non-anonymous users’ purchase intentions, but only emotional motivations affect anonymous users’ intentions. Table 1: Studies on gamers’ and virtual world users’ motivations to purchase virtual items Social aspects involve the value of virtual items that reflects the gamer’s social structures with other individuals and within the community (or communities). For example, the possession of a rare treasure item may increase status and respect within the gamer community. Even though at times viewed as a separate aspect from functional and hedonic motivations, it has been argued that social aspects belong to either functional or hedonic motivations (Holbrook, 1996). Therefore, we have placed them under the two main motivations. Prior studies have presented three game-related social attributes: social self-expression (Ho and Wu, 2012; Park and Lee, 2011), social relationship support (Ho and Wu, 2012; Lehdonvirta, 2005), and rarity (Lehdonvirta, 2009). The value of social self-expression may be derived from pure enjoyment (e.g., artistic contributions) or the aim of making a certain impression on others (e.g., status). As these are clearly distinguishable from each other, we have decided to apply both functional and hedonic self-expression (cf. Holbrook, 1996). Some virtual items support social relationships by enhancing communications and maintaining relations (Ho and Wu, 2012). In this study, we have conceptualized these aspects as team play support because the most essential dimension of social relationships in FPS games relates to teamwork and collaboration. Rarity reflects also (partially) social structures, as possessing items that are rare within the game community can produce hedonic value (Lehdonvirta, 2009). 2.4 Conceptual Model The conceptual model of this study is summarized in Figure 1. We chose to form our model based on the previous studies that focused on gamers’ purchase motivations (Ho and Wu, 2012; Lehdonvirta, 2005, 2009; Park and Lee, 2011) because they are the closest ones to the context of this study. Therefore, we aimed to integrate all relevant concepts from those four studies into our research model. We believe that our approach provides the most useful conceptual frame for studying gamers’ motivations to purchase game items for two main reasons. First, our conceptual model includes the central motivational aspects: functional, hedonic, and social (Guo and Barnes, 2011; Ho and Wu, 2012; Kim et al., 2011; Lehdonvirta, 2009). Second, our approach affords a multidimensional frame, which implies an interaction between individual 507 Lauri Frank, Markus Salo, Anssi Toivakka gamers’ motivations and virtual item attributes. Therefore, to tap into the gamers’ context- specific motivations, we amplified the multidimensional frame with rather specific game item attributes that are expected to drive purchase decisions. Functi ona l M oti va ti ons H edoni c M oti va ti ons Quality 1 Visual Appeal 1, 3 Pr ice 4 Ga m e I tem Audio Appeal 3 Per for mance Advantage 1, 2, 3, 4 Pur cha se Playfulness 1 Pow er Advantage 1, 4 I ntenti on Humor Str ategic Planning Stor y 3 Game Balance Cultur al Ref er ences 3 Team Play Suppor t 1, 2 Rar ity 3 Functional Self-Expr ession 1 Socia l M oti va ti ons H edonic Self-Expr ession 1 1: Ho and Wu (2012) 2: Lehdonvirta (2005) 3: Lehdonvirta (2009) 4: Park and Lee (2011) Figure 1: The conceptual model of this study In our conceptual model, the functional motivations are expressed with functional attributes: quality, price, performance advantage, power advantage, team play support, and functional self- expression. The hedonic motivations depend on hedonic attributes: visual appeal, audio appeal, playfulness, story, cultural references, rarity, and hedonic self-expression. All of these attributes are based on prior studies (as described in Section 2.3). Additionally, we decided to add three attributes that we thought were particularly relevant for FPS game items—strategic planning (functional), game balance (functional), and humor (hedonic)—as some game items are designed to facilitate strategic planning or to balance gamers’ different skill levels, and FPS gamers buy game items sometimes just for the sake of humor. There are, of course, some attributes that we chose to either combine with other ones or exclude from our study. For example, we combined customizability (suggested by Lehdonvirta, 2009) with visual appeal, since both of these aspects have been conceptualized similarly in previous studies (as a comparison of Guo and Barnes (2011) and Park and Lee (2011) shows). Additionally, we decided not to include the concept of character identification because it is specific to role-playing games but not so essential for FPS games. 3 Method To examine gamers’ perceptions of their own motivations to purchase virtual game items, we applied a quantitative approach utilizing an online questionnaire and cluster analysis. The rationale for choosing the quantitative approach and the questionnaire was its suitability for studying individual persons’ perceptions, beliefs, and attitudes (Jenkins, 1985; Straub et al., 2004). 3.1 Data Collection We collected the data using an online questionnaire. An introductory text to the questionnaire described with illustrative examples what we meant by the terms FPS games and virtual items in FPS games. The main questionnaire items, as statements, were written to represent the conceptual model of this study; they were adapted and modified mainly from the studies by Ho and Wu (2012), Kim et al. (2011), Lehdonvirta (2009), and Park and Lee (2011). We requested 508 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers the respondents to express their agreement or disagreement on an ordinal five-point Likert scale (from 1 = strongly disagree to 5 = strongly agree) with statements describing different motivations to purchase virtual FPS game items. These 23 statements are presented in Appendix 1. Additionally, the questionnaire included 12 questions about individual gamers’ backgrounds, such as gender, age, primary status, time spent on FPS games, money spent on virtual FPS game items, other reasons for purchases, and future purchase intentions of virtual FPS game items. The data was collected in October 2013. A link to the questionnaire was posted in Valve’s Team Fortress 2 forums, Sony’s PlanetSide 2 forums, and Facebook (for public sharing). Users could respond to the questionnaire anonymously. In total, 98 gamers completed the questionnaire. Before the actual collection, we conducted a small pilot focus group session with three active Team Fortress 2 gamers. This focus group answered and reviewed the questionnaire items by thinking aloud their perceptions and opinions. The pilot session aimed to modify and verify the suitability of the proposed questionnaire items as well as to examine gamers’ willingness to answer. The pilot group provided only a few suggestions for covering the most prominent motivations for purchasing game items: Based on the feedback, one item was deleted and two items focusing on team play support and strategic planning were added. Generally, the pilot respondents were able to complete the questionnaire rather easily. 3.2 Data Analysis The data was analyzed using the SPSS software. To evaluate how well the questionnaire items measured functional and hedonic motivations, we calculated Cronbach’s alphas. As the values for both functional and hedonic motivations exceeded 0.8, the reliability of the item measurements could be considered satisfactory (Nunnally and Bernstein, 1994). To identify distinct gamer groups, the responses were submitted to a cluster analysis. Cluster analysis is used to identify homogeneous groups, when the number of groups or group membership for the cases is unknown. One of the typical ideas of clustering is to minimize within-group variation and maximize between-group variation (Vassilikopoulou et al., 2005). In a number of studies in different disciplines, such analysis has been found useful in developing typologies of individuals. One important use of clustering is to identify different groups of buyers’ regarding their behavioral characteristics (Punj and Stewart, 1983). In this study, clustering aimed to divide or segment gamers into relevant homogeneous groups based on the gamers’ ratings on the statements regarding motivations to purchase virtual game items. We applied Ward’s hierarchical method for cluster formation and Euclidean distance for distance measurement. The analysis resulted in four different clusters of gamers (cluster sizes: C1=32; C2=35; C3=25; C4=4). We chose to distinguish four clusters by considering previous studies and following the pattern of the clustering process. The resulting four clusters were then interpreted and prepared for reporting the results based on the between-group differences in the mathematical means of the measured items. Finally, we compared the potential differences in gamer background (intention to purchase game items in the future, age, and primary status) among the different clusters. We estimated the prospective statistically significant differences between the distributions in different clusters by applying cross-tabulations with Pearson’s chi-squares. Overall, the summary of our research process is illustrated in Table 2. Stage Description Development of the We developed the model based on previous studies that focused on 509 Lauri Frank, Markus Salo, Anssi Toivakka conceptual model individuals’ motivations to purchase virtual game items. The questionnaire items were adapted and modified mainly from Formulation of the previous studies (with a few additions related to the FPS game questionnaire items context). We piloted the questionnaire with a small focus group to fine-tune the Pilot: Focus group wordings, ensure the coverage of the motivational attributes, and find out gamers’ willingness to participate. The questionnaire link was submitted to different forums relevant for Online questionnaire FPS gamers. We applied cluster analysis to identify different gamer groups Cluster analysis regarding their motivations to purchase virtual game items. We used cross-tabulations to examine whether there were statistically Cross-tabulations significant differences among the gamer groups related to the gamers’ intention to purchase, age, and/or primary status. Table 2: Summary of the research process: The main stages and their descriptions 3.3 Respondents The background information of the respondents is summarized in Table 3. On average, the respondents estimated that they had spent 57 Euros for virtual goods in FPS games within the last six months. As expected, the reported amounts varied a lot: from 0 Euros to 600 Euros. In our sample, the majority of the respondents were male (96.9%), students (67.7%), and 30 years old or under (75.5%). These distributions can be considered to reflect FPS gamers because online games related to weapons and war typically attract young males. Similarly, many previous studies on online games have had male-centric samples, and it has been stated that the majority of heavy gamers are young men (Kirriemur and McFarlane, 2004, according to Park and Lee, 2011). The participants named fifteen different FPS games as their favorite and ten different games as the FPS games they based their responses on. For the latter, the most frequently mentioned games were Team Fortress 2 (53) and Planet Side 2 (22). Altogether, the gamers who opened the questionnaire web link were from 22 countries: mostly from Finland and the United States, but also from Canada, Sweden, Estonia, Denmark, Germany, and Brazil. Male Gender 95 (96.9 %) Female 3 (3.1 %) Under 20 34 (34.7 %) 20–30 Age 40 (40.8 %) Over 30 20 (20.4 %) N/A 4 (4.1 %) Work 40 (41.7 %) Primary status Student 65 (67.7 %) Unemployed 7 (7.1 %) Average time used for playing games per week 22.6 hours Average time used for playing FPS games per week 13.5 hours Average money spent for virtual game items within 6 months 65 € (ranging from 0-600 €) Average money spent for virtual FPS game items in past 6 months 57 € (ranging from 0–600 €) Table 3: Background information of the respondents 510 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers 4 Results Based on the cluster analysis, we identified four distinct clusters as gamer groups: three groups of buyers and one group of non-buyers. The average means of the gamer groups for each questionnaire item are illustrated in Figure 2. The four groups are first labeled and described and then compared regarding the gamer groups’ background information. 4.1 Group I: Aesthetes Group I involved gamers who strongly valued specific hedonic motivations: visual appeal, humor, and hedonic self-expression. In particular, the respondents wanted to purchase items that made their game character look better. As a contrast, the respondents rated most functional motivations very low (expect for item quality, which was rated rather highly among all buyer groups). For example, they did not value game items for their prospective effects in performance advantage, power advantage, strategic planning, or team play support at all. Consistent with these findings, we labeled this group of gamers as aesthetes. The group accounted for 32 respondents who were mainly students (69%), a lot of them less than 20 years of age (50%). 4.2 Group II: Adventurers Group II had some similar characteristics with the first group: this group contained gamers who valued visual and audio appeal, playfulness, humor, and hedonic self-expression. However, as the main difference compared to the first group, they reported an average agreement with many functional motivation attributes. Overall, this group highlighted hedonic motivations but did not downplay the functional motivations. Therefore, this group was named adventurers. The group included 35 respondents who were mainly students (80%), a lot of them less than 20 years of age (46%). 4.3 Group III: Performers Group III contained gamers who especially valued those motivational attributes that were related to performance and power advantage. These gamers reported only an average agreement with several hedonic motivation statements but, interestingly, there were no particularly low ratings for any motivational aspect. Compared to adventurers, this group valued more functional and less hedonic motivations. According to these findings, we labeled this group as performers. This group accounted for 25 respondents. The majority (60%) of these respondents were young adults between 20 and 30 years of age. Among them there were almost equal numbers of students (52%) and those in working life (48%). 4.4 Group IV: Critics Group IV included very different gamers containing only four respondents. These gamers were non-buyers and strongly disagreed with all reasons for game item purchases. The average ratings for all motivational statements were extreme low (equal to or less than 2). None of these respondents planned to purchase virtual items within the next six months. According to these insights, we labeled this group as critics. Two of the critics were over 30, while two were less than 30 years of age. Both students and workers were included. 511 Lauri Frank, Markus Salo, Anssi Toivakka Figure 2: Comparison of the gamer groups 4.5 Background Information of the Groups There are two statistically significant differences in the background variables related to the groups. First, the intention to purchase game items within the next six months significantly differed among the gamer groups according to our chi-square tests. As presented in Table 4, the majority of the aesthetes (67.7%) and adventurers (68.6%) reported that they were likely to purchase game items in the near future. As for the two other groups, a smaller share of the performers (40%) and none of the critics (0%) intended to buy game items within the next six months. Group Purchase intention in the next 6 months: Not likely Likely Total I Aesthetes % within group 32.3 % 67.7 % 100.0 % II Adventurers % within group 31.4 % 68.6 % 100.0 % III Performers % within group 60.0 % 40.0 % 100.0 % IV Critics % within group 100.0 % 0.0 % 100.0 % Table 4: Cross-tabulation: Purchase intention in the next 6 months and gamer groups Second, the cross­tabulation (Table 5) and chi­square tests indicated that age groups differed significantly between the gamer groups: gamers under 20 years of age formed the largest group of aesthetes and adventurers, whereas the majority of performers and critics were older than 20 512 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers years. We also investigated the differences of the respondents’ primary status (student, unemployed or employed), but found no statistically significant differences. Group Under 20 20-30 Over 30 Total I Aesthetes % within group 50.0 % 40.6 % 9.4 % 100.0 % II Adventurers % within group 45.5 % 33.3 % 21.2 % 100.0 % III Performers % within group 8.0 % 60.0 % 32.0 % 100.0 % IV Critics % within group 25.0 % 25.0 % 50.0 % 100.0 % Table 5: Cross-tabulation: Age and gamer groups 5 Discussion This article contributes to existing knowledge by presenting a new typology of gamers according to their motivations to purchase virtual game items. Previous studies have reported empirical investigations about the main motivations for virtual item purchases among gamers in general, but they have not taken a stand on the prospective individual differences of purchase motivations. Therefore, our typology assists researchers to understand different gamer groups and providers of games and similar virtual service environments to communicate and market virtual items in more suitable ways. 5.1 Theoretical Contribution: A Typology of Gamers In the empirical part of our study, we found three distinct groups of game-item buyers and one group of non-buyers. Based on these findings, we developed a typology of gamers that is illustrated in Figure 3. Even though the extant gamer typologies do not examine any purchase motivations, we used them to compare and contrast our typology as follows. 513 Lauri Frank, Markus Salo, Anssi Toivakka High I Aesthetes I I Adventurers nder 30 years U Students Under 30 year s ers Likely to buy Students Likely to buy or kers w I I I Perform ver 20 years O hat likely to buy Students & ew H edoni c Som m otivations I V Critics We did not find a gr oup w ith high functional and low Not likely to buy hedonic motivations. Low Low Functional High m otivat ions Figure 3: Our typology of gamers regarding their purchase motivations Interestingly, we could not find a group of game-item buyers that would emphasize merely functional motivations and, simultaneously, downplay hedonic motivations (positioned in the lower right-hand corner of Figure 3). Our findings depart from prior knowledge, since the extant gamer typologies have identified high functionality-oriented gamer groups labeled as dominators or killers (Bartle, 1996; Hamari and Tuunanen, 2014). We expected such a group to exist also regarding gamers’ purchase behavior, especially in the context of fast-paced and performance-centric FPS games. Additionally, in contrast to the previous typologies, we did not find strong social motivations for game item purchases. Even though some gamers are socializers and their general gaming behavior is motivated by socializing (Bartle, 1996; Hamari and Tuunanen, 2014; Yee, 2006), it seems that such a motivation does not currently reach purchasing behavior, at least in FPS games. We consider this finding somewhat paradoxical because some FPS games accentuate social aspects and provide gamers with various game items as social tools to facilitate communication and teamwork. The first group of our typology, aesthetes, is positively oriented toward game item purchases. As these gamers highlighted hedonic aspects and disregarded functional aspects, they shared some similarities with gamers who play games to reach immersion by, for example, escapism or getting absorbed in the game (Bartle, 1996; Hamari and Tuunanen, 2014). Thus, our findings extend this prior knowledge about highly hedonic-centric gamers to the context of in-game purchases. The second group of gamers, adventurers, appreciated various hedonic attributes in game items, but also saw some potential motivational boosts from functional attributes. Adventurers also reported a high likelihood of purchasing game items in the future. This group could be 514 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers interpreted to reflect exploration, which on the one hand concentrates on appeal, curiosity, and playing around, but on the other hand may involve some interests related to rationality and problem-solving (Hamari and Tuunanen, 2014). The third group, performers, was motivated by functional aspects, especially performance and power, with a lesser focus on hedonic aspects. Performers resemble achievement-centric gamers (Bartle, 1996; Hamari and Tuunanen, 2014; Yee, 2006), who focus specifically on in-game goals and advancement in the game. On average, they reported to be only somewhat likely to purchase game items in the near future. When they do, it seems that they purchase game items mainly to perform better, but they also appreciate the additional playfulness and visual enjoyment that the items might bring. Even though these motivations are partly in line with the group referred to as dominators or killers (Bartle, 1996; Hamari and Tuunanen, 2014), such a group would probably use game items just as tools to do damage to others (i.e., for purely functional motivations). Finally, our typology presented an important, yet previously unmentioned gamer group: critics. This gamer group is radically different from the others: even though critics might enjoy playing the actual game, they basically disagreed with any motivations to purchase virtual game items. It seems that they would not even like to have the option to purchase game items. There may be a variety of specific reasons behind such critical behavior; some gamers oppose game item purchases because they consider it to be harmful in preserving the games’ “magic circle” (Castronova, 2004, 192) or perceive it as cheating (Lehdonvirta, 2005). 5.2 Practical Implications There are at least four implications for the providers of games and similar virtual service environments. First, game providers could take advantage of the resulting typology by customizing their game item offerings according to the gamer types. Currently, many game providers already sell game items for different purposes (e.g., for performance boost or aesthetic appeal), but providers could take even further steps to offer gamers what they really wish to purchase. Second, many game providers and designers seem to assume that players are likely to spend money on virtual items that raise their performance quickly and increase their power in the game (Fields and Cotton, 2012; Lehdonvirta, 2009; Oh and Ryu, 2007). However, we could not find support for these assumptions. In contrast, we found that hedonic aspects motivated the gamers that were most likely to purchase game items. Therefore, we suggest game providers carefully revisit their potential assumptions on functional motivations. Third, our findings indicate that hedonic motivations are highly essential for game item purchases—especially visual appeal, humor, playfulness, and hedonic self-expression. Previously, aesthetic items have been assumed to be essential mostly in rather visually-oriented virtual worlds such as Habbo Hotel, where users can buy decorative furniture or cute pets (Lehdonvirta, 2009; Kim et al., 2011). Naturally, one would expect that individuals’ perceptions of visual appeal regarding cheerful virtual worlds are different from those regarding quite harsh game environments, such as FPS games. However, our findings contradicted this assumption and, thus, may help FPS game providers to promote certain hedonic aspects suitable for FPS games. Fourth, there is a group of gamers who quite radically critique the current system of game item sales. Even though this group seems to be extremely difficult to convert into game-item buyers, 515 Lauri Frank, Markus Salo, Anssi Toivakka at least game providers and designers could acknowledge these gamers and try to reduce the amount of their negative associations regarding purchasing game items. 