Measuring Online Community Health in Enterprise-owned Technical Support Forums Mikhail Timofeev A Dissertation submitted to the University of Dublin, Trinity College in fulfilment of the requirements for the degree of Master in Science (Computer Science) July 2023 2 Declaration I declare that this thesis has not been submitted as an exercise for a degree at this or any other university and it is entirely my own work. I agree to deposit this thesis in the University’s open access institutional repository or allow the library to do so on my behalf, subject to Irish Copyright Legislation and Trinity College Library conditions of use and acknowledgement. Mikhail Timofeev Dated: July 1, 2023 Acknowledgments I would like to thank my entire family who have supported and encouraged my education from the very beginning. I would like to thank my friends, colleagues and TCD staff for all their help and support over the years. Mikhail Timofeev University of Dublin, Trinity College July 2023 3 4 Abstract Enterprise-owned online communities (OCs) are purpose-built virtual spaces that support various business activities, such as customer support, marketing and product development. These communities are formed with the objective of meeting the company’s goals, while ensuring customer engagement and satisfaction. However, measuring the effectiveness of these OCs has been a challenge for companies due to the lack of publicly available diagnostic tools and fragmented research in this area. Researchers have proposed various indicators and drivers of OC success and health, but empirical verification and theoretical justification are lacking for many of them. This study aims to develop a theoretical understanding of online community health (OCH) and connect it with previously suggested diagnostics suitable for measuring OCH in technical support (TS) OCs. The study uses a two-phase exploratory case study to diagnose OCH of existing TS OCs owned by two established software producers. The first phase involved conducting interviews with OC managers to identify diagnostics and form a list of computable metrics. In the second phase, online activity patterns and health factors were analysed using quantitative methods. Both phases were concluded by comparing the OCs and their evaluated OCH. The results indicate the usefulness of the proposed diagnostic tools and their ability to diagnose online community health. The study expands the current notion of OCH and proposes ways to improve OC dynamics through actively focusing on user role compositions. 5 Contents Chapter 1. Introduction and Overview ................................................... 17 1.1 Introduction .................................................................................................... 17 1.2 Background and Context ................................................................................ 17 1.3 Statement of the Problem ............................................................................... 19 1.4 Research Questions ......................................................................................... 20 1.5 The Boundaries of this Study .......................................................................... 21 1.6 Research Contributions ................................................................................... 21 1.7 Thesis Structure .............................................................................................. 22 Chapter 2. Literature Review ................................................................. 24 2.1 Introduction .................................................................................................... 24 2.2 Definition of an OC ......................................................................................... 24 2.3 Participation in OC ......................................................................................... 34 2.4 Enterprise-owned TS OC ................................................................................ 38 2.5 OC Success ...................................................................................................... 43 2.6 OC Health ....................................................................................................... 48 2.7 OC Success-Health Relationship ..................................................................... 52 2.8 Measuring OCH ............................................................................................... 57 2.9 Summary ......................................................................................................... 62 6 Chapter 3. Methodology .......................................................................... 64 3.1 Introduction ..................................................................................................... 64 3.2 Philosophical Considerations ........................................................................... 65 3.3 Research Approach .......................................................................................... 70 3.4 Interview Methodology .................................................................................... 74 3.5 Case Study Methodology ................................................................................. 75 3.6 Research Design .............................................................................................. 78 3.7 Data Collection ............................................................................................... 79 3.8 Interview Design .............................................................................................. 80 3.8.1 Designing ................................................................................................... 80 3.8.2 Sampling ................................................................................................... 81 3.8.3 Setting ....................................................................................................... 82 3.8.4 Administration .......................................................................................... 83 3.9 Validation in Case Study ................................................................................ 84 3.10 Ethical Considerations ................................................................................... 86 3.11 Role of the Researcher ................................................................................... 87 3.12 Summary ....................................................................................................... 87 Chapter 4. Analysis ................................................................................. 88 4.1 Introduction ..................................................................................................... 88 4.2 Interview Analysis: NC .................................................................................... 88 4.2.1 Background of NC ..................................................................................... 88 4.2.2 Curation of NC ......................................................................................... 91 7 4.2.3 Health and Success of NC ......................................................................... 94 4.3 Interview Analysis: AF .................................................................................... 96 4.3.1 Background of AF ..................................................................................... 96 4.3.2 User Groups and MVPs in AF .................................................................. 99 4.3.3 Internal/External Moderation in AF ....................................................... 101 4.3.4 Members of AF ........................................................................................ 104 4.3.5 Health and Success of AF ........................................................................ 108 4.4 OC Data Analysis .......................................................................................... 111 4.4.1 Data Analysis of NC ................................................................................ 111 4.4.2 Data Analysis of AF ................................................................................ 116 4.4.3 Comparing OC Data Sets ........................................................................ 118 4.4.4 OCH Dimensions and Diagnostic Factors ................................................ 124 4.5 Health Factor Analysis ................................................................................... 125 4.5.1 Maturity ................................................................................................... 126 4.5.2 Generation (of Content) ........................................................................... 126 4.5.3 Consumption (of Content) ....................................................................... 134 4.5.4 Interactivity ............................................................................................. 141 4.5.5