5.3 Limitations and Future Topics There are certain limitations regarding this study. First, our sample size could have been larger. However, our sample was sufficient enough for our research task to identify different gamer groups. Second, our sample consisted mainly of young males. Even though young men are currently the dominant user group for FPS games, it would be important to study other demographic groups that prospectively play FPS and other games in the future. Third, we focused merely on the context of FPS games. Our focus on a certain game genre could have affected our results—for example, the fast-paced nature of FPS games could emphasize some motivations more than others. In the future, we encourage researchers to examine whether our findings are applicable to other virtual service environments than just games. For example, it would be interesting to compare our typology of gamers against similar typologies of virtual world item buyers. Also, as this study focused on computer games, it would be tempting to examine whether the device makes any difference to gamers’ purchase motivations. Thus, future studies could focus on gamers’ motivations to conduct mobile in-app and in-game purchases. Finally, it would be worthwhile to dive deep into the perceptions and motivations of the group labeled critics. Researchers could explore the reasoning behind critics’ negative attitudes toward in-game purchases with qualitative methods such as laddering interviews. 516 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers References Bartle, R. (1996). Hearts, Clubs, Diamonds, Spades: Players Who Suit Muds. Journal of Virtual Environments 1 (1). Castronova, E. (2004). The Right to Play. New York Law School Law Review 49 (1), 185-210. Guo, Y. & Barnes, S. (2011). Purchase behavior in virtual worlds: An empirical investigation in Second Life. Information & Management 48 (7), 303-312. Fields, T. & Cotton, B. (2012.) Virtual Goods - An Excerpt from Social Game Design: Monetization Methods and Mechanics. URL: http://www.gamasutra.com/view/feature/135067/ Hamari, J. & Lehdonvirta, V. (2010). Game Design as Marketing: How Game Mechanics Create Demand for Virtual Goods. Journal of Business Science and Applied Management 5 (1), 14-19. Hamari, J. & Tuunanen, J. (2014). Player Types: A Meta-synthesis. Transactions of the Digital Games Research Association 1 (2), 29-53. Ho, C-H. & Wu, T-Y. (2012). Factors Affecting Intent to Purchase Virtual Goods in Online Games. International Journal of Electronic Business Management 10 (3), 204-212. Holbrook, M. B. (1996). Customer Value – A framework for Analysis and Research. Advances in Consumer Research 23 (1), 138-142. Jenkins, A. M. (1985). Research Methodologies and MIS Research. In: Research Methods in Information Systems, Ed. by E. Mumford et al., Holland: Elsevier Science Publishers B.V., 103-117. Kim, H-W., Chan, H. C. & Kankanhalli, A. (2012). What Motivates People to Purchase Digital Items on Virtual Community Websites? The Desire for Online Self-Presentation. Information Systems Research, articles in advance, 1-14. Kim, H-W., Gupta, S. & Koh, J. (2011). Investigating the Intention to Purchase Digital Items in Social Networking Communities: A Customer Value Perspective. Information & Management 48 (6), 228-234. Lehdonvirta, V. (2005). Real-Money Trade of Virtual Assets: Ten Different User Perceptions. In: Proceedings of Digital Arts and Culture (DAC 2005). 52-58. Lehdonvirta, V. (2009). Virtual Item Sales as a Revenue Model: Identifying Attributes that Drive Purchase Decisions. Electronic Commerce Research 9 (1-2), 97-113. Lehdonvirta, V., Wilska, T-A. & Johnson, M. (2009). Virtual Consumerism: Case Habbo Hotel. Information, Communication & Society 12 (7), 1059-1079. Mäntymäki, M. & Salo, J. (2011). Teenagers in Social Virtual Worlds: Continuous Use and Purchasing Behavior in Habbo Hotel. Computers in Human Behavior 27 (6), 2088-2097. Mäntymäki, M. & Salo, J. (2013). Purchasing Behavior in Social Virtual Worlds: An Examination of Habbo Hotel. International Journal of Information Management 33 (2), 282-290. Nunnaly, J. C. & Bernstein, I. H. (1994). Psychometric Theory. New York: McGraw-Hill. 517 Lauri Frank, Markus Salo, Anssi Toivakka Oh, G. & Ruy, T. (2007). Game Design on Item-selling Based Payment Model in Korean Online Games. In: Proceedings of Digital Games Research Association Conference (DiGRA), 650-657. Park, B-W. & Lee, K. C. (2011). Exploring the Value of Purchasing Online Game Items. Computers in Human Behavior 27 (6), 2178-2185. Punj, G. & Stewart, D. W. (1983). Cluster Analysis in Marketing Research: Review and Suggestions for Application. Journal of Marketing Research 20 (2), 134-148. Reeves, B., Malone, T. W. & O’Driscoll, T. (2008). Leadership’s Online Labs. Harvard Business Review 86 (5), 58-66. Shang, R-A., Chen, Y-C. & Huang, S-C. (2012). A Private Versus a Public Space: Anonymity and Buying Decorative Symbolic Goods for Avatars in a Virtual World. Computers in Human Behavior 28 (6), 2227-2235. Straub, D., Gefen, D. & Boudreau, M-C. (2004). The IS World Quantitative, Positivist Research Methods Website. Ed. by D. Galletta. URL: http://dstraub.cis.gsu.edu/quant/ (visited on 26th Nov 2014) Superdata (2012). Worldwide virtual goods market reaches $15 billion. Monetization still a four letter word. URL: http://www.superdataresearch.com/blog/monetization-is-a-four-letter- word/ (visited on 12th Nov 2014) Vassilikopoulou, A. I., Siomkos, G. J. & Mylonakis, J. (2005). Clustering Consumers According to Their Attitudes on Corporate Social Responsibility. International Journal of Business Governance and Ethics, 1 (4), 317-328. Wasko, M., Teigland, R., Leidner, D. E. & Jarvenpaa, S. (2011). Stepping into the Internet: New Ventures in Virtual Worlds. MIS Quarterly 35 (3), 645-652. Yee, N. (2006). Motivations of Play in Online Games. Cyber Psychology & Behavior 9 (6), 772-775. 518 Why Buy Virtual Helmets and Weapons? Introducing a Typology of Gamers Appendix 1: Questionnaire Main Attribute Questionnaire Item Motivation Question: Do you agree or disagree with the following reasons for buying virtual goods in FPS games? (Five-point Likert scale, from 1 = strongly disagree to 5 = strongly agree.) They function reliably. Functional Quality They are high quality. They are reasonably priced. Price They have good value for the money. They raise my performance quickly. Performance They help my team to win. Character competency They increase my power in the game. Strategic planning They facilitate strategic planning in the game. Game balance They help to keep game balance. They help to work as a team. (Social) Team play support They provide effective communication tools for the game. (Social) Functional self-expression They make me respected by other players. They are aesthetically appealing. Hedonic Visual appeal They make my character look better. Sound effects They have enjoyable sound effects. They make the game more exciting. Playfulness They stimulate my curiosity. They increase immersion in the game. Humor They add humor to the game. Story They fit well with the game lore. Cultural reference They can add cultural nuances to the game. (Social) Rarity They are rare. (Social) Hedonic self-expression They make my character look cooler for others. Table 6: Online questionnaire statements 519 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Designing Tablet Banking Apps for High-Net-Worth Individuals: Specifying Customer Requirements with Prototyping Christian Ruf University of St.Gallen Institute of Information Management, Switzerland christian.ruf@unisg.ch Andrea Back University of St.Gallen Institute of Information Management, Switzerland christian.ruf@unisg.ch Henk Andreas Weidenfeld Laboremus, Norway henk@laboremus.no Abstract Private banks with high-net-worth customers see a great potential in mobile information technology to provide more transparency in the advisory process. Previous literature has mainly focused on gathering requirements with regard to mobile banking applications targeted for retail customers or with regard to advisory services in physical proximity. This paper focuses on an mFAS which is designed for the private banking customer segment and facilitates location- independent customer relationships on a tablet. Furthermore, we specify previously established requirements with the Requirements Abstraction Model. In this study, we evaluated the requirements with a focus group involving seven domain experts. The results of this workshop suggest that most of the specified requirements meet the recommended practice for requirements specification. However, the experts only partly agreed that the presented requirements meet the completeness criterion, which guides future research endeavors. Keywords: Requirements Engineering, Tablet Banking, Mobile App, Prototyping 520 Christian Ruf, Andrea Back, Henk Andreas Weidenfeld 1 Introduction In Switzerland, 12.7% or 435,000 of households possess wealth exceeding CHF 1 million. During the recent financial crisis many such high-net-worth individuals (HNWI) lost faith in financial institutions and in their relationship managers (RM) (Gemes, Ammann, & Lenzhofer, 2010). Consequently, HNWIs are demanding more transparency and simplicity (Oehler & Kohlert, 2009). Financial institutions are taking various countermeasures in order to address these customers’ concerns. Both practitioners (KPMG, 2013; PwC, 2013) and researchers (Inbar Noam, 2012; Nussbaumer, Matter, & Schwabe, 2012) believe that information technology (IT) is one of the measures that may facilitate more transparent financial advisory services. Consequently, introducing a mobile application (app) in financial advisory services might be a first step in this direction. However, in order to develop such mobile apps, recent articles have primarily focused on gathering the requirements of retail customers (Yousafzai, Pallister, & Foxall, 2003), or on advisory processes in physical proximity (Nussbaumer et al., 2012). This paper focuses on mobile apps in location-independent situations addressing the needs of the HNWI segment. In order to develop successful mobile apps, or software artifacts in general (Aurum & Wohlin, 2005), the literature acknowledges that the requirements engineering (RE) process, which involves the elicitation and management of requirements for designing software, is a prerequisite (Vijayasarathy & Turk, 2008). Accordingly, successful endeavors allocate a significantly higher amount (28 percent) of resources to RE (Hofmann & Lehner, 2001). The Requirements Abstraction Model (RAM) from (Gorschek & Wohlin, 2006) introduces an integrated approach for specifying customer requirements (CR) which should address these challenges in RE. Thus, the goal of this paper is twofold and incorporates both theoretical as well as practical contributions. First, we specify CR for a mobile app targeted for private banking customer segments with the RAM. Second, by developing a prototype according to the specified requirements, we pursue an iterative evaluation and present the findings in three focus groups. The final focus group, involving seven experts, validates whether the requirements meet the IEEE recommended practice for requirements specification (IEEE, 1998). The following research question illustrates our goal: What are specified customer requirements (CR) for a mobile app that meet the quality criteria of the recommended practice for requirements specification? We structure the remainder of this paper as follows: First, we elaborate how mobile apps facilitate financial advisory services in Section 2. Furthermore, we also discuss a theoretical foundation regarding RE and previously elicited requirements with regard to a mobile app for HNWI. Second, following the theoretical discussion in Sections 2, we introduce the research design, chosen design science research (DSR) approach, and the method in Section 3. Third, we present the results of our iterative evaluation with 3 focus groups in Section 4 and subsequently discuss the findings in Section 5. Finally, Section 6 provides limitations, conclusions and outlook for future studies. 521 Designing Tablet Banking Apps for High Net Worth Individuals 2 Related Work 2.1 Mobile Financial Advisory Service (mFAS) When speaking of mobile financial advisory services (mFAS), we refer to the interactions between relationship managers (RM) and high-net-worth individuals (HNWI) who possess investable assets exceeding $1 million. According to the ISO standard (ISO, 2011) a financial advisory service consists of various process steps. In this study, we specifically focus on the monitoring and reviewing of the financial plan. Within these process steps, considering the recent technological advances, mobile applications (apps) provide viable alternatives to email or phone calls, e.g. access to RMs or personal financial information on the tablet from anywhere at any time. Despite the acknowledged relevance of such an mFAS for the HNWI segment (KPMG, 2013; PwC, 2013), the literature so far has only captured requirements for the retail banking customer segment (Yousafzai et al., 2003) or for advisory services in physical proximity (Nussbaumer, Matter, & Schwabe, 2012). Hence, this study aims at addressing this gap and specifies requirements for an mFAS for HNWI specifically. 2.2 Requirements Specification with the Requirements Abstraction Model (RAM) and with Prototyping Requirements engineering (RE) captures complete and correct needs of various stakeholders and consequently to facilitate documentation of these needs (Byrd, Cossick, & Zmud, 1992). In order to develop mobile apps successfully, the RE poses a critical prerequisite. Hence, failing to apply a comprehensive RE may lead to project failures or costly change requests later throughout the project execution phase (Pohl, 2008). In order to manage successful RE, Gorschek and Wohlin (2006) introduced the Requirements Abstraction Model (RAM), an approach for specifying requirements. However, despite preliminary evaluations, they propose that researchers and practitioners should further instantiate and validate the usefulness of the proposed RAM (Gorschek et al., 2007). We aim at specifying customer requirements with the RAM on the Feature Level to the Function Level and consequently provide a theoretical contribution. This model contains 4 Abstraction Levels (Gorschek & Wohlin, 2006). Goal Level. The Goal Level consists of general requirements which refer to the value creation process of an organization meeting the demand of customers. Due to the generic characteristic it is questionable whether the Goal Level actually composes actual requirements, but rather general guidelines. Feature Level. The Feature Level consists of general characteristics. Such characteristics include technical functionality and behavior, tangible or intangible outcomes, design elements of the process and resources requirements of the service provider. Function Level. Functions refer to specific characteristics. Compared to the Feature Level, such characteristics should be more specific and precise. 522 Christian Ruf, Andrea Back, Henk Andreas Weidenfeld Component Level. This level relates to information how the developers should actually implement the requirements from the Function Level. In this study, we did not specify the requirements on this level, as we did not implement our artifact in a real-life context. The scope of this study is to specify requirements on the first four levels of the RAM. 2.3 Customer Requirements (CR) for Mobile Financial Advisory Service (mFAS) A previous study (Ruf, Back, Bergmann, & Schlegel, 2014) elicited and prioritized customer requirements (CR) on the Feature Level for an mFAS. A multi-method approach was followed, including a literature review, expert interviews and focus groups. Overall, the stakeholders included in the study were the following: Project Sponsor, Senior Consultant, Social Media Manager, Investment Advisor, Relationship Manager, HNW customer, Independent Investment Advisor, and Director. Based on the feedback from the practitioners, as well as the desk research, the following requirements were identified. (CR1) Access to experts. As a Feature Level requirement, customers should not only be able to contact personal RMs, but also financial experts and investment advisory teams. The mFAS, therefore, should provide such a network in the mobile app. (CR2) Information quality. Regarding information quality, the previous findings suggest that customers are already well-informed and demand aggregated and personalized information. Furthermore, the information provided on the platform should be timely and available at the fingertips. (CR3) Proactivity. As a next requirement, customers expect RMs to inform them proactively about new financial trends and topics, as well as events which are relevant for them. Hence, the mFAS should facilitate this information exchange between customers and RMs in a proactive way. (CR4) Situational use and social presence. Furthermore, mFAS should enable a more effective and personalized communication for international customer relationships. The findings suggest that both practitioners and researchers believe that mFAS might be especially beneficial in such customer relationships. Furthermore, the findings also identified some challenges: Slow performance of the mobile network might lead to quality problems when using social presence features, such as desktop sharing and co-browsing, and might consequently lead to poor customer experience. Clearly, such challenges need to be addressed when developing mFAS. (CR5) Transparency. With regard to transparency, researchers have previously elicited the requirement for documenting the information exchange between customers and RMs. According to this requirement, customers need to be able to access previous calls or product recommendations and assess whether these suggestions have actually improved the financial performance. Furthermore, if RMs initiate such recommendations, the way in which they meet 523 Designing Tablet Banking Apps for High Net Worth Individuals the pre-defined investment strategy needs to be transparent, and lie within the risk tolerance of customers. (CR6) Privacy. Banks and RMs are both eager to gain more insights into customer behaviors by analyzing data such as recent transactions. However, previous studies have highlighted that customers need to be in control of the kind of data the banks and RMs collect and analyze. Hence, customers should be able to control and configure such data collection and analysis practices in the mobile app. In the subsequent section, the way in which the requirements specification process of these six CR (CR1-6) was pursued is discussed in detail. 3 Research Design In Section 3.1 the research endeavor is highlighted and the design science research (DSR) method from Peffers et al. (2007) is described. Section 3.2 provides details on the development and evaluation cycles of the prototype. 3.1 Design Science Research (DSR) Activity 1: Identification of the problem and motivation (DONE). The motivation for the topic is provided in the introduction (Section 1) of this paper. Providing mFAS will become crucial in order to provide customers with a transparent advisory process and ultimately to meet customer expectations with regard to such a service. Activity 2: Definition of objectives and requirements for the artifact (DONE). Previously published work (Ruf et al., 2014) has elicited CR following the RAM of Gorschek and Wohlin (2006). As a result, the researchers have derived various CR from a multi-method approach which included empirical findings involving domain experts and customers. The results of this activity were introduced in Section 2.3 above. Activity 3: Design of an artifact (DONE). In this study, we designed a prototype with specified CR for mFAS. The following Section 3.2 highlights details on the research approach and chosen method. This research project involved experts from various banks in Switzerland and did not receive funding from a particular bank. Hence, we argue that the findings are more generalizable and unbiased than if the project had been funded by a single project partner. Activity 4: Demonstration (OPEN). The artifact has been demonstrated with an experiment involving participants and potential customers; this ended in December 2014 (Ruf, Back, & Wittmann, 2015). We are currently in the process of analyzing the data. Activity 5: Evaluation (ONGOING). Following the experimental demonstration, we plan to evaluate the artifact with customers in cooperation with a bank in Switzerland. However, this evaluation is still in the planning process and is dependent on the results of the experimental 524 Christian Ruf, Andrea Back, Henk Andreas Weidenfeld demonstration. Furthermore, we believe that each activity should include a separate evaluation process. Hence, we present the evaluation of the specified CR in Section 4 of this study. Activity 6: Communication (ONGOING). We plan to communicate our findings and results on a continuous basis and get valuable feedback from peer-reviewed conferences and journals. As the scope of this study refers to Activity 3 and Activity 5 of the DSR, we provide further details on how we built our prototype and planned a first evaluation cycle. 3.2 Chosen Research Approach and Method Regarding Activity 3, we designed an artifact based on specified CR. We conducted three design-and-evaluation iterations (Activity 5) which are described in further detail. Figure 1 depicts our procedure in developing and evaluating the CR. Development Phase 1: Design of mock-ups and a first clickable prototype. For the design of the user interface, we chose Adobe Illustrator. We developed the advisory process, the navigation, and the look and feel of it. Subsequently, we used these interfaces to build a first clickable prototype with InVision software. This allowed us to simulate the advisory process by linking the interfaces and navigation sites. We evaluated this prototype in a first iteration. Evaluation Phase 1: Focus group with researchers. With this first evaluation, we ensured that the prototype included the previously elicited System Requirements presented in Section 2.3. We incorporated small changes, such as switching the language from German to English, and adapting the look and feel of the menu. In total, three Research Associates and a Professor provided feedback regarding the completeness and consistency of implementing the CR in this prototype. The participants had previous knowledge in the domain of either interactive design or the financial industry. Development Phase 2: Design interactive prototype v1. Based on the input and feedback from the first evaluation, we were able to further specify the CR and design an interactive prototype accordingly. Where possible, we used Axure RP in combination with HTML5 and JavaScript to develop this interactive prototype. Furthermore, we coded the social presence features, such as desktop sharing and the chat function with PHP and created a MySQL database. Evaluation phase 2: Focus group with Research Associates and Master’s Students. In the second evaluation and iteration, we presented the interactive prototype and the customer journeys to Research Associates and master’s students who were either involved in user experience projects or the requirements elicitation process for such an mFAS. We were able to specify the CR and gain a more comprehensive understanding. 525 Designing Tablet Banking Apps for High Net Worth Individuals Figure 1: Design and evaluation phases Development Phase 3: Design interactive prototype v2. This process involved an incremental improvement of the interactive prototype from the previous development phase. Final Evaluation: Focus group with seven domain experts. For the final evaluation, we invited seven experts with extensive industry experience. We summarized the roles and experiences of these experts in Table 1. During the focus group, we presented the final prototype, gathered additional feedback in order to specify CR, and consequently evaluated its consistency and completeness. We organized the focus group for the final evaluation on June 26th 2014. The session lasted two hours. Three Research Associates were responsible for recording the minutes. Following the discussion, the participants were asked to fill out a questionnaire, for which the experts evaluated the CR with regard to the quality criteria of the recommended practice for requirements specification (IEEE, 1998). The experts were asked to agree or disagree whether the specified CR met the suggested quality criteria for requirements specification using a scale where 1=“I completely disagree”, 2=“I disagree”, 3=“I partly agree”, 4=“I agree” and 5=“I completely agree”. Table 3 in Section 4 summarizes the survey questions and the findings from this final evaluation. 526 Christian Ruf, Andrea Back, Henk Andreas Weidenfeld Position Domain experience Organization # Employees Head of Banking Consulting More than 10 years Consulting Firm < 50 Senior Manager IT Architecture More than 10 years Private Bank 1,500 Head of Online Private Banking 8 years Universal Bank >10,000 Head of Private Banking 5 years Universal Bank 1,000-1,500 Manager IT Architecture 5 years Private Bank 1,500 Software Developer 5 years Universal Bank 1,000-1,500 Assistant Manager Online 2 years Universal Bank 5,000-5,500 Channels Table 1: Focus group with seven experts for the final evaluation 4 Results During the first two DSR cycles, we specified the CR as summarized in Table 2. These specified CR and the prototype are presented in Figures 2 and 3. Figure 2: Prototype with the customer requirements (CR1,4,5) transparency, access to experts, social presence and situational use With regard to (CR1) access to experts, the focus group with the domain experts suggested that depending on the importance of customers, they should be able to contact experts and investment advisory team members directly. Hence, whether RMs serve as a single point of contact really depends on how much wealth customers have or how important they are. 527 Designing Tablet Banking Apps for High Net Worth Individuals Accordingly, RMs should be able to customize this feature. Furthermore, the evaluation cycles revealed that customers should only be able to use chat. Thus, only RMs should be able to initiate video and desktop sharing features (Figure 2). Feature Level* Description* Feature Level Function Level (continued) (CR1) Access to RMs are the single The RM is a single point Customers are able to experts point of contact. of contact, but is able to request a meeting and customize the chat or send messages. accessibility of the Video calls are initiated advisory team. by the RM. (CR2) Information The information on The platform includes The platform visualizes quality the platform is timely, both research the portfolio and the and aggregates news information and pre-defined investment according to the information of the strategy. individual customer’s customer’s current risk profile. portfolio. (CR3) Proactivity The service supports Such recommendations Customers are able to the RM sending out include rebalancing accept or decline such product requests but also invitations and request recommendations. invitations to exclusive additional information. events. (CR4) Situational use Customers are able to Such interactions include If the mobile network is and social access the personal chat and desktop not fast enough for using presence RM from anywhere, at sharing. Video such features, this any time. conferencing is not a should be graphically priority. highlighted. (CR5) Transparency In order to address Transparency relates The product site displays information and both to the product all relevant information interest asymmetries, recommendation and to in a comprehensive way the mFAS provides a the entire for the customer. transparent advisory communication between Furthermore, the process. RMs, customers and the communication center financial advisory team. archives client touch points. (CR6) Privacy While privacy is Customers need to be On the first login, critical for customers, aware of what kind of customers are able to RMs require insights data the app collects and configure the data about their clients. how it is analyzed. collection and data The mFAS should analysis practices. balance these two requirements. Table 2: Specified customer requirements (CR) for mFAS, *feature requirements in a previous study (Ruf et al., 2014). 528 Christian Ruf, Andrea Back, Henk Andreas Weidenfeld The Feature Level requirement (CR2) information quality refers to the aggregation of research information and investment advice. In our evaluation process, the conclusion was that such information relates not only to investment ideas and corresponding products, but also to the clients’ current investment portfolio. Consequently, the mFAS should match the investment ideas and research information according to this portfolio information and provide a more customized and personalized service (Figure 3). Regarding (CR3) proactive information, this should always include buy and sell orders combined. Practitioners also refer to such buy and sell orders as rebalancing. Furthermore, clients are interested in exclusive events to which RMs might also invite them. With regard to such proactive information, clients should be able to quickly accept or decline such recommendations. In our prototype, this CR was implemented with three simple buttons; customers could accept the recommendation, decline it (Figure 3), or request additional advice. During the evaluation, we also specified the Feature Level requirement (CR4) situational use and social presence. Previous studies have emphasized the relevance of an mFAS for managing international client relationships. Our findings suggest that videoconferencing, or being able to see the other person, is not a main priority. Desktop sharing or co-browsing features are more relevant, in order to provide a better advisory service. Furthermore, the mFAS should notify customers if the performance of the mobile network is not sufficient for using such features. For example, if the customer does not have wireless or 3G network access, the desktop sharing and co-browsing features are disabled. In our prototype, the availability of chat and social presence features was highlighted with a green circle around the portrait picture (Figure 2). Regarding (CR5) transparency, we designed a dedicated communication center which incorporated the entire communication streams between customers, RMs and the expert or investment advisory team members. Consequently, customers were able to verify whether the investment proposals and recommendations from previous interactions had actually resulted in increased financial performance. We also designed the product site according to transparency criteria. The product recommendations contained the transaction costs associated with a trade and information on how the product fit with the person’s risk tolerance, risk profile and pre-defined investment strategy (Figures 2 and 3). 529 Designing Tablet Banking Apps for High Net Worth Individuals Figure 3: Prototype with the customer requirements (CR2,3,5) information quality, transparency and proactivity Finally, we also specified the last Feature Level requirement (CR6) privacy. We discussed the importance of privacy with regard to collecting and analyzing customer data. While financial institutions and RMs in particular try to collect and analyze data for a better understanding of customers, privacy issues remain one of the top concerns of customers. Hence, we implemented a notification at the beginning of the login process. With a simple click, customer could adjust their privacy settings and decide what kind of personal data they wanted to share with the financial institution. Following the requirements specification process and the design of the prototype as depicted in Figures 2 and 3, we asked the participants to evaluate the CR according to the recommended practice for requirements specification (IEEE, 1998). We present the results of this final evaluation in Table 3. The experts positively evaluated the specified CR as being (1) consistent and correct, (2,3) unambiguous, (4) modifiable, and (5) traceable as well as transparent. Regarding the quality criteria (6) ranked for importance and (7) measurable, the experts only partly agreed with our findings. Finally, compared to the other quality criteria, the experts were more skeptical with regard to the (8) completeness of our specified CR. Hence, some of experts disagreed or only partly agreed that our specified CR are complete. These findings give rise to discussion, which is addressed in the following section. 530 Christian Ruf, Andrea Back, Henk Andreas Weidenfeld The specified customer requirements (CR) … Feedback (1)…are consistent and meet the customer and stakeholder agree needs. (2)…can only be interpreted one way. agree (3)…are unambiguous. agree (4)…are modifiable. agree (5)…are transparent and traceable. agree (6)…are ranked for importance. partly agree (7)…are easily transformed into measurable performance partly agree indicators. (8)…are complete. disagree/partly agree Table 3: Results from the final evaluation and the focus group 5 Discussion When looking at the results from Table 3 in Section 4, the conclusion that can be drawn is that by applying the RAM model we successfully specified CR that met most of the quality criteria. The experts agreed with our specified CR being correct, consistent, unambiguous, modifiable, transparent and traceable. Hence, we argue that the RAM model provided a useful framework in the RE process. While the experts positively evaluated most of the quality criteria and, hence, agreed with how we specified CR and built our prototype, the results indicate that the presented CR might be only partially complete. Regarding the completeness criteria, some of the experts either disagreed or only slightly agreed. There might be several reasons for this critical assessment. First, our presented CR were still generic and abstract. The CR would need to be specified on the Component Level of the RAM in order to provide more complete and specific requirements in the business context of each practitioner, as suggested by Gorschek and Wohlin (2006). Secondly, the final evaluation also provided us with new requirements, which had not been considered thus far. One statement provided during the evaluation was the following: “Depending on the customer needs, we should allow the customers to design their own app with the features and functions they need” . For example, a trader might want to execute the transaction personally, while the RM should facilitate these transactions for other customers. Thirdly, we only elicited customer-related requirements (CR1-6). Accordingly, business processes, the existing information systems and other stakeholders within an organization also have requirements which were not addressed in this study. Such additional requirements might also originate from the political environment. One of the experts mentioned the following: “New regulatory frameworks are a huge challenge for us. Which customers are we able to consult with the new financial intermediary and consulting regulation?” To sum up, we believe that specifying requirements on the Component Level in a real-life context, as well as 531 Designing Tablet Banking Apps for High Net Worth Individuals capturing requirements from additional stakeholders, would have resulted in more positive feedback with regard to the completeness criterion. Regarding the quality criterion (6) modifiable and transformable into key performance indicators, we want to highlight an item of feedback from the focus group: “At the end of the day, we need to be able to make money with this service. How are we going to price such an app?” Clearly, the CR presented in this study did not provide specific figures on increasing customer satisfaction, financial performance or profits. By addressing this limitation, we believe that the feedback from the experts with regard to this criterion would have been more positive. Finally, the presented CR1-6 were not prioritized on a quantitative scale. Hence, only the relative importance of these CR in the focus group could be assessed. For example, in the opinion of the group and based on previous findings (Ruf et al., 2014), privacy is the top concern and a prerequisite which must be addressed when developing mFAS. While privacy issues are clearly of significant importance, proactivity is less of a priority. However, such a qualitative assessment did not completely meet the criterion “ranked for importance”. 6 Conclusions, Limitations and Future Research In this study, we aimed at specifying customer requirements (CR) for a mobile financial advisory service (mFAS) with the instantiation of a prototype. In order to achieve this goal, we conducted 3 development and evaluation cycles. The final evaluation included a focus group with seven domain experts. Besides the specified CR (1) access to experts, (2) information quality, (3) proactivity, (4) situational use and social presence, (5) transparency, and (6) privacy, we also captured new ideas on how to improve our prototype. Furthermore, the evaluation also revealed how effectively the specified CR met the recommended practice for requirements specification (IEEE, 1998). Our findings suggest that our CR are consistent, correct, unambiguous, modifiable, traceable and transparent. However, the experts were more skeptical with regard to the completeness criterion. Consequently, we believe that future studies should also address different stakeholder requirements, such as the environment, business processes, and the existing information systems in an organization in order to improve the completeness of the presented CR. Apart from that, we believe that the provided CR provides insights on how practitioners design mFAS in their organizational context. It would be particularly interesting to evaluate how the proposed CR also applies to different segments, such as retail or affluent customers. In our study, we developed a mobile app that runs in the browser of tablets. Future studies might also evaluate how the specified CR are applicable to mobile apps on smartphones. Furthermore, our results show an instantiation of the Requirements Abstraction Model (RAM) from Gorschek and Wohlin (2006), combined with a prototyping approach. By applying the proposed model for specifying CR, we instantiated the model and acknowledge its usefulness. Furthermore, we combined the specification process with a prototyping approach in three 532 Christian Ruf, Andrea Back, Henk Andreas Weidenfeld iterations. Hence, we argue that the RAM is a useful method for capturing and specifying requirements. Despite the presented results and contributions, we also want to discuss some limitations. The evaluation phases of our CR and prototype included Research Associates, a Professor, and Master’s students, as well as seven experts with significant industry experience. While we made sure to include only experienced people in our evaluation process who had good knowledge and understanding of customer needs, the involvement of HNWI in the evaluation cycles would have provided us with additional valuable feedback. However, we only had limited access to HNWI and thus were not able to address this limitation in our study. Consequently, future research endeavors should incorporate additional feedback from this customer segment. Notably, we are currently in discussion with various banks in order to get access to HNWI clients for a future validation process. References Aurum, A., & Wohlin, C. (2005). Requirements Engineering: Setting the Context. In A. Aurum & C. Wohlin (Eds.), Engineering and Managing Software Requirements SE - 1 (pp. 1–15). Springer Berlin Heidelberg. doi:10.1007/3-540-28244-0_1 Byrd, T. A., Cossick, K. L., & Zmud, R. W. (1992). A Synthesis of Research on Requirements Analysis and Knowledge Acquisition Techniques. MIS Quarterly, 16(1), 117. doi:10.2307/249704 Gemes, A., Ammann, C., & Lenzhofer, A. (2010). Private Banking - After the Perfect Storm. booz&co. Retrieved November 6, 2014, from http://goo.gl/YHVocL Gorschek, T., Garre, P., Larsson, S. M., & Wohlin, C. (2007). Industry evaluation of the Requirements Abstraction Model. Requirements Engineering, 12(3), 163–190. doi:10.1007/s00766-007-0047-z Gorschek, T., & Wohlin, C. (2006). Requirements Abstraction Model. Requirements Engineering, 11(1), 79–101. doi:10.1007/s00766-005-0020-7 Hofmann, H. F., & Lehner, F. (2001). Requirements engineering as a success factor in software projects. IEEE Software, 18(4), 58–66. IEEE. (1998). IEEE Recommended Practice for Software Requirements Specifications. IEEE Std 830-1998. doi:10.1109/IEEESTD.1998.88286 Inbar Noam, O. (2012). Lowering the line of visibility: incidental users in service encounters. Behaviour & Information Technology, 31(3), 245–260. Retrieved from 10.1080/0144929X.2011.563796 533 Designing Tablet Banking Apps for High Net Worth Individuals ISO. (2011). ISO 22222:2005 - Personal Financial Planning. Retrieved from http://goo.gl/P8qqXa KPMG. (2013). Success through innovation - achieving sustainability and client-centricity in Swiss private banking. Retrieved April 9, 2014, from http://www.kpmg.com/ Nussbaumer, P., Matter, I., Reto à Porta, G., & Schwabe, G. (2012). Designing for Cost Transparency in Investment Advisory Service Encounters. Business & Information Systems Engineering, 4(6), 347–361. doi:10.1007/s12599-012-0237-1 Nussbaumer, P., Matter, I., & Schwabe, G. (2012). “Enforced” vs. “Casual” Transparency -- Findings from IT-Supported Financial Advisory Encounters. ACM Transactions on Management Information Systems, 3(2), 1–19. doi:10.1145/2229156.2229161 Oehler, A., & Kohlert, D. (2009). Financial Advice Giving and Taking—Where are the Market’s Self-healing Powers and a Functioning Legal Framework When We Need Them? Journal of Consumer Policy, 32(2), 91–116. doi:10.1007/s10603-009-9099-4 Peffers, K., Tuunanen, T., Rothenberger, M. A., & Chatterjee, S. (2007). A Design Science Research Methodology for Information Systems Research. Journal of Management Information Systems, 24(3), 45–77. doi:10.2753/MIS0742-1222240302 Pohl, K. (2008). Requirements Engineering: Grundlagen, Prinzipien, Techniken (Vol. 2). dpunkt Verlag GmbH. PwC. (2013). Navigating to tomorrow: serving clients and creating value - Global Private Banking and Wealth Management Survey 2013. Retrieved from www.pwc.com/wealth Ruf, C., Back, A., Bergmann, R., & Schlegel, M. (2014). Elicitation of Requirements for the Design of Mobile Financial Advisory Services – Instantiation and Validation of the Requirement Data Model with a Multi-method Approach. In 48th Hawaii International Conference on System Sciences (Kauai, Hawaii). Ruf, C., Back, A., & Wittmann, M. (2015). Is an App Better than an Email? Developing Trust in a Mobile Financial Advisory Service : Design and Evaluation of a Prototype. In Wirtschaftsinformatik Proceedings 2015 (pp. 225–239). AIS Electronic Library (AISeL). Vijayasarathy, L., & Turk, D. (2008). Agile Software Development: A survey of early adopters. Journal of Information Technology Management, 19(2), 1–8. Yousafzai, S. Y., Pallister, J. G., & Foxall, G. R. (2003). A proposed model of e-trust for electronic banking. Technovation, 23(11), 847–860. doi:10.1016/S0166-4972(03)00130-5 534 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Business Models Innovation for SMEs: platforms, tools and research Chair: Christer Carlsson, Professor IAMSR/Abo Akademi University, Finland Panelists Harry Bowman, Professor TU Delft, The Netherlands & IAMSR/Abo Akademi University, Finland Jukka Heikkila, Professor, University of Turku, Finland Timber Haaker, Innovalore, The Netherlands Abstract In this session we discuss the relevance of Business Model Innovation for SMEs, the role and significance of tools and platforms, as well as the relation between BMI innovation and performance. BM Innovation is poorly defined and used as an open concepts, mainly related to either start up companies or to large corporations. A focus on specific enterprises like micro- enterprises, family businesses or female entrepreneurs is often lacking. In this session we will present results and insights from the European Horizon2020 project ENviSION (Understanding and supporting business model innovation for small businesses through ICT-based tooling). ENviSION aims at supporting business model innovation by small and medium sized enterprises. ENviSION will advance understanding of how small businesses do business model innovation, but also develop ICT-based tools to enable business model innovation. The ultimate goal is to bring ICT-based tooling for business model innovation to at least three million enterprises across Europe. BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Cloud Computing: Towards Business e-Model Enlightenment Trevor Clohessy (Chair & Panel), Researcher, Lero - The Irish Software Research Centre, National University of Ireland Galway, Ireland trevor.clohessy@nuigalway.ie Dr. Thomas Acton (Panel), Head of School of Business & Economics, National University of Ireland Galway, Ireland thomas.acton@nuigalway.ie Dr. Lorraine Morgan (Panel), Senior Researcher, Lero - The Irish Software Research Centre, National University of Ireland Galway, Ireland lorraine.morgan@nuigalway.ie Barry Reddan (Panel), Senior Manager, Hewlett Packard barry.reddan@hp.com Dr. Ultan Sharkey (Panel), CEO, Sharkey cONSULTING ultan@sharkeyconsulting.ie Panel Outline The forthcoming years will be a crucial period for the development of cloud computing. Built upon foundations in virtualisation, distributed computing, utility computing, networking, and more recently, web and software services, ‘cloud’ represents a shift to computing as a service, enabling a fundamental change in how information technology is provisioned. From a business model perspective, the rationale for provisioning or consuming cloud computing is compelling in terms of agility, innovation, lower costs, and scalability. To reap the benefits of cloud computing, organisations often require parallel innovation in business models, organisational processes, structures and skills. However, recent research suggests that both cloud service providers and service users are concurrently experiencing substantial difficulties in their attempts to effectively leverage the transformational business capabilities afforded by cloud computing. These difficulties can often be rooted in a reluctance or inability of organisations to alter their existing business models when attempting to leverage the nascent capabilities of digital technologies. This is compounded by the fragmented fuzzy descriptors underlying the business model concept, the rapidly evolving cloud computing technological landscape and the inherent complexities of cloud computing architectures. Moreover, concrete examples of how cloud computing can benefit enterprises and customers from a business model perspective are required. For organisations wishing to leverage the propitious capabilities associated with cloud computing, it is imperative from the outset that IS researchers are proactively involved in every discussion regarding the paradigm. The panel session will provide a current snapshot of leading business model research in cloud computing, capturing both provision and consumption perspectives. The panellists will discuss the state of the art research on the business model value of cloud via the following agenda:  An overview of the current cloud computing business model landscape  Demonstrate exemplars of how enterprises have successfully leveraged cloud  Demonstrate challenges which are currently stagnating cloud growth  Suggest a research agenda for both IS researchers and practitioners Chair and Panel Biographies Thomas Acton is Head of School, and senior lecturer at the J.E. Cairnes School of Business & Economics, National University of Ireland, Galway, Ireland. His research interests lie in cloud computing, mobility, decision support systems, and usability and acceptance. He has also served as associate editor on a number of journals, including the European Journal of Information Systems. Tom has also organised and chaired a number of national cloud industry workshops. Trevor Clohessy is a researcher with Lero, the Irish Software Research Centre. His research interests include cloud computing, business models, strategy and business transformation. His research on business models for cloud computing have been published at a number of academic and practitioner fora, including ECIS and Bled. Trevor has served as an editorial advisory board member for IGI’s “Handbook of Research on Architectural Trends in Service-Driven Computing”. Lorraine Morgan is senior research fellow with Lero, the Irish Software Research Centre. Her principle research interests are cloud computing, open innovation, open source software and crowdsourcing. Lorraine’s research has been published in leading journals, including the European Journal of Information Systems, Journal of Strategic Information Systems, Database for Advances in Information Systems and Information and Software Technology. Barry Reddan is a Senior Manager leading the Hewlett Packard's Enterprise Security Products consulting Practice in EMEA. Barry has worked in the IT Security Sector for 13 years and interfaces with cross vertical enterprise accounts and public sector clients on a PAN-EMEA basis to help to shape their security strategies, ensuring that clients are positioned to mitigate risk attached to potential security breaches, by leveraging industry leading HP ArcSight SIEM technology. SIEM integration with Cloud based applications is a very hot market driven requirement which Barry and team regularly work to deliver solutions against. Ultan Sharkey is CEO of Sharkey Consulting, a specialty e-commerce consulting firm with a global client base. Leveraging cloud solutions, Ultan has worked with small to medium corporate enterprises to deliver class-leading web strategies and presences, systems optimisation, systems integration and configuration, and scalable systems solutions. BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Digital Wellness Services for Young Elderly: New Frontiers for Mobile Technology Chair: Christer Carlsson, Professor IAMSR/Abo Akademi University, Finland christer.carlsson@abo.fi Panelists: Harry Bowman, Professor TU Delft, The Netherlands & IAMSR/Abo Akademi University, Finland W.A.G.A.Bouwman@tudelft.nl Doug Vogel, Professor Harbin Institute of Technology, China isdoug@hit.edu.cn Pirkko Walden, Professor IAMSR/Abo Akademi University, Finland pirkko.walden@abo.fi Panel Outline The panel on Digital Wellness Services for the Young Elderly will address the needs for the young elderly to develop a sustained use of wellness routines in order to reduce the risk of suffering from functional impairment with advancing age. The young elderly is the age group 60-75 years, which is expected to be 97 million within the EU by 2020. Recent studies have shown that functional impairment in the young elderly age group will carry an increased risk for extensive functional impairment in the following, the senior 75+ , age group at high and increasing cost for public health and social care. In Finland, with a population of 5.3 million, the cost for public health and social care for the ageing was 3.8 B€ in 2014. Digital wellness services can be produced and delivered over mobile smartphones with back- end cloud service support. This will make the services ubiquitous, affordable and flexibly adaptive to a multitude of user needs – the services can be extensively tailored to very different user needs for young elderly from different countries, in different cultures, with different socio-economical background, with different technology skills, etc. Recent studies have shown that the mobile platform for the wellness services should be omnivore, i.e. it should connect to and support a wide variety of digital devices and services, for which there should be information and knowledge support from the cloud services. A first omnivore platform with 100+ interfaces has been developed and tested with young elderly in Finland. The wellness services need to be adopted and used continuously by the young elderly for wellness routines to be formed and sustained. This requires the development of new forms for user centred design of digital services – a co-creation of digital wellness services with the young elderly. The services can be sustained only if there is an ecosystem of service developers, digital platform and cloud services developers and operators, information and knowledge developers and providers, wellness consultants and trainers, system integrators with public health and social care systems, etc. In other words, the ecosystems for digital wellness services offer embryos for a wellness service industry; there is demand from a potential market of 97 million consumers in the EU countries (and about 1 billion consumers globally) for wellness solutions that are win-win-win: (i) wellness services will improve the quality of life for the young elderly [“it is nicer to get older if you are in good shape”]; (ii) sustainable wellness routines will reduce the risk for functional impairment with increasing age, which will have a significant impact on the development of the cost for public health and social care; (iii) the growth of digital wellness services will offer revenue streams for hundreds (first, then thousands) of SMEs that form the digital wellness ecosystems. The panel will explore several challenges and opportunities for digital wellness services: the development of mobile technology, the design of digital wellness services, the distribution and adoption of wellness services among the young elderly, the forming of wellness routines through sustainable wellness services, the impact of wellness routines on the risk for functional impairment, the forming of ecosystems for digital wellness services – and probably a number of emerging, new issues. The Panel will be open to all Conference participants who are welcome to follow and comment on the introductions by the panellists and to contribute findings, experience, ideas and proposals to the theme of the Panel [which is an ongoing research program]. 2 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia eHealth: Security, Privacy and Reliability Co-chairs Nilmini Wickramasinghe, Professor, Deakin University and Epworth Health Care, Australia Juergen Seitz, Professor, Baden-Wuerttemberg Cooperative State University, Germany Panelists: Urban Schrott, IT Security and Cybercrime Analyst, Communications manager at Reflex, ESET Ireland, Safetica UK & Ireland Vladislav Rajković, Professor Emeritus University of Maribor, Slovenia Igor Košir, Strategic progams director, Smartcom, Slovenia Abstract Today, data can be securely stored in the cloud as well as in a telematics infrastructure. Encryption algorithms which have been used to protect against brute force attacks may be not good enough to ensure security for future cryptanalysis computing capabilities. What happens if encrypted data are collected now and decrypted later? How do we have to encrypt data today to ensure they are secure for the next for example twenty or thirty years? These are reasonable time frames which exist in the area of health data given that most countries insist health data is maintained at last for the life of the individual. BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia eHealth: Towards independent living Co-chairs Nilmini Wickramasinghe, Professor, Deakin University and Epworth Health Care, Australia Juergen Seitz, Professor, Baden-Wuerttemberg Cooperative State University, Germany Panelists: Luuk P.A. Simons, Professsor Delft University of Technology, Netherlands Vladislav Rajković, Professor Emeritus University of Maribor, Slovenia Mateja Sajovic, Head of IT infrastructure, University Medical Center Ljubljana Vesna Prijatelj, Business director of Independent hospitals of University Medical Center Ljubljana Igor Košir, Strategic programs director, Smartcom, Slovenia BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia e-Solutions to Cluster Analysis and Knowledge Sharing Chair Michael Walsh, Cluster Manager it@cork European Tech Cluster, Ireland michael@itcork.ie Panellists Marc Pattinson, Manager Inno Group, Sophia Antipolis, France m.pattinson@inno-group.com John Hobbs, Senior Lecturer, Department of Management & Enterprise Cork Institute of Technology, Ireland John.hobbs@cit.ie Tamara Högler, Head of Innovations and International Affairs CyberForum e.V., Germany hoegler@cyberforum.de Darja Radić, Project Manager Automotive Cluster of Slovenia darja.radic@poly4emi.eu 1 e-Solutions to Cluster Analysis and Knowledge Sharing The emergence of Smart Specialisation and regional innovation policy as part of national industry policy is a result of more than four decades of analysis and empirics, which have reshaped the understanding of the role played by innovation in economic development and in particular its relationship with geography. Cluster polices (Porter, 1990) and more recently Triple Helix Cluster policies (Leydesdorff, 2012) which combine the roles of ‘academia,’ ‘Government’ and ‘industry’ have had a significant impact in regard to innovation and economic growth, providing a more coordinated approach. What started as a relatively narrow sectoral and science based R&D way of thinking about innovation policy has developed into a much more multi-dimensional policy approach involving industry, institutions, agencies and cross border collaborations. Industry clusters have been particularly useful in helping policy makers understand how certain sectors can grow through developing regional strengths. This panel session is focused on ‘e-Solutions to Cluster Analysis and Knowledge Sharing,’ and explores how eTechnologies are becoming more ingrained in the day to day running of cluster organisations and state mandated development agencies. The panellists presentations relate to two distinct areas: 1) ‘Cluster Analysis,’ focused on the assessment of cluster networks and developing polices aimed at economic growth, and 2) Cluster ‘Knowledge Sharing’ relating to the day to day operations of a cluster, communications between its members and policy supports to realise shared visions. Cluster Analysis is a focus of two panellists, Marc Pattinson, Inno Group, France and John Hobbs, Cork Institute of Technology, Ireland. Marc Pattinson’s work is primarily engaged in the areas of innovation, regional economic development, programme evaluation and cluster support and Marc is an approved Gold label cluster auditor. The inno group serves research institutions, clusters, businesses, policy institutions and development agencies across Europe and Internationally. Marc will present the analysis toolkit used by the European Cluster Observatory when analysing clusters across Europe developed on behalf of DG GROWTH. With its origins in Harvard Business School through the work of Professor Michael Porter, the observatory continues to develop as a toolkit for developing innovative cluster policy initiatives http://ec.europa.eu/growth/smes/cluster/observatory/index_en.htm. John Hobbs presents V-LINC an eSolution methodology for identifying, recording and analysing the linkages that firms in clusters engage in. It categorises business linkages, and groups them by geographic scope. Furthermore, V-LINC records the business significance of linkages based on the perceptions of firm personnel who engage in the linkages with other companies and organisations. Data for V-LINC analysis of linkages is collected through structured interviews of company personnel. V-LINC maps give a visual representation of the relative reliance on local, national, European or Global linkages of a company, or when combined, of a cluster. V-LINC facilitates policy development at regional and national levels, through the aggregation of data from a sample of firms. Initial applications of V-LINC have been conducted on European ICT clusters through the BeWiser FP7 funded project www.be-wiser.eu. 2 Cluster ‘Knowledge Sharing’ is a focus of Tamara Högler, CyberForum e.V., Germany and Darja Radić, Automotive Cluster of Slovenia, both of whom have practical experience as cluster managers working in Industry. Tamara Högler’s work is primarily focused on innovation and internationalisation. She will present CyberForumś approach to boost regional innovation by trust-building, pro-active connecting of companies as potential cooperation partners and supporting their internationalisation activities. She will describe the Business Roaming Agreement, that can be seen as a soft-landing platform for companies that want to become international but lack either experience in this topic and / or a vast “internationalisation budget”, as well as illustrate the potential of EU-funded projects as appropriate tools for innovation and internationalisation. Some perspectives from the Smart BusinessIT initiative (www.smartbusiness-it.de) a regional Cluster Accelerator Programme developed in Baden-Wuerttemberg. and the UPSIDE FP7 funded project (www.upside-project.eu) will also be shared. Darja Radić, maintains that knowledge is no longer limited by geography space, rather knowledge flows globally, without borders. Companies need to capture the best know- how available globally, in locations beyond their regional and national boundaries. Linkages between clusters in different locations which offer complementary strengths can provide access to the most advanced technologies and know-how. There is a need to provide better opportunities for cluster in relation to cross-regional and tran-regional cooperation. New policy approaches can enhance such cooperation through policy measures, which support job creation via industrial transformation. We have to better understand the policy framework of clusters from different regions, and various cross- sectoral solutions which are essentially a combination of knowledge, experiences, industrial landscape and conducive framework conditions. Ms. Radić presents the peer review approach, experiences and lessons learned within the poly4emi project http://www.poly4emi.eu/. 3 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia How to publish your conference paper in a journal? Co-chairs Roger Bons, Director, Bons Academic Services, The Netherlands & 2015 Bled eConference Research Track Chair Johan Versendaal, Professor of Extended Enterprise Studies, Research Centre Technology & Innovation, HU University of Applied Sciences Utrecht & Professor of E-Business, Faculty of Management, Science and Technology, Open University of the Netherlands & 2015 Bled eConference Research Track Co-Chair Andreja Pucihar, Associate Professor Faculty of Organizational Sciences, University of Maribor, Slovenia & 2015 Bled eConference PC Chair Panelists Hans-Dieter Zimmermann, Co-Editor in Chief Electronic Markets - The International Journal on Networked Business, www.electronicmarkets.org Narciso Cerpa, Editor in Chief Journal of Theoretical and Applied Electronic Commerce Research www.jtaer.com Nilmini Wickramasinghe, Editor in Chief International Journal of Networking and Virtual Organisations www.inderscience.com/jhome.php?jcode=ijnvo Ronald S. Batenburg, Editorial Member International Journal of Organisation Design and Engineering www.inderscience.com/jhome.php?jcode=ijode Jože Zupančič, Editor in Chief Organizacija – Journal of Management, Informatics and Human Resources organizacija.fov.uni-mb.si 1 Panel Outline The purpose of this panel is to guide authors of conference papers how to make their papers publishable in an academic journal. Therefore editors of the five partner journals of the Bled eConference will be present on the panel: Electronic Markets - The International Journal on Networked Business, Journal of Theoretical and Applied Electronic Commerce Research, International Journal of Biomedical Engineering and Technology, International Journal of Organisation Design and Engineering, Organizacija – Journal of Management, Informatics and Human Resources. The journal’s representatives will give an overview about their respective journals addressing: scope, different formats of publications, submission, review, and publication process, editorial structures, rankings, etc. In addition, the editors will give detailed advice addressing the requirements and the processes to transform a conference paper into a publishable paper in their respective journals. Finally, authors will have the opportunity to address the editors with specific questions. 2 BACK 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Social media for business: current state and future trends Co-Chairs Jari Salo, Professor, Oulu Business School Finland Hans-Dieter Zimmermann, Professor, FHS St. Gallen, Switzerland Presenters Gregor Zupan, Information Society, Statistical Office of the Republic of Slovenia Marko Perme, Director, Agilcon, Slovenia Urban Schrott, Ireland - IT Security and Cybercrime Analyst, Communications Manager at Reflex / ESET Ireland / Safetica, United Kingdom and Ireland Panel Outline Social media is transforming every aspect of business. Organizations do not have just a Twitter account or a Facebook page, but they are also using social media tools to support their everyday business processes and decisions. Social media has changed the way how organizations are doing business and it is expected to bring more changes and challenges in the future. The objective of this panel is to provide discussion on current state of social media applications for business, to highlight some of best practices, to discuss current issues and to identify further trends of social media usage for business. The panellists will share their insights, experiences and predictions. BACK 28th Bled eConference #WellBeing June 7 - 10, 2015; Bled, Slovenia Future University Chair: Mirjana Kljajić Borštnar, Assistant Professor & Vice Dean for Research, Faculty of Organizational Sciences, University of Maribor, Slovenia Andreja Pucihar, Associate professor & Vice dean for international cooperation, Faculty of Organizational Sciences, University of Maribor, Slovenia Presenters Johan Versendaal, Professor of Extended Enterprise Studies, Research Centre Technology & Innovation, HU University of Applied Sciences Utrecht & Professor of E- Business, Faculty of Management, Science and Technology, Open University of the Netherlands Roger Bons, Director, Bons Academic Services, The Netherlands & 2015 Research Track Chair Tomi Ilijaš, Director, Arctur, Slovenia Amira Mujanović, Blaž Sašek, Students, University of Maribor, Slovenia Rok Kepa, Blaž Vidmar, Students, University of Ljubljana, Slovenia Information Literacy of Students: Preliminary Results of a Survey on a Slovenian Sample Alenka Baggia, Mirjana Kljajić Borštnar, Andreja Pucihar, Danica Dolničar, Tomaž Bartol, Andrej <1st Author Name>, <2nd Author Name> Šorgo, Bojana Boh in cooperation with Saša Aleksej Glažar, Vesna Ferk Savec, Mojca Juriševič, Blaž Rodič, Irena Sajovic, Margareta Vrtačnik Workshop Outline In the past decade education sector has faced tectonic changes. Not only interactions among stakeholders in the education processes have changed, new technologies, methods of teaching, learning and new business models are being explored. From the pedagogy perspective past years were devoted to researching virtual learning environments, social media, gamification etc. Expectations that new technologies and approaches would yield better results have proven to be overrated for both students and faculty. Furthermore, universities and students are faced with virtual competitors, pressure to provide students with excellent knowledge, ready-to-use competencies and entrepreneurial skills. The pressure to shift from university – a promotor of new knowledge, progress and innovation to university – a corporation training centres is enormous. Is “university as we know it” on the verge of extinction? The panellist will seek answers to questions of how to approach education that would yield deep knowledge and understanding of concepts, and at the same time provide practical skills and competencies? What are the roles of university, industry, government and students? How do we adjust and reinvent methodologies to support different learning goals and expected outcomes? How do we adjust organizational structures and processes, define new business models that will work in the Future University? What are the roles and responsibilities of the stakeholders in this new environment? With speakers coming from university, industry and students we will try to assess the current state of university and pave the way for the Future University. 2 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Modelling the European Data Economy: ”Data Re-use as a Model for Europe” Chair Daniel Bachlechner Senior Researcher Fraunhofer ISI, Germany Co-chair Sven Abels CEO ASCORA GmbH Germany Co-chair Helena Uršič Researcher Leiden University, Centre for Law and Digital Technologies The Netherlands 1. What is EuDEco? EuDEco (Modelling the European Data Economy) is a Coordination & Support Action receiving funding under the European Union’s Horizon 2020 programme. EuDEco’s aim is to assist the European science and industry in understanding and exploiting the potentials of data reuse in the context of big and open data. EuDEco partners believe that a self-sustaining data market would importantly increase the competitiveness of Europe. To be able to extract the benefits of data reuse to build useful applications and services, it is first critical to assess the underlying economic, societal, legal, and technological framework conditions and related challenges. Building on a thorough understanding of the framework conditions and by analyzing viable use cases and business patterns, EuDEco’s final goal is to deliver a model of the data economy. The model should include suggestions and recommendations addressing the main legal, contractual, societal and technological concerns and challenges in the EU. 2. Why this topic should be addressed at eConference in Bled? Use cases and business models play a crucial part of our research. As EuDEco partners believe our analysis can only add value if it is based on real-life examples, we are looking for user experiences from different sectors, industries and member states to help us validate and refine the model. eConference is an excellent opportunity to get in touch with various players in the field and to discuss some obstacles they have come across on the way towards data re-use economy. In line with that, the idea of EuDEco’s workshop is to establish a platform for participants to share their knowledge and experience on data re-use. EuDEco would facilitate the discussion by explaining the trends, stating our opinion and bringing up certain aspects the users have not addressed yet. 3. Proposed agenda  Modeling data re-use for the EU (5 minutes) A short description of the project and our research objectives  A short panel discussion about three perspectives of data re-use (30 minutes) Helena Uršič would take the role of a moderator. She would briefly present the panelists and lead the discussion, making sure all the main issues are addressed at the beginning and the audience is given the opportunity to interact with the panelist, if necessary.1 Technological perspective – Dr. Sven Abels (8 minutes): Within Big Data the units of measure is changing rapidly. Amount of data was measured in Gigabytes (GB) not too long ago and is now measured in Petabytes (PB) or even Zettabytes (ZB). Considering variety, information is also beginning to look differently from the way it did decades ago, since traditional IT has grown up in a database-centric operating model, where data fits neatly into rows and columns. It is valuable to point out that as outstanding as the challenges are the opportunities associated with this amount of data. Organizations worldwide are facing explosive information growth as new information is created and handled. New information access methods bring new challenges and require new tools and solutions for use, management and protection of these increased volumes. The barrage of new information drives new decisions that pertain to IT infrastructure needs, for instance, primary storage and backup and recovery. It also calls for the use of optimization 1 If there are external speakers involved in the panel, they could participate by expressing their opinion on the Eudeco initiative and elaborating how the EU could benefit from a streamlined data re-use model etc. 2 technologies such as compression, data deduplication, deep infrastructure integration and flexible methods for moving data efficiently through the network. Socio-economical perspective – Dr. Daniel Bachlechner (8 minutes): As of today, Big Data can be considered a buzzword and is often used to describe the mass collection and analysis of data. Companies such as Google and Facebook are heavily making use of this strategy, e.g. to place ads or to learn about the behavior of their users. The EU is slow in embracing the data revolution compared to the USA. According to the EC (EC, Communication on data-driven economy, 2014), “Big data technology and services are expected to grow worldwide to US$16.9 billion in 2015 at a compound annual growth rate of 40% – about seven times that of the information and communications technology (ICT) market overall.” The number of specialist big data staff working will increase significantly over the next years. While the Commission clearly acknowledges the importance of big data, European companies are fighting some great obstacles. They are facing entry barriers such as the complexity of the current legal environment and insufficient access to large datasets and enabling infrastructure. Legal perspective – Helena Uršič (8 minutes): The opportunities and challenges related to the emerging Big Data Economy are regulated to at least some extent. Not everything is allowed, obviously, but the question is whether the existing legal frameworks sufficiently enable all potential of the Big Data Economy and sufficiently address potential risks and negative side effects. When technological developments and socio-economic arguments call for changes in the existing legal frameworks, the question becomes which legal framework is desirable from an ethical, social and economic perspective. Legal aspects can be divided into two major topics, i.e., data related legal frameworks (personal data protection, IP-law, database law and liability/tort law) and human rights related legal frameworks (privacy, equal treatment, freedom of expression, human dignity, etc.). These areas come together closest in the domain of privacy and personal data protection, yielding this domain the most important within this project.  Parallel discussions in smaller groups After the introductory presentations, the panelists and participants/audience would split into three groups, each of them focused on one perspective. The groups would discuss the challenges mentioned in the first part with the focus on real-life examples of data re-use. The EuDEco representatives would act as facilitators of three perspective- specific groups discussions, encouraging the participants to consider different aspects and provide a joint statement that would be shared with two other groups in the closing part.  Closing The workshop would close with comments/statements of the EuDEco representatives/participants about the results of the group discussions and some final remarks. 3. Outcomes/goals 3 This workshop has two goals. First, it aims to raise the awareness of complexity of data re-use and its technological, socio-economical and legal aspects. In this purpose it facilitates the discussion and encourages the workshop participants to express their opinion on data re-use. Secondly, it strives for practical examples, best practices and user experiences to help participants as well as EuDEco team validate and reconcile their understanding of data re-use in the EU. 4 28th Bled eConference #eWellBeing June 7 - 105, 2015; Bled, Slovenia Design and Application of a Methodology for Unstructured Data Automated Analysis in Word-of- Mouth Lucie Šperková University of Economics in Prague, the Czech Republic lucie.sperkova@vse.cz Abstract The proposed dissertation research is focused on the automated analysis of unstructured data for marketing purposes. Specifically, the objectives will focus on developing, and the application of methodology for modelling unstructured data, their sentiment in specific marketing methods e-Word-of-Mouth, and their combination with other data from collaborative CRM. It is a specific use of data / text / opinion mining tools for the analysis of unstructured content, their application to marketing research, as well as possibly to Business, Customer and Competitive Intelligence. By their use an increase in performance, efficiency, marketability, return of investment and better negotiations with customers is expected. For these activities data from web 2.0, call centres of companies and other in-house data will be used. One of the reasons for using more data sources is a multichannel environment through which customers interact with the company and with which the companies had previously worked fragmented. Keywords: Unstructured Data, Word-of-Mouth, Sentiment Analysis, Opinion-mining, Voice-of-Customer 1 Introduction We live in a period of socio-economic changes, where the most influential factor is the Internet. Web 2.0 has changed the Internet paradigm since it involves the user as a creator of content, which he helps to organize, share, change and criticize. Web 2.0 offers new business approaches and supports knowledge management. (Choi and Scott, 2013) Amongst the most used services are social media including social networks, but also online forums for acquisition and aggregation of customer reviews. There customers share their opinions, references, feedbacks and other observations, the so- called Word-of-Mouth (WoM). It is therefore a logical step to monitor these actions and process obtained information for further use. Data generated from Internet communication contain huge potential for marketing research and knowledge acquisition. Modern marketing techniques include the use of information technology, CRM and online marketing that seeks to replace the techniques performed manually by 1 Lucie Šperková observation, inquiry, experimentation or other sophisticated marketing campaigns. Opinion mining techniques deal with customers' attitudes, opinions, moods and emotions toward the brand, product or competitors that share on social networks, forums, blogs, product reviews etc. (Liu, 2012; Pang and Lee, 2008; Kaur, 2013). This process should be, however, not accidental, but supported by a methodology that will clearly show how to deal with such data from collection to evaluation. If the online data are expanded by additional data, which the customer shares with company voluntarily through e-mail messages or calls to call centres, or other available sources about competition, suppliers, partners, industry trends, etc., a large database is available for automated marketing research methods, which integrates everything in one place and provides a unified view of the problems with the possibility of deeper analysis and evaluation of the success of the methods. Here, companies can unlock the potential to increase their competitiveness in the market. 2 Definition of the subject area Companies are generally aware of the existence of unstructured data and their significance for the business. There are already conferences organized on this topic which trying to grab this concept, e.g. Gartner Gartner Business Intelligence & Analytics Summit Barcelona, 2013, TDWI Konferenz München, 2013, Teradata Universe Prague, 2014, Gartner BI Summit London, 2013. However, they do not yet have settled rules how to work with these data and in particular how to use them to obtain a higher market share. The whole concept can be effective only if it presents a clear and consistent picture of what is happening in and around the company. If companies use automated analyses of unstructured data, they are able reduce cost and time-consuming manual and comprehensive analyses conducted by people like reading posts and search links in them. Analysis of unstructured data regarding the customer can help find the priority clients, problems relating to products, customer sentiment, difficulties in servicing services, find the next best step in business, identify activities and customers of competitors, their reactions, etc. Large IT companies, who offer different solutions for unstructured data analysis, face various challenges in their businesses. As I experienced in practice, the biggest problem that they face is the inability to sell this software. During my practice, it became clear that companies have a lot of data, but do not know exactly what to analyse. Large commercial solution integrate many modules and techniques promising huge opportunities, but the lack of any methodology and case studies that clearly show "how to transform this data (in a very broad sense of the term) into actionable knowledge which can be used to follow a particular goal." (Gopal et al., 2011) Given that marketing spends a large number of financial budgets of companies, the creation of a consistent methodology for Word-of-Mouth and its combination with other data, e.g. from calls to a call centre, is more than suitable. In addition, the need for new methods of marketing is influenced by other factors, such as customer loyalty, satisfaction and customer-oriented approach. With the growth of unstructured data and social networks also social marketing is actual topic. It is supported by some Customer Relationship Management (CRM) systems. The most common method used by companies is social network monitoring. Generally, 2 Design and Application of a Methodology for Unstructured Data Automated Analysis in WoM measuring the return of investment should not satisfied with statistics such as impressions, the number of followers on the network. This data should be read in the wider context, thus analyse the text and its sentiments and relationships, and thereby examine customer loyalty, behaviour and satisfaction. Therefore marketers have come to recognize WoM’s importance and understand how to generate a coordinated, consistent response that reaches the right people, at the right time, with the right content and setting. The value of WoM was identified and put into practice at the time when there was no internet and has been an important scientific subject over 50 years, e.g. (Helm and Schlei, 1998; Braun and Reingen, 1987). Today it mainly deals with the so-called Voice of Customer, which is also discussed in research of (Griffin and Hauser, 1993; Evangelopoulos and Visinescu, 2012; Peng et al., 2012). The Internet originated a new concept, a kind of classical innovation of WoM, e-Word-of-Mouth, which has received considerable attention in both various enterprises and academic circles (Choi and Scott, 2013; Han and Niu, 2012; Dellarocas et al., 2007) and more. Also, within the context of the past and current financial crisis and continuously emerging social network sites, it can be argued that there is a need to redefine models of eWoM communications, in order to more effectively target stakeholders so as to better serve the needs of managers, entrepreneurs, firms and society. 3 Preliminary definition of the objectives of the dissertation The above results of the analysis of the current state of research in the field of marketing methods and existing models for the analysis of unstructured data and their sentiment confirm the sense of the application of methodology for unstructured data automated analysis in WoM and their combination with analysis of data from collaborative CRM. The thesis is intended to be interdisciplinary. It combines the aspect of the public websites with the in-house aspect. On this basis, it was determined also the main aim of the thesis: Creation of a methodology for unstructured data analysis to tackle marketing methods Word-of-Mouth and its combination with analysis of data from the sources of collaborative CRM. This includes objectives defined in chapter 4 containing the proposal of methods for achieving them. Key research questions The current state of research on the subject defines the following key scientific questions for the application of unstructured data analysis into existing marketing methods of WoM. The research questions will be refined after the full literature review and baseline study from the following: RQ1: Can we significantly streamline marketing research methods using WoM by analysis of unstructured data? RQ2: Is there clearly given a uniform methodology and chosen models for automated analysis of unstructured data in the WoM use? 3 Lucie Šperková RQ3: How can be these analyses of WoM combined with the data from the Collaborative CRM? RQ4: How to methodically replace the current marketing methods of WoM by analysis of unstructured data and combine them with other analysis of data from the CRM systems? RQ5: What are the requirements for the selection of models and creation of methodology for combination of WoM with analysis of data from collaborative CRM? RQ6: Have proposed models and methodology increase competitiveness, efficiency of management processes and negotiations with customers and ROI? 4 Proposed methods for achieving the objectives of the dissertation Methods of achieving the objectives of the proposal are based on the commonly used methods of scientific work, (Molnar et al., 2012). Several approaches will be used, due to the interdisciplinary character of the thesis. Research will be performed through applied research due to the fact that social marketing is a social science in which it will be applied informatic approach. 4.1 Collection and analysis of theoretical data Collection and analysis of theoretical data is performed as based on exploratory research, i.e. on the collection, classification and description, and then synthesis of information. Based on the research I identified a compilation of a prior research. A method of analysis is used to analyse: - Current WoM marketing techniques, methodologies and their use in practice. - Specific factors affecting WoM. - Current methods for analysing unstructured data in collaborative CRM. - Representatives offering tools for analysing unstructured data and their sentiment. - Suitable Algorithms, methods or models applicable to final methodology. The method of analysis is followed by a synthesis method that is used to: - Definition of requirements for the selection of models and methodology for automated analysis of unstructured data and their sentiment in WoM and analysis of data from CRM. - Evaluate the applicability of current methods and approaches from third parties. - Selection of appropriate marketing techniques and methodologies for the application of automated analysis of unstructured data into WoM and their combination with data from CRM. - Selection of models that will be used for the analysis of unstructured data and sentiment. 4 Design and Application of a Methodology for Unstructured Data Automated Analysis in WoM - The creation of the new methodology that will support the modelling of unstructured data to tackle marketing methods of WoM and its combination with analysis of data from CRM. 4.2 Empirical survey A survey based on empirical methods (observation, in-depth interview) among marketing professionals engaged in WoM in selected agencies / companies and companies offering ready-made solutions based on monitoring and analysis of unstructured data will serve as a method for determining the level of practical solutions of marketing models and methods for WoM. As part of the evaluation of this survey I will evaluate: - Whether is possible to use the current marketing research methods of WoM as a template for a set of methodology for the analysis of unstructured data. - Information about the level and practicality of offered tools, technologies and products to address the issue. - How the current solution of unstructured data analysis for marketing purposes may be useful for the formation of uniform methodology. - How to properly combine the analysis of data from a public site with analysis of internal data from CRM. 4.3 Methodology compilation To achieve the objectives it will be necessary to choose appropriate models and construct a methodology for unstructured data in WoM. The input will be a well described WoM marketing methodology from a previous exploration phase (descriptive approach) and CRISP-DM methodology used in data-mining. 2-3 opinion-mining models based on Machine Learning algorithms will be selected. Use of empirical learning and inductive inference on the acquired data will result in a general description of the concept, which will be methodically described and formalized. The methodology will include a technological part and also a process part for application in marketing management both on the internet and in the CRM environment. For the selection of models and a set of methodology large amounts of data in unstructured form is needed. On the data analysis and simulation will be performed. 4.4 Applications and theoretical and practical validation of defined and compiled methodology Compiled methodology will need to be simulated on the real data. It will use in particular a what-if analysis distinguishing the impact of different input situations. Case study on concrete data about customers, products, and other data available from marketing, both from social networks and CRM will be carried out. Full paper about this case study should verify by the truth-value, (re)applicability, consistency and neutrality of the methodology. Case study will use the methods suggested by (Yin, 2013). Case studies should also show the impact of an unstructured data analysis to negotiation with the customers. Analysis of unstructured data should increase the effectiveness of 5 Lucie Šperková process management and negotiations with customers, learning about their needs and providing information to the possibility of their broader satisfaction. Thus become a source of financial benefit, and increase the competitiveness of companies in the market. To determine this, I will use empirical methods, statistical survey, qualitative research, comparison of the original marketing approaches to the compiled case study, interviewing the persons responsible in marketing. Methodology should be authorized by responsible persons in marketing and incorporated as a formalized methodology for marketing research. 5 Preliminary results In the first half of 2014 was conducted a project with Hewlett and Packard and their solution Autonomy for unstructured data analysis. The pilot project involved analysing unstructured data from Facebook public pages of Czech bank institutions. This research led to two full papers, one for the Marketing Identity conference in Slovakia (Šperková, 2014a) and in the Czech reviewed journal Acta Informatica Pragensia (Šperková, 2014b). Articles have been published with the support of internal grant called Innovative Research that described the state of banking institutions in the Czech Republic and their approach to analysing texts from social networks. Research confirmed my hypothesis that Czech financial institutions have no concept how to approach such analysis. The paper suggests a simple methodology how to perform these analyses, but not yet in an automated form. The paper also emphasizes the importance of WoM and its analysis by information technology. It discusses the role of sentiment analysis in WoM and other Machine Learning method clustering. The results show that it has a meaning to analyse customers' feedback and contributions and look for some patterns or search at individual level. People use networks of banking institutions relatively actively in order to solve problems, because in their opinion banks are not able to solve them through other marketing channels such as call centres or a personal visit to the branch. If companies are able to use data from internet and combine them with data from CRM, they will be able to communicate with customers faster and more accurately on personal level. 6 Future development In order to expand the expediency of current literature one strand of future research will include a broader analysis of social media / review sites / blogs practice among companies in order to establish how this may be integrated into the collaborative CRM and the data mixed together. An analysis of views and experiences from industry experts is also proposed, as well as a more detailed case study from another company, specifically in terms of eWoM from other channels than internet, as well as how they may be used to meet strategic objectives, as compared with traditional marketing methods. Next step will be further research conducted in cooperation with a Czech insurance company in order to analyse their calls to the call centre. However, the transliteration of the calls for further analysis will be problematic. This case study will be the first step for the compilation of a methodology which will also include in-house data. 6 Design and Application of a Methodology for Unstructured Data Automated Analysis in WoM The cooperation with digital marketing experts is also in process. The aim is to gain a well described current methodology of WoM approaches conducted in marketing nowadays. Acknowledgement This paper was prepared thanks to the IGA grant VSE IGS F4/18/2014. References Brown, Jacqueline J. and Peter H. Reingen. (1987) Social Ties and Word-of-Mouth Referral Behavior. Journal of Consumer Research, 14 (December), 350-362. Choi, J. H., & Scott, J. E. (2013). Electronic word of mouth and knowledge sharing on social network sites: a social capital perspective. Journal of theoretical and applied electronic commerce research 8(1), 69-82. Dellarocas, C., Zhang, X. M., & Awad, N. F. (2007). Exploring the value of online product reviews in forecasting sales: The case of motion pictures. Journal of Interactive marketing, 21(4), 23-45. Evangelopoulos, N., & Visinescu, L. (2012). Text-mining the voice of the people. Communications of the ACM, 55(2), 62-69. Gopal, R., Marsden, J. R., & Vanthienen, J. (2011). Information mining—Reflections on recent advancements and the road ahead in data, text, and media mining. Decision Support Systems, 51(4), 727-731. Griffin, A., & Hauser, J. R. (1993). The voice of the customer. Marketing science, 12(1), 1-27. Han, X., & Niu, L. (2012). Word of mouth propagation in online social networks. Journal of Networks, 7(10), 1670-1676. Helm, S., & Schlei, J. (1998). Referral potential-potential referrals. An investigation into customers' communication in service markets. Track 1 Market Relationships. In Proceedings 27th EMAC Conference, Marketing Research and Practice (pp. 41-56). Kaur, H. (2013). Opinion Mining Task and Techniques: A Survey. International Journal of Advanced Research in Computer Science, 4(3). Liu, B. (2012). Sentiment analysis and opinion mining. Synthesis Lectures on Human Language Technologies, 5(1), 1-167. Molnár, Zdeněk et al. (2012) Pokročilé metody vědecké práce. Profess Consulting. Pang, B., & Lee, L. (2008). Opinion mining and sentiment analysis. Foundations and trends in information retrieval, 2(1-2), 1-135. Peng, W., Sun, T., Revankar, S., & Li, T. (2012). Mining the “voice of the customer” for business prioritization. ACM Transactions on Intelligent Systems and Technology (TIST), 3(2), 38. Šperková, Lucie. (2014) Analýza nestrukturovaných dat z bankovních stránek na sociální síti Facebook. Acta Informatica Pragensia, 3(2), 154-167. Sperkova, Lucie. (2014) Word of Mouth Analysis on Facebook in Banking. In Marketing Identity: Explózia inovácií. Trnava. Yin, R. K. (2013). Case study research: Design and methods. Sage publications. 7 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Analytics on Feedback Creation: The Application of Learning Analytics on Formative Assessments Justian Knobbout MSc Utrecht University of Applied Sciences, the Netherlands justian.knobbout@hu.nl Abstract Learning analytics’ main objective is optimizing learning by analysing learners’ data from various sources. Although analytics is applied within educational institutes for some time now, much attention is paid to improving processes at institutional level rather than at micro level – at which the actual learning happens. One way of shifting this focus is by the application of learning analytics on formative assessments. This type of assessments provides feedback to learners during the course of their learning process. This paper provides an brief overview of research on learning and formative assessments, as well as a methodology for research within this area. Moreover, the author describes a case-study in which the effects of formative testing on the predictability of final grades are researched. Keywords: Learning Analytics, Formative Assessment 1 Introduction In recent years, the availability of big data led to the (further) development of a variety of analytical techniques and technologies, for instance, data mining, cluster analysis and machine learning (Maltby, nd). Nowadays, the analysis of big data is everyday practice – at least in certain industries. In other industries, for example higher educational institutes (HEIs), big data analytics is yet an upcoming topic of interest. Data usages within HEIs has been quite inefficient in the past, however, this is likely to change (Siemens & Long, 2011). 8 Analytics on Feedback Creation The analytical spectrum within educational institutes consist of three main areas. That is, 1) educational data mining, 2) academic analytics and 3) learning analytics (Siemens, 2011). Whereas the former primarily focuses on deductive measures and seeks patterns in historic data, the latter two are more oriented towards understanding and explaining what is happening within educational systems and processes. Academic analytics can be compared with business intelligence and takes place at institutional, national, and international level (Siemens, 2011). Learning analytics takes place at course- and departmental level, thereby benefiting learners and lecturers. Buckingham Shum (2012) makes a similar differentiates between analytics at macro, meso, and micro level. A now widely used definition of learning analytics is provided during the 1st International Conference on Learning Analytics and Knowledge (LAK), namely: “the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimizing learning and the environment in which it occurs.” (Ferguson, 2012). Another definition is given by Duvel (2012), who states that learning analytics is about “collecting traces that learners leave behind and using those traces to improve learning”. Both definitions stress the importance of improving learning from the learners’ perspective. Learning analytics can be used for different purposes, that is, prediction and reflection (Greller & Drachsler, 2012). Predictive models can be used to model student activities and provide early interventions. This can lead to the development of early-warning- systems, as described by e.g. MacFadyen and Dawson (2009), who designed a system based on data from the Learning Management System (LMS) in order to identify students at risk of failing a course. Reflection, on the other hand, can lead to improved monitoring of learning processes and offer students more personalized information on their progress (Greller & Drachsler, 2012). Both ways, researchers apply data and analytics “as a means to become better teachers and help learners become better learners” (Siemens & Gasevic, 2012). SURF is a Dutch organization supporting research within educational institutes. From 2012 on, SURF initiates researches related to learning analytics. Based on the results of seven grassroots projects, an ideal cycle of learning analytics was formulated (SURF, 9 Justian Knobbout 2013), see Figure 1. First, the purpose of the data collection must be decided upon as well as which data is required and the data source. Measuring and analyzing data without a well-defined goal is not effective and should be prevented. During the second phase of the cycle the data is collected and stored. This can already provide points-of- interest to focus on in the next phase; data analysis and visualization. Statistical tests are performed in order to test the hypotheses posed at the beginning of the analytics cycle. Finally, an intervention, based on the results of the analysis, takes place. The effects of the intervention can be measured by continuing the cycle, starting again from the beginning. Figure 1: Learning Analytics Cycle (SURF, 2013). 2 Problem Definition The primary goal of educational institutes is to educate learners. All analytic activities within these institutes should therefore support this goal. However, “there are very few compelling examples of analytics being used at scale to benefit learners” (MacNeill, Campbell & Hawksey, 2014). Until recently, much attention is paid to processes at institutional, i.e. meso level. Although this might benefit learners in an indirect way, it seems more effective to engage in activities which help students perform better directly, that is, during the courses they follow momentarily instead of a later phase of their studies. 10 Analytics on Feedback Creation Cross and Angelo (1988) elaborate on different assessment techniques which can be applied within traditional college classrooms in order to gauge various types of students’ skills and competencies. The purpose of these techniques is to learn about student learning and improve teaching rather than grading. With summative tests, students receive feedback – often in the form of a grade – after they finished their course. The authors, however, suggest assessments are used to provide formative feedback (p. 23). This helps students and teachers to define goals and assess progress toward them. By doing so during the course, it is possible to change strategies and make changes based on given feedback. As stated by Kizilcec, Piech and Schneider (2013), “frequent, formative testing enable learners to reflect on their knowledge state and actively retrieve information in a way that facilitates learning”. Research shows that formative assessments lead to significant learning gains (Black & Wiliam, 1998: p. 3). In order to be efficient, feedback must meet three conditions (Sadler, 1989); the learner must (1) have a concept of the standard being aimed for, (2) compare the current level of performance with this standard, and (3) engage in the right action to close the gap between them. In recent literature, there is an increase in interest in the application of leaning analytics to formative assessment (Tempelaar et al., 2014; Tempelaar, Rienties & Giesbers, 2015). The researchers found that computer-assisted formative testing might be a better predictor to identify underperforming students and academic performance than basic Learning Management System data. This requires, however, for data to be quickly available. The use of computers and digital testing is therefore paramount. Second best sources are e-tutorial systems, entry tests, and prior education data. The use of learning analytics can support the feedback objectives stated by Sadler (1989). That is, analytics can help to measure the current performance of learners and compare this to the group of learners they belong. Moreover, they can identify activities which support their learning process and activities which do not. This can help to plan the actions required to meet the desired learning outcome. From a lecturer’s point of view, analytics provide insight in the performance of the entire group, help to distinguish between effective and efficient learning activities, and identify students who 11 Justian Knobbout are at risk of failing the summative assessment. This, in turn, should lead to beneficial interventions. See Figure 2 for a schematic view on the relationship between the three feedback objectives and potential learning analytic-supported areas. Figure 2: Schematic View on Relationship Between Feedback Objectives and Learning Analytics. In order to further comprehend the effects of learning analytics on formative testing, an extensive research will be conducted. The research takes four years and forms the researcher’s PhD project. The objective is to provide an answer to the following research question: “In what way can learning analytics be applied and configured in order to analyze and improve formative assessment processes within higher educational institutions?” 3 Methodology The units of analysis in this research are learners and lecturers. This follows the definition of learning analytics as defined by Siemens (2011) and Buckingham Shum (2012). The research takes a deductive approach, of which Bhattacherjee (2012) provides a generalized process incorporating three phases, that is; (1) exploration, (2) research design, and (3) research execution (see Figure 3). During the explorative phase, research questions are further explored, a literature review is conducted, and relevant theories are 12 Analytics on Feedback Creation identified. Some proof-of-concept will be held at a case organization in order to show the practical applicability of learning analytics on formative testing. The literature review will be performed using Kitchenham’s (2004) method to systematic review literature. The second phase relates to the selection of a research design, which provides a framework regarding the collection and analysis of data (Bryman & Bell, 2007). Moreover, the research design deals with the research method and sampling strategy (Bhattacherjee, 2012). The result of these two phases are a research proposal, in this case a PhD project proposal which will be filed to the Graduate School of Open University Heerlen, the Netherlands. The research is executed during the third phase of the process. The activities performed during this phase depend on the chosen research design. A comparative design might be useful to draw distinctions between different cases, e.g. courses, departments or institutes. On the other hand, an experimental design might be relevant when analyzing the effects of an intervention, e.g. the use of a formative assessment application, on the results of a treatment group, which in turn are compared to the results of a control group. Figure 3: Generalized Research Process (Bhattacherjee, 2012) 4 Preliminary and expected results This section describes the preliminary as well as the expected results. 13 Justian Knobbout 4.1 Preliminary results To position learning analytics within the domain of formative testing, a small case-study was conducted. The researcher, who is a part-time lecturer as well, analyzed data from one of the courses given. Goal of the course Entrepreneurship - part of the minor Technical Commercial Engineering, followed by 3rd and 4th year students from various studies - is to write a business plan to set up a business in order to commercialize a product or service of the student’s choice. Several formative assessments were given during the ten weeks of the course’s duration. One assessment was to hand in a concept plan, four week in advance of the final deadline. This way, there was plenty of time for the lecturer and students to provide feedback and make changes to the plan, respectively. Of the 23 students, 15 (65%) did hand in a concept, eight (35%) did not. Of the latter group, six students (75%) failed the course, against two (13%) of the group who sent their concept plan. Using a Fisher’s Exact Test1, it is calculated that there is a significant ( p = .006) effect between handing in a plan and passing the course. By applying this knowledge in due courses, students who do not send a concept can be warned that they are at risk of failing, hopefully making them realize there is work to do in order to avoid a low grade. Moreover, at the end of the course an online questionnaire was provided to the same group of minor students ( n = 23), of which 17 filled it out; response-rate is therefore 74%. The majority of respondents (82%) thought it was useful to work on assessments whilst the course was still running. Even a bigger group though (94%) said they expected feedback on each assessment given. The case-study shows that learning analytics can be applied on formative testing, thereby identifying important assessments and students-at-risk. Students acknowledge the usefulness of it as well, as long as they receive feedback on their input. 4.2 Expected results The intended results of the research are: (1) an overview of state-of-the-art research on learning analytics as well as digital and formative assessments, (2) a conceptual model 1 Normally, a Pearson Chi-Square test would be used to analyze the effect of two nominal variables but as the expected results of one of the cells is less than five, a Fisher’s Exact Test is used (Weaver, 2013). 14 Analytics on Feedback Creation on learning analytics and formative testing and, (3) an extended study on the effects of formative testing on students’ results in order to both design and validate the model. 5 Future development The research will elaborate on the application of learning analytics on formative assessment processes within higher educational institutes. Comparative case studies and experiments will provide answers to the question in what way learning analytics can effectively be applied to formative testing. The focal unit of this research will be learners and their lecturers. Van Leeuwen et al. (2014) propose to investigate whether there are differences in benefits of supporting tools between different types of teachers. This suggestion provides a different view on learning analytics, which often focusses on - various types of - learners, whereas teacher characteristics are consider to be static and uniform. Another research suggestion is provided by Tempelaar et al. (2015), who propose to investigate the best way to “present feedback based on learning disposition data in combination with technology-generated data to students”. This presses the issue that the generation of feedback alone is not the only thing which matters; how to provide the feedback to students is another important aspect. References Bhattacherjee, A. (2012). Social science research: principles, methods, and practices. Black, P., & Wiliam, D. (1998) . Inside the black box: Raising standards through classroom assessment. Granada Learning. Bryman, A., & Bell, E. (2007). Business Research Methods 2e. Oxford university press. Cross, K. P., & Angelo, T. A. (1988). Classroom Assessment Techniques. A Handbook for Faculty. Duvel, E. (2012). Learning Analytics and Educational Data Mining. Retrieved from https://erikduval.wordpress.com/2012/01/30/learning-analytics-and-educational- data-mining/ on 1 February 2015 Ferguson, R. (2012). Learning analytics: drivers, developments and challenges. International Journal of Technology Enhanced Learning, 4(5), 304-317. Greller, W., & Drachsler, H. (2012). Translating learning into numbers: A generic framework for learning analytics. 15 Justian Knobbout Kizilcec, R. F., Piech, C., & Schneider, E. (2013, April). Deconstructing disengagement: analyzing learner subpopulations in massive open online courses. In Proceedings of the third international conference on learning analytics and knowledge (pp. 170-179). ACM. Macfadyen, L. P., & Dawson, S. (2010). Mining LMS data to develop an “early warning system” for educators: A proof of concept. Computers & Education, 54(2), 588-599. MacNeill, S., Campbell, L. M., & Hawksey, M. (2014). Analytics for Education. Reusing Online Resources: Learning in Open Networks for Work, Life and Education, 154. Maltby, D. Big Data Analytics. Sadler, D. R. (1989). Formative assessment and the design of instructional systems. Instructional science, 18(2), 119-144. Siemens, G. (2011). Learning and Academic Analytics. Retrieved from http://www.learninganalytics.net/?p=131 on 29 October 2014. Siemens, G., & Gasevic, D. (2012). Guest Editorial-Learning and Knowledge Analytics. Educational Technology & Society, 15(3), 1-2. Siemens, G., & Long, P. (2011). Penetrating the fog: Analytics in learning and education. Educause Review, 46(5), 30-32. SURF (2013). Learning Analytics in het hoger onderwijs: Mogelijkheden en aandachtspunten. Tempelaar, D. T., Heck, A., Cuypers, H., van der Kooij, H., & van de Vrie, E. (2013, April). Formative assessment and learning analytics. In Proceedings of the Third International Conference on Learning Analytics and Knowledge (pp. 205-209). ACM. Tempelaar, D. T., Rienties, B., & Giesbers, B. (2014). In search for the most informative data for feedback generation: Learning analytics in a data-rich context. Computers in Human Behavior. Van Leeuwen, A., Janssen, J., Erkens, G., & Brekelmans, M. (2014). Supporting teachers in guiding collaborating students: Effects of learning analytics in CSCL. Computers & Education, 79, 28-39. Weaver, B. (2013). Assumptions/Restrictions for Chi-square Tests on Contingency 16 Analytics on Feedback Creation Tables. Retrieved from https://sites.google.com/a/lakeheadu.ca/bweaver/Home/statistics/notes/chisqr_as sumptions on 29 January 2015 17 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia A conceptual framework for service system interaction Buddhi Pathak Henley Business School, University of Reading, UK B.N.Pathak@pgr.reading.ac.uk Abstract There is a growing research concern on how service ecosystems form and interact. This research thus aims to explore the service ecosystem formation and interaction as well as its underlying nature of value co-creation. This work develops an initial conceptual framework for assessing service system interactions that includes the various stages of value co-creation within ecosystem context. How the conceptual framework will further be developed and future plan are also presented. Keywords: Service systems, Value co-creation, S-D logic, Service Science 1 Introduction The discussion on how value is created and appropriated in our society is once again gaining the wider attention in academia and industry after the introduction of Service- dominant logic (SDL) in 2004 and, also the emerging concepts of Service Science (SS) (short for Service Science, Management, Engineering and Design, also known as SSMED). SDL (Vargo and Lusch, 2004; 2008) is a theoretical approach that describes paradigm shift from goods-dominant logic (manufacturing) to service-dominant logic. SS is an interdisciplinary domain to study, improve, create and innovate in service to understand how service systems interact and co-create value (Spohrer and Kwan, 2008; Vargo et al., 2010). The traditional approach of one way delivery of value (e.g. Porter, 1985) was initially challenged by Moore in 1993 where he proposed the idea of business ecosystem- a metaphor that originated from ecology. His argument is that firm operates between upstream (e.g. supplier) and downstream (e.g. customer) and it also make use of government agencies and other stakeholders, therefore, a firm’s value creation capability is not linear but affected by all of its interaction with its community, which he termed it as a firm’s ecosystem. This concept of business ecosystem is termed as service 18 A conceptual framework for service system interaction ecosystem in SDL (Vargo and Lusch, 2011), and in Service Science it is defined as service system (Spohrer and Kwan, 2008). The discussion of value co-creation in service ecosystem (Vargo and Lusch, 2011; Akaka et al., 2012; Edvardsson et al., 2012; Autio and Thomas, 2014) stresses ecosystem as spontaneously sensing and responding (Vargo and Lusch, 2011). According to Lusch and Nambisan (2015; p.162) service ecosystem is ‘a relatively self- contained, self-adjusting system of mostly loosely coupled social economic actors connected by shared institutional logics and mutual value creation through service exchange.’ Although, there are some research (e.g. Adner and Kapoor, 2010) that illustrates the nature of value co-creation in ecosystem, the literature on business ecosystem, or ecosystem henceforth, is not well developed with regards to formation of ecosystem (Autio and Thomas, 2014), the nature of value co-creation in ecosystem (Vargo and Lusch, 2011) and factors that enable and disable value creating activities in ecosystems (Akaka et al., 2012; Edvardsson et al., 2012). Therefore, this work aims to identify research opportunity in this area. The rest of the paper is organised as follows: in the problem definition section (section 2), the research problem as well as the research questions are identified; case study methods are adopted for this research and is explained in the methodology section (section 3). The conceptual framework is presented in preliminary result section (section 4). And section 5 outlines future plan. 2 Problem Definition In recent years, the growing competition amongst businesses and the need to reduce operation costs have triggered firms to look outside for innovation ideas and technologies from their suppliers, independent inventors, customer and government agencies (Chesbrough, 2003). Innovation network and ecosystems established by companies with this objective have raised growing interest for research (Ostorm et al., 2010; Nambisan, 2013; Lusch and Nambisan, 2015). Business ecosystems are important resources for competitiveness and business innovation (Moore, 1996; Adner & Kapoor, 2010; Adner et al., 2013), nonetheless, existing research lack insights into how these business (eco) system function to co-create value for stakeholders (Miraglia and Visnjic, 2012; Autio and Thomas, 2014). There are a number of studies (e.g. Tax and Stuart, 1997) that attempt to research service system, however, they are mainly driven by Goods dominant logic (Vargo and Lusch, 2004) i.e. not consideration of assessment of value in context and do not reflect the role of social forces (Edvardsson and Tronvoll, 2013) in service systems. There are some industrial and governmental initiatives such as digital ecosystem initiation in Europe but, there is a lack of theoretical and practical insights that guide firms in developing deeper understanding of their ecosystem. The mechanistic approach of service systems has been changed overtime. Service Science which considers Service dominant logic (SDL) as theoretical departure point defines service (eco) system as interaction amongst actors (e.g. people, machine, and organisation) who use resources to co-create value in a given context (Vargo et al., 19 Buddhi Pathak 2010). The central point of service ecosystem is to enhance value co-creation and increasingly the value is co-created in densely complex network of actors referred as service ecosystem (Mele et al., 2010). But it is not clear how this service ecosystem functions and value is co-created in this dense network. Similarly, there are limited scholarly works (e.g. Vargo and Akaka, 2012; Akaka et al., 2012) that illustrate the function of service ecosystem. These researches are also limited to non-empirical conceptual/analytical frameworks. Therefore, there is a need to illustrate these concepts form empirical lens. Scholars such as Ostorm et al., (2010); Miraglia and Visnjic, (2012); Edvardsson and Tronvoll, (2013) and Autio and Thomas, (2014) have called for research in the areas of service system formation, value co- creation and factors that impact vlaue co-creation in ecosystem contexts. Thus, this work aims to explore and understand how service systems are formed and how their components interact. Further, in line with Autio and Thomas (2014) this research also aims to identify factors that facilitate and inhibit value co-creation in service ecosystem. Therefore, to achieve the research aim the following research questions are proposed. Q1. How service systems are formed and how do their components interact? The objective of this is to advance undertanding of formation of service system components. The outcome of objective will be service system interaction framework. Q2. What are the factors that enable value co-creation in ecosystems? The objective of this is to identify key factors that facilitate value co-creation in ecosystem context. The outcome of this will be a number of key value enabling factors. Q3. What are the factors that inhibit value co-creation in ecosystem? The objective of this is to identify key factors that prevent value co-creation in ecosystem context. The outcome of this will be a number of key value inhibiting factors. Q4. How Information Systems is useful to reduce service ecosystem inhibiting factors? The objective of this is to explore information systems' usefulness with regards to service systems interactions and how it facilitates these interactions. The outcome can be seen as a number of recommendations to implement Information Systems in different parts of service systems. The next section looks on how the objectives could be achieved. 3 Methodology This research aims to explore how service systems are formed and interact, which are empirically underexplored, a situation that can be studied with a case study research (Yin, 2013). Eisenhardt (1989, p.534) define case study as 'a research strategy which focuses on understanding the dynamics present within single settings.' Case study research approach is not a new method to study service ecosystems. Edvardsson et al., (2012) for instance, use case study method to investigate social structure which is embedded in service systems. Similarly, the work of West and Wood (2013) also use case study to examine evolving an open ecosystem. The chosen case study should meet the criteria of service ecosystem identified by Vargo and Lusch (2011; 185), which is 'a spontaneously sensing and responding spatial and 20 A conceptual framework for service system interaction temporal structure of largely loosely coupled, value-proposing social and economic actors interacting through institutions, technology, and language to 1) co-produce service offerings, 2) engage in mutual service provision, and 3) co-create value'. Further, we use multi case studies to enhance the results which provide more compelling evidence and produce more robust conclusions than a single case study (Eisenhardt 1989). This research uses case study from IT industry and data will be collected from document analysis and interviews. From the interview and document analysis, it is expected to get a number of service system components, enabling and inhibiting factors. It is expected to get this results by Autum 2015. The result will be used to reframe initial service system framwork. It is anticipated to achieve the research objectives by answering other research questions by the end of 2016. The next section presents the initial conceptual framework will be used as a underlying concpets for the research. 4 Preliminary/Expected results Following the logic of Service Science Worldview and SDL (Spohrer and Kwan, 2008; Vargo and Lusch, 2004; 2008), we have proposed a conceptual framework (Figure 1) that describes the main components of service systems and how they interact. According to Vargo and Akaka (2012) service ecosystems are not pre-existing or fixed, it is continually being formed and reformed through the enactment of practices. Actors are key to understanding practices, who integrate their resources with other actors to co- create value in service systems (Edvardsson et al. 2012). Therefore, SDL (Vargo and Lusch, 2008) emphasizes the design of value propositions, resource integration, and the co-creation of value. Kwan (2010) and Spohrer and Kwan (2008) define service system worldview as interaction between service provider and customer which is connected by value propositions. In this logic, service experience is considered as a main motive of this overall interaction. On this perspective, we argue that our conceptual framework presented in Figure 1 explores the service system interaction process in which it shows what happens in the value proposition stage and beyond, and defines the overall experience of service provider as well as customer. In Figure 1, value proposition is a single reason why actors interact with one and other; the aim is to fulfil resource requirement for both parties involved. Value proposition is defined as a kind of shared information among the entities that forms the communications between entities (Spohrer and Kwan, 2008; Vargo et al., 2010). As service ecosystems are continually being formed through enactment of practices a central aspect of the practice is the integration of resource (Vargo and Akaka, 2012). Resource integration is defined as ‘how organisations, households and individuals integrate and transform micro-specialised competences into complex services that are demanded in the marketplace' (Vargo and Lusch, 2008: 7). Moreover, resource integration includes resources and context; resources are important in service system as they enable intended activities (Lobler, 2013; 423). Vargo and Lusch (2008) suggest that knowledge is the central basis of value creation and competitive advantage, and that knowledge and skills represent ‘operant resources’, whereas natural resources and goods or tangible resources which must be acted on to be 21 Buddhi Pathak beneficial represent ‘operand resources’. Context is listed in 10th foundational premise of SDL (Vargo and Lusch, 2008). Dey (2001; 5) defines context as ‘it is all about the whole situation relevant to an application and its set of users’. With regards to value co- creation, context denotes to the applicable aspects of a situation, which are relevant for the resource integrating activities. The co-creation of value is driven by the ability to access, adapt and integrate resources within networks of multiple stakeholders; therefore, it is dependent on the previous stage of resource integration. The final but very significant step is the realisation of benefits for both service provider and customer. After experiencing service, customers may be able to provide feedback that is obviously helpful to shape or modify new service system. Service Value provider Customer proposition Operant Resource Context & integration Operand Resource s Value co- creation Value Realisation/ Feedback Figure 1 Service system interaction 5 Future Development The conceptual framework is developed from the literature review represent the first part of the study. The next work involves testing this framework with the real practices using case study research method. Once the data is collected through interview and document analysis, the framework will be developed further based on findings. Based on this result, the research questions will begin to be answered. Some of the future work includes; the identification of case study, research on case study, document analysis, 22 A conceptual framework for service system interaction interview, further development of framework, analysing research questions based on findings, and writing thesis. References Adner, R., and Kappor, R. (2010). Value creation in innovation ecosystems; how the structure of technological interdependence affects frim performance in new technology generations. Strategic Management Journal, 31 (3), p. 306-333. Adner, R., Oxley, J. E. & Silverman, B.S. (2013). Collaboration and competition in business ecosystem, Advances in Strategic Management, 30, Emerald Group. Akaka, M.A., Vargo, S.L & Lusch, R.F. (2012). An exploration of network in value cocreation: a service-ecosystems view. Review of Marketing Research, 9: 13-50. Autio, E. & Thomas, L. D. W. (2014). Innovation ecosystems: Implication for innovation management? In Dodgson, M., Gann D.M., & Philips N. (Eds.) The Oxford Handbook of Innovation Management, [Online: Accessed on: 09/08/2014]. Available from; http://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780199694945.001.0 001/oxfordhb-9780199694945 Chesbrough, H. (2003). Open innovation: The new imperative for creating and profiting from technology. Boston: HBS Press. Dey, A.K. (2001). Understanding and using context. Personal and Ubiquitous Computing. 5 (1), p.4-7. Edvardsson, B. and Tronvoll, B., (2013). A new conceptualisation of service innovation grounded in S-D logic and service systems, International Journal of Quality and Service Science, 5 (1), pp.19-31. Edvardsson, B., Skalen, P., & Tronvoll, B., (2012). Service systems as a foundation for resource integration and value co-creation. In Special issues-Toward a better understanding of the role of value in markets and marketing. Review of Marketing Research, 9.p. 79-126. Eisenhardt, K.M. (1989). Building theories from case study research, Academy of Management Review. 14 (4), pp.532-550. Kwan, S. K. (2010). Service Systems, Value Propositions, and Service Innovation. Presented in Case Studies in Service Innovation Conference SSMEnetUK. Manchester Business School. June 14-15, Manchester, UK. Lobler, H, (2013). Service-dominant networks: An evolution from service-dominant logic perspective. Journal of Service Management, 24 (4).p. 420-434. Lusch, R.F. and Nambisan, S. (2015). Service innovation: a service dominant logic perspective. MIS Quarterly. 39 (1). p.155-175/March. Mele, C., Russo-Spena, T., & Colurcio, M., (2010). Co-creating value innovation through resource integration. International Journal of Quality and Service Sciences. 2 (1), p. 60-78. 23 Buddhi Pathak Miraglia, S. and Visnjic, I. (2012). Ecosystems value mapping and analysis, Cambridge Service Alliance, University of Cambridge. [Online: Accessed on: 24/08/14]. Available from; http://www.cambridgeservicealliance.org/research/ecosystems- value-mapping-and-services.html. Moore, J.F. (1993). Predators and Prey: A new ecology of competition. Harvard Business Review. [Online: Accessed on: 01/02/2015]. Available from; https://hbr.org/1993/05/predators-and-prey-a-new-ecology-of-competition/ar/1. Moore, J.F. (1996). The death of competition: Leadership & strategy in the age of business ecosystems. New York: Harper Business. Nambisan, S. (2013). Information technology and product/service innovation: A brief assessment and some suggestions for future research. Journal of the Association for Information Systems. Vol.14 Special Issue, pp. 215-226. Ostrom, A.L., Bitner,M.J., Brown, S.W., Burkhard, K.A., Goul, M., Smith-Daniels, V., Demirkan, H.,& Rabinovich, E. (2010). Moving forward and making a difference: Research priorities for the science of service. Journal of Service Research. 13(1), 4–36. Porter, M. E., (1985) Competitive Advantage, New York: Free Press. Spohrer, J. and Kwan, S.K. (2008). Service Science, Management, Engineering, and Design (SSMED): Outline & References. [Online: Accessed on: 01/06/14 ]. Available from: http://www.cob.sjsu.edu/ssme/ssmed.pdf . Tax, S.S. and Stuart, I. (1997). Designing and implementing new services: the challenges of integrating service systems, Journal of Retailing, 73, pp. 105-34. Vargo S.L. and Akaka M.A., (2012). Value cocreation and service systems (re) formation: A service ecosystems view. Service Science. 4(3). P.207-217. Vargo, S.L., & Lusch, R.F. (2004). Evolving to a new dominant logic for marketing. Journal of Marketing, 68(January), 1–17. Vargo, S.L., & Lusch, R.F. (2008). From goods to service(s): Divergences and convergences of logics. Industrial Marketing Management, 37(3), 254–259. Vargo, S.L., & Lusch, R.F. (2011). Its all B2B…and beyond: Toward a systems perspective of the market. Industrial Marketing Management. 40, pp. 181-187. Vargo, S.L., Lusch, R.F. & Akaka, M. A. (2010). Advancing service science with service-dominant logic; clarifications and conceptual development. In Maglio et al. (eds.), Handbook of service science: Research and Innovation in the Service Economy. New York: Springer. West, J. & Wood, D. (2013). Evolving an open ecosystem: the rose and fall of the Symbian platform. in Adner, R. Oxley, J. E. & Silverman, B.S. (eds.) Collaboration and competition in business ecosystem. Advances in Strategic Management. 30. Emerald Group. Yin, R.K. (2013). Case study research: Design and methods (5th ed.). London: Sage Publication. 24 28th Bled eConference #eWellBeing June 7 - 10, 2015; Bled, Slovenia Social CRM adoption and its influence on organizational performance - SMEs perspective Marjeta Marolt University of Maribor, Faculty of Organizational Sciences, Slovenia Marjeta.Marolt@Fov.Uni-Mb.si Abstract In the social media era customers expect from companies to engage them in their everyday activities. In order to fulfill customer requirements, organizations need to exploit new technologies, social media being one of them, particularly in the domain of customer relationship management. In line with this the research proposal focuses on investigation of factors influencing the adoption of the social customer relationship management (social CRM) and its implications on business performance. In this phase of the study we developed an initial conceptual research model that will serve as a foundation for interviews with consultants, IT providers and B2C orientated SMEs in Slovenia. We draw on theories that have been applied in the past in similar studies. Finally, we briefly present identified limitations and expected results and contribution. Keywords: social CRM, adoption, organizational performance, SMEs 1 Introduction Social media technologies were initially targeted at individuals to communicate and share content with each other. The overall usage among consumers, extensive opportunities and customer expectations (Rainie, Purcell, & Aaron, 2011) have motivated organizations to start using social media technologies for business purposes. They see potential in using social media technologies to improve customer relationship management, however many of them are still cautious in the adoption approach (Kiron, Palmer, Nguyen Phillips, & Berkman, 2013). According to Kaplan & Haenlein (2010) many organizations do not have complete vision of what social media technologies are and how it can be used effectively. Researchers have been interested in CRM technology adoption and the relationship between CRM technology and organizational performance for a long time. Some have demonstrated that CRM technologies are most effective when combined with other firm resources and CRM processes (Jayachandran, Sharma, Kaufman, & Raman, 2005; Reinhold & Alt, 2012; Trainor, Andzulis, Rapp, & Agnihotri, 2014). The emergence of social media technologies enables organizations a new way of interactions with their 25 Marjeta Marolt customers. Besides social networking sites, other popular technologies including wikis, blogs, microblogging, podcasts, social games, photo sharing, video sharing and social bookmarking can be viewed as social media technologies. Harrigan & Miles (2014) highlighted that social media technologies seem to fit SMEs intuitive way of managing customer relationships. Because of SMEs unique needs and usage of different low-cost programs in order to accomplish similar tasks as with off-the-shelf CRM solutions, social media technologies can also be a substitute for CRM technology (Cappuccio, Kulkarni, Sohail, Haider, & Wang, 2012). Different social media technologies have a unique set of functionalities (Kaplan & Haenlein, 2010; Kietzmann, Hermkens, McCarthy, & Silvestre, 2011) which may affect CRM processes in different manners. For example, an organization can approach to customers using social networking sites and uncover the specific needs using blogs (Andzulis, Panagopoulos, & Rapp, 2012). Using social media technologies to support CRM presents many advantages for SMEs, including global reach with minimal efforts, instant feedback and improved communication, financial affordability due to lower cost compared with off-the-shelf CRM solutions, continuous interaction with customers, excellent service provided to customers through web, minimal human resources to manage client relationships compared to off-the-shelf CRM solutions (Cappuccio, Kulkarni, Sohail, Haider, & Wang, 2012). However, concerns about return on investment (ROI), negative brand exposure (word of mouth), security issues, requirements for additional funds holds organizations back from allowing social media to become part of their CRM strategy as well as business strategy (Baird & Parasnis, 2011; Bernoff & Schadler, 2010). The specific problem that this study addresses is that SMEs lack understanding how the extent of social CRM adoption may improve organizational performance and under which circumstances. This may result in organization’s incapability to reap significant benefits from social CRM experimentation and adoption (Bughin, Byers, & Chui, 2011) or even worse, experience negative consequences, such as negative brand exposure (Baird & Parasnis, 2011). 2 Problem definition Research on CRM adoption has been conducted from a variety of perspectives. In the information systems literature CRM studies have mainly focused on technology or information system perspective. However, some of them have highlighted the importance of organizational and managerial factors when considering the adoption of CRM (Hung et al., 2010; Wang & Swanson, 2008). Contrary, in the marketing literature CRM studies addressed organization-wide perspective. Only several studies identified a range of technological and organizational factors that have influence on the CRM adoption among SMEs (e.g. Alshawi, Missi, & Irani, 2011; Sophonthummapharn, 2009). Even though CRM is a complex innovation a number of studies on CRM threated the adoption decision typically as a binary choice problem (adopt or not adopt). For instance Hung, Hung, Tsai, & Jiang (2010) in their survey asked the respondents to indicate whether they adopted CRM systems or not. According to Lehmkuhl & Jung (2013) there is no generally accepted definition of social CRM and therefore people 26 Social CRM adoption and its influence on organizational performance - SMEs perspective have a different understanding of what social CRM is. While social media adds another layer of complexity to CRM practice, adoption decision is likely to be more complex and, thus, in our view, to relay the research simply on binary measure is inadequate. Furthermore, Trainor, Andzulis, Rapp, & Agnihotri (2014) propose to develop a measurement approach that will capture usage intensity or the extent to which the technologies are used within an organization. Drawing on the observation from Damanpour & Schneider (2008) that innovation is not truly adopted until “it has actually been put into use in the adopting organization” (p. 497) in this study extent of social CRM adoption is conceptualized as the actual use of social media technologies within the CRM in its broader sense, which includes not only technology but also CRM processes. Despite the ever-increasing usage of social media technologies in the context of CRM among organizations, the performance outcomes of social CRM are largely underexplored (Trainor et al., 2014). Organizations are executing numerous CRM processes to achieve their strategic objectives. There are a range of information technologies which have potentials to improve processes and consequently organizational performance (Porter & Millar, 1985). While recent research tends to include customer-related outcomes (Chang, Park, & Chaiy, 2010; Trainor et al., 2014), the organizational performance should not be measured only through effectiveness and efficiency (Chang et al., 2010). Additionally, while IT contributes to organizational performance (Kohli & Grover, 2008; Melville, Kraemer, & Gurbaxani, 2004), it appears that there is no agreement among IT scholars how it contributes to performance and therefore they seek different links between information technology and organizational performance (Chang et al., 2010; Chen et al., 2014; Melville et al., 2004). Since social CRM is becoming one of the most important tools for SMEs to perform CRM (Cappuccio et al., 2012), there is no sufficient understanding of what factors influence on the SMEs’ adoption decision, how they affect the extent of social CRM adoption and what the implications of the extent of social CRM adoption are. While social media has much greater impacts on B2C (business to customer) than B2B (business to business) relationships (Kumar & Reinartz, 2012) and the measurement of technology usage should be different for B2B or B2C relationships (Trainor et al., 2014) we focus our study on SMEs and their B2C relationships. 3 Methodology The first step of our research is qualitative data collection. More specifically, conceptual research model will integrate the findings from existing literature and insights from semi-structured interviews with consultants, IT providers and B2C orientated SMEs in Slovenia. For the interviews SMEs will be specifically selected because we want to get insights from different levels of social CRM adoption. After the development of the conceptual research model the hypotheses and initial questionnaire used to collect quantitative data will be developed. To increase the validity of the measurement instrument the questions will be partially developed based on validated measures available in the literature. The questionnaire will be initially reviewed by two experts from academic background: an expert in electronic business and an expert in SME management. Then a small pilot study will be conducted. After revision of the questionnaire the invitation letter will be sent out. For the quantitative 27 Marjeta Marolt phase a random sample of 1000 SMEs will be obtained. We will collect data using both, the web and paper-based questionnaire. The marketing, sales, or customer service executives, general managers and owners of the B2C focused SMEs in Slovenia have been chosen to be the key informants. 4 Initial conceptual research model With the integration of TOE framework, DOI theory, RBV theory, we developed a conceptual research model presented in Figure 1, followed by explanation of the key elements and their relationships. Factors Technological factors Relative advantages Extent of social CRM adoption Compatibility Breadth Complexity  Campaign management  Lead management Technology readiness  Contact management  Offer management Organizatonal factors  Contract management,  Retention management Organizational performance Top management support  Service management  efficiency  Complain and feedback  sales performance Organizational innovativeness management  customer satisfaction Customer centric strategy Depth Assimilation of customer Environmental factors data from all organization- customer interactions Customer readiness Competitor pressure Figure 1: Initial conceptual research model For the accurate and complete picture of social media use in the CRM context across the organization Trainor et al. (2014) suggest to employ a measurement that capture the extent to which the technologies are used within an organization. While social media technologies are used to support different CRM processes (Reinhold & Alt, 2012) and are rarely integrated with legacy systems (Sussin, Thompson, Sarner, & Hopkins, 2013) we follow the work by Zhu & Kraemer (2005) and (Wu, Mahajan, & Balasubramanian, 2003) on the extent of use of e-business and we conceptualise the extent of social CRM adoption in two dimensions as breadth and depth. In our study, breadth refers to CRM processes where social media technologies are adopted while depth refers to assimilation of customer data from all organization-customer interactions (Jayachandran et al., 2005). Following the Wu et al. (2003) idea of intensity of e-business adoption in our research model we use for the measurement of breadth dimension CRM processes. Namely, according to Reinhold & Alt (2012) social CRM approaches are not separated from CRM processes. Furthermore, Kiron et al. (2013) in their analysis illustrated that social media is used to support CRM processes to a large or even great extent. Namely through social media technologies organizations can have more influence on customer 28 Social CRM adoption and its influence on organizational performance - SMEs perspective loyalty and ‘word of mouth’ advertising (Greenberg, 2010). Standard CRM processes that were identified as relevant for most industries are used in the conceptual research model. These processes are campaign, lead, contact, offer, contract, retention, service and complain and feedback management (Gebert, Geib, Kolbe, & Brenner, 2003; Reinhold & Alt, 2012). Furthermore we incorporated into the research model the major factors from previous research that are found to have a significant impact on the adoption of innovation and likely to affect the breadth and the depth of social CRM adoption. These factors are grouped in three contexts: technological, organizational and environmental factors. The potential factors included in the technological context of our research model are relative advantage, compatibility, complexity and technology readiness. The first three factors are based on DOI theory by Rogers (2003) and we included them because they are consistently found to have a significant impact on the adoption of innovation. In this context we also included technology readiness that is based on TOE framework. The potential determinant factors included in the organizational context of our research model are top management support, organizational culture, customer centric strategy. The potential determinant factors included in the environmental context of our research model are customer readiness and competitor pressure. Determinant included in the organizational and environment context are based on TOE framework. Additionally, we include into the research model organizational performance to gain an overview on the entire chain of social CRM adoption composed of adoption factors, extent of adoption and organizational performance. In this study organizational performance focuses on three key aspects of customer relationship outcomes, namely efficiency, sales performance and customer-related outcomes (e.g. Chang et al., 2010; Melville et al., 2004; Wu et al., 2003). 5 Limitations We also identified several limitations:  Our study is limited on the user perspective and provides partial insights on consultant and provider perspective. In qualitative research phase insights from 5 consultants and social CRM providers and 5 users (SMEs) will be collected while in quantitative phase only 1000 randomly selected SMEs will participate.  This study will be limited to the B2C focused SMEs in Slovenia. Our concerns about generalization are eased by the fact that Slovenia B2C orientated SMEs seem not significantly different from the overall European SMEs.  This research will relay on survey responses provided by one key informant per organization. 6 Expected results and contribution This study will contribute to theory by conceptualizing the extent of social CRM adoption and its influence on business performance. This conceptualization will offer a broader understanding of the extent of social CRM and the better understanding of factors influencing on the extent of social CRM adoption as well as its influence on organizational performance. Our study will also have practical implications. With better understanding of factors influencing on the social CRM adoption the marketing, sales, or customer service 29 Marjeta Marolt executives, general managers or owners of the SMEs will know which factors have to be taken into account when considering the adoption of social CRM and how the extent of social CRM adoption can influence organizational performance. Furthermore, IT providers will have additional insights from the current adoption situation so they will be able to provide more suitable social CRM solutions. Last but not least, social CRM consultants can provide better support during the adoption and implementation phases. References Alshawi, S., Missi, F., & Irani, Z. (2011). Organisational, technical and data quality factors in CRM adoption — SMEs perspective. Industrial Marketing Management, 40(3), 376–383. doi:10.1016/j.indmarman.2010.08.006 Andzulis, J. M., Panagopoulos, N. G., & Rapp, A. (2012). A review of social media and implications for the sales process. Journal of Personal Selling and Sales Management, 3, 305–316. Baird, C. H., & Parasnis, G. (2011). From social media to Social CRM: reinventing the customer relationship. Strategy & Leadership, 39(6), 27–34. doi:10.1108/10878571111176600 Bernoff, J., & Schadler, T. (2010). Empowered. Harvard Business Review, 95–100. Retrieved from http://www.hbr.org Bughin, J., Byers, A. H., & Chui, M. (2011). How social technologies are extending the organization. Retrieved from https://www.mckinseyquarterly.com Cappuccio, S., Kulkarni, S., Sohail, M., Haider, M., & Wang, X. (2012). Social CRM for SMEs: Current Tools and Strategy. Springer Berlin Heidelberg, 422–435. Chang, W., Park, J. E., & Chaiy, S. (2010). How does CRM technology transform into organizational performance? A mediating role of marketing capability. Journal of Business Research, 63(8), 849–855. doi:10.1016/j.jbusres.2009.07.003 Chen, Y., Wang, Y., Nevo, S., Jin, J., Wang, L., & Chow, W. S. (2014). IT capability and organizational performance: the roles of business process agility and environmental factors. European Journal of Information Systems, 23(3), 326–342. doi:10.1057/ejis.2013.4 Damanpour, F., & Schneider, M. (2008). Characteristics of Innovation and Innovation Adoption in Public Organizations: Assessing the Role of Managers. Journal of Public Administration Research and Theory, 19(3), 495–522. doi:10.1093/jopart/mun021 Gebert, H., Geib, M., Kolbe, L., & Brenner, W. (2003). Knowledge-enabled customer relationship management: integrating customer relationship management and knowledge management concepts. Journal of Knowledge Management, 7(5), 107– 123. doi:10.1108/13673270310505421 Greenberg, P. (2010). The impact of CRM 2.0 on customer insight. Journal of Business Industrial Marketing, 25(6), 410–419. doi:10.1108/08858621011066008 30 Social CRM adoption and its influence on organizational performance - SMEs perspective Harrigan, P., & Miles, M. (2014). From e-CRM to s-CRM. Critical factors underpinning the Social CRM activities of SMEs. Small Enterprise Research, 21(1). doi:10.5172/ser.v21i1.5496 Hung, S.-Y., Hung, W.-H., Tsai, C.-A., & Jiang, S.-C. (2010). Critical factors of hospital adoption on CRM system: Organizational and information system perspectives. Decision Support Systems, 48(4), 592–603. doi:10.1016/j.dss.2009.11.009 Jayachandran, S., Sharma, S., Kaufman, P., & Raman, P. (2005). The Role of Relational Information Processes and Technology Use in Customer Relationship Management. Journal of Marketing, 69(4), 177–192. doi:10.1509/jmkg.2005.69.4.177 Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of Social Media. Business Horizons. doi:10.1016/j.bushor.2009.09.003 Kietzmann, J. H., Hermkens, K., McCarthy, I. P., & Silvestre, B. S. (2011). Social media? Get serious! Understanding the functional building blocks of social media. Business Horizons. doi:10.1016/j.bushor.2011.01.005 Kiron, D., Palmer, D., Nguyen Phillips, A., & Berkman, R. (2013). Social Business: Shifting Out of First Gear | MIT Sloan Management Review. Retrieved from http://sloanreview.mit.edu/reports/shifting-social-business/ Kohli, R., & Grover, V. Business Value of IT: An Essay on Expanding Research Directions to Keep up with the Times, 9 Journal of the Association for Information Systems (2008). Retrieved from http://aisel.aisnet.org/jais/vol9/iss1/1 Kumar, V., & Reinartz, W. (2012). Strategic Customer Relationship Management Today. In Customer Relationship Management (pp. 3–20). Berlin, Heidelberg: Springer Berlin Heidelberg. doi:10.1007/978-3-642-20110-3 Lehmkuhl, T., & Jung, R. Towards Social CRM – Scoping the Concept and Guiding Research, BLED 2013 Proceedings (2013). Retrieved from http://aisel.aisnet.org/bled2013/14 Melville, N., Kraemer, K., & Gurbaxani, V. (2004). Review: information technology and organizational performance: an integrative model of it business value. MIS Quarterly, 28(2), 283–322. Retrieved from http://dl.acm.org/citation.cfm?id=2017219.2017226 Porter, M. E., & Millar, V. A. (1985). How Information Gives You Competitive Advantage. Competitive Advantage: Creating and Sustaining Superior Performance, 63(4), 149–160. Retrieved from http://www.hbs.edu/faculty/Pages/item.aspx?num=4322 Rainie, L., Purcell, K., & Aaron, S. (2011). The Social Side of the Internet | Pew Research Center’s Internet & American Life Project (p. 40). Retrieved from http://www.pewinternet.org/Reports/2011/The-Social-Side-of-the-Internet.aspx 31 Marjeta Marolt Reinhold, O., & Alt, R. Social Customer Relationship Management: State of the Art and Learnings from Current Projects, BLED 2012 Proceedings (2012). Retrieved from http://aisel.aisnet.org/bled2012/26 Rogers, E. (2003). Diffusion of innovations (5th ed.). New York: Free Press. Sophonthummapharn, K. (2009). The adoption of techno-relationship innovations: A framework for electronic customer relationship management. Marketing Intelligence & Planning, 27(3), 380–412. doi:10.1108/02634500910955254 Sussin, J., Thompson, E., Sarner, A., & Hopkins, J. (2013). The Five Stages of Social CRM Adoption (pp. 1–10). Retrieved from http://my.gartner.com/portal/server.pt?open=512&objID=260&mode=2&PageID= 3460702&resId=1836723&ref=QuickSearch&sthkw=5+Stages+of+Social+CRM+ Adoption+2011 Trainor, K. J., Andzulis, J. (Mick), Rapp, A., & Agnihotri, R. (2014). Social media technology usage and customer relationship performance: A capabilities-based examination of social CRM. Journal of Business Research, 1201–1208. doi:10.1016/j.jbusres.2013.05.002 Wu, F., Mahajan, V., & Balasubramanian, S. (2003). An Analysis of E-Business Adoption and its Impact on Business Performance. Journal of the Academy of Marketing Science, 31(4), 425–447. doi:10.1177/0092070303255379 Zhu, K., & Kraemer, K. L. (2005). Post-Adoption Variations in Usage and Value of E- Business by Organizations: Cross-Country Evidence from the Retail Industry. Retrieved from http://pubsonline.informs.org/doi/abs/10.1287/isre.1050.0045 32 28th Bled eConference #eWellBeing June 7 - 105, 2015; Bled, Slovenia Research plan ‘Optimising implementations of innovations in Operating Room Processes’ Abstract Navin Sewberath Misser Bsc Msc University Medical Centre Utrecht / HU University of Applied Sciences, The Netherlands Navin.SewberathMisser@hu.nl Keywords: Technological Innovations, Operating Rooms, Healthcare, Implementation guidelines 1. Introduction Surgeries in operating theatres are complex en dynamic processes which are supported by various actors and technologies. The surgical process can be divided in three major processes: pre-operative, peroperative, and post-operative processes (Girotto, Koltz, & Drugas, 2010; Inspectorate for Healthcare, 2008). Many of these operative processes are supported by technology and/or medical devices. Medical devices including software are tools to be used for diagnostic or therapeutic purposes or to support bodily functions (Dutch Association of hospitals NVZ, 2011). Stefanidis (Stefanidis, Fanelli, Price, & Richardson, 2014) distinguishes innovations in processes or techniques and used technology during surgery. The Dutch Hospital Association developed an agreement for ‘Medical Technology and devices’ with a focus on safe application of medical technology within hospitals (Dutch Association of hospitals NVZ, 2011). This agreement implies implementing policies in acquiring, implementing, using and disposing medical devices and technology within hospitals. The Dutch Inspectorate of Healthcare announced that audits on the implementation of this agreement in hospitals will continue in 2014 and 2015 (Inspectorate for Healthcare, 2014). Technological innovations for supporting processes in preoperative and postoperative processes are introduced regularly such as logistic processes or ergonomic tools. In contrast to the use of medical devices, general policies to implement technological innovations for supportive processes have not been addressed. Observations and discussions with 35 Navin Sewberath Misser stakeholders such as quality departments, scrub and anaesthesia nurses show that implementations of new technologies in OR’s vary from being ad-hoc to systematically planned across various departments within the hospital. In order to address the implementation of technological innovations, specifically for supportive processes, a PhD research project is proposed with the following research question: What are guidelines and situational factors related to a successful implementation of technological innovations within operating rooms? 2. Method Peffers developed a design science research methodology with six phases (see Figure 1) (Peffers, Tuunanen, Rothenberger, & Chatterjee, 2007). Hevner developed a research framework in which practical problems are identified (the variety and complexity of introducing technological innovations in ORs) and analyzed resulting in justified contributions to the knowledge base and to practice (Hevner, March, Park, & Ram, 2004). For this research the design science method will be used, combined with Hevner’s guidelines (see figure 1). Figure 1 Design science research method Source: (Peffers et al., 2007) In the first phase the problem is identified and a research proposal will be written. The next phase will commence with a systematic literature review on situational factors that influence the success of innovations according to a predefined protocol that fits the context. This protocol will consist of a research question, a search strategy and predefined criteria to analyze the results (Kitchenham & Charters, 2007). To determine success factors of major interventions and innovations, a case study on a relocation to a new OR complex will be written (Bryman & Bell, 2007). Based on these outcomes an implementation model will be designed in phase three. In the following phases the model will be evaluated and improved in projects and recorded in cases. Other evaluation methods will be determined pending the projects. 36 Optimising implementations of innovations in Operating Room Processes for Bled eConference: Doctoral Consortium 3. Expected results The expected result is an implementation model for technological innovations within an OR-environment. Based on principles of Business and IT-alignment for new IT implementations, factors are: process, technology and ICT, monitoring and control, organisational departments, human factors (Venkatraman, Henderson, & Oldach, 1993). The model will provide guidelines and (situational) factors to implement innovations within the OR. References Bryman, E., & Bell, E. (2007). Business Research Methods. New York, USA: Oxford University Press. Dutch Association of hospitals NVZ. (2011). Convenant Veilige Toepassing van Medische Technologie in het ziekenhuis. Girotto, J. a, Koltz, P. F., & Drugas, G. (2010). Optimizing your operating room: or, why large, traditional hospitals don’t work. International Journal of Surgery (London, England), 8(5), 359–67. doi:10.1016/j.ijsu.2010.05.002 Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design science in information systems research. MIS Quarterly, 28(1), 75–105. Inspectorate for Healthcare. (2008). Standaardisatie onmisbaar voor risicovermindering in operatief proces. Inspectorate for Healthcare. (2014). Veilig gebruik van medische technologie krijgt onvoldoende bestuurlijke aandacht in de ziekenhuizen. Kitchenham, B., & Charters, S. (2007). Guidelines for performing Systematic Literature Reviews in Software Engineering. Keele University (Vol. 2). doi:10.1145/1134285.1134500 Peffers, K., Tuunanen, T., Rothenberger, M. a., & Chatterjee, S. (2007). A Design Science Research Methodology for Information Systems Research. Journal of Management Information Systems, 24(3), 45–77. doi:10.2753/MIS0742- 1222240302 Stefanidis, D., Fanelli, R. D., Price, R., & Richardson, W. (2014). SAGES guidelines for the introduction of new technology and techniques. Surgical Endoscopy, 28(8), 2257–71. doi:10.1007/s00464-014-3587-6 Venkatraman, N., Henderson, J. C., & Oldach, S. (1993). Continuous strategic alignment: Exploiting information technology capabilities for competitive success. European Management Journal, 11(2), 139–149. doi:10.1016/0263- 2373(93)90037-I 37 28th Bled eConference #eWellBeing June 7 - 105, 2015; Bled, Slovenia Research plan ‘Optimising implementations of innovations in Operating Room Processes’ Abstract Navin Sewberath Misser Bsc Msc University Medical Centre Utrecht / HU University of Applied Sciences, The Netherlands Navin.SewberathMisser@hu.nl Keywords: Technological Innovations, Operating Rooms, Healthcare, Implementation guidelines 1. Introduction Surgeries in operating theatres are complex en dynamic processes which are supported by various actors and technologies. The surgical process can be divided in three major processes: pre-operative, peroperative, and post-operative processes (Girotto, Koltz, & Drugas, 2010; Inspectorate for Healthcare, 2008). Many of these operative processes are supported by technology and/or medical devices. Medical devices including software are tools to be used for diagnostic or therapeutic purposes or to support bodily functions (Dutch Association of hospitals NVZ, 2011). Stefanidis (Stefanidis, Fanelli, Price, & Richardson, 2014) distinguishes innovations in processes or techniques and used technology during surgery. The Dutch Hospital Association developed an agreement for ‘Medical Technology and devices’ with a focus on safe application of medical technology within hospitals (Dutch Association of hospitals NVZ, 2011). This agreement implies implementing policies in acquiring, implementing, using and disposing medical devices and technology within hospitals. The Dutch Inspectorate of Healthcare announced that audits on the implementation of this agreement in hospitals will continue in 2014 and 2015 (Inspectorate for Healthcare, 2014). Technological innovations for supporting processes in preoperative and postoperative processes are introduced regularly such as logistic processes or ergonomic tools. In contrast to the use of medical devices, general policies to implement technological innovations for supportive processes have not been addressed. Observations and discussions with 1 Navin Sewberath Misser stakeholders such as quality departments, scrub and anaesthesia nurses show that implementations of new technologies in OR’s vary from being ad-hoc to systematically planned across various departments within the hospital. In order to address the implementation of technological innovations, specifically for supportive processes, a PhD research project is proposed with the following research question: What are guidelines and situational factors related to a successful implementation of technological innovations within operating rooms? 2. Method Peffers developed a design science research methodology with six phases (see Figure 1) (Peffers, Tuunanen, Rothenberger, & Chatterjee, 2007). Hevner developed a research framework in which practical problems are identified (the variety and complexity of introducing technological innovations in ORs) and analyzed resulting in justified contributions to the knowledge base and to practice (Hevner, March, Park, & Ram, 2004). For this research the design science method will be used, combined with Hevner’s guidelines (see figure 1). Figure 1 Design science research method Source: (Peffers et al., 2007) In the first phase the problem is identified and a research proposal will be written. The next phase will commence with a systematic literature review on situational factors that influence the success of innovations according to a predefined protocol that fits the context. This protocol will consist of a research question, a search strategy and predefined criteria to analyze the results (Kitchenham & Charters, 2007). To determine success factors of major interventions and innovations, a case study on a relocation to a new OR complex will be written (Bryman & Bell, 2007). Based on these outcomes an implementation model will be designed in phase three. In the following phases the model will be evaluated and improved in projects and recorded in cases. Other evaluation methods will be determined pending the projects. 2 Optimising implementations of innovations in Operating Room Processes for Bled eConference: Doctoral Consortium 3. Expected results The expected result is an implementation model for technological innovations within an OR-environment. Based on principles of Business and IT-alignment for new IT implementations, factors are: process, technology and ICT, monitoring and control, organisational departments, human factors (Venkatraman, Henderson, & Oldach, 1993). The model will provide guidelines and (situational) factors to implement innovations within the OR. References Bryman, E., & Bell, E. (2007). Business Research Methods. New York, USA: Oxford University Press. Dutch Association of hospitals NVZ. (2011). Convenant Veilige Toepassing van Medische Technologie in het ziekenhuis. Girotto, J. a, Koltz, P. F., & Drugas, G. (2010). Optimizing your operating room: or, why large, traditional hospitals don’t work. International Journal of Surgery (London, England), 8(5), 359–67. doi:10.1016/j.ijsu.2010.05.002 Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design science in information systems research. MIS Quarterly, 28(1), 75–105. Inspectorate for Healthcare. (2008). Standaardisatie onmisbaar voor risicovermindering in operatief proces. Inspectorate for Healthcare. (2014). Veilig gebruik van medische technologie krijgt onvoldoende bestuurlijke aandacht in de ziekenhuizen. Kitchenham, B., & Charters, S. (2007). Guidelines for performing Systematic Literature Reviews in Software Engineering. Keele University (Vol. 2). doi:10.1145/1134285.1134500 Peffers, K., Tuunanen, T., Rothenberger, M. a., & Chatterjee, S. (2007). A Design Science Research Methodology for Information Systems Research. Journal of Management Information Systems, 24(3), 45–77. doi:10.2753/MIS0742- 1222240302 Stefanidis, D., Fanelli, R. D., Price, R., & Richardson, W. (2014). SAGES guidelines for the introduction of new technology and techniques. Surgical Endoscopy, 28(8), 2257–71. doi:10.1007/s00464-014-3587-6 Venkatraman, N., Henderson, J. C., & Oldach, S. (1993). Continuous strategic alignment: Exploiting information technology capabilities for competitive success. European Management Journal, 11(2), 139–149. doi:10.1016/0263- 2373(93)90037-I 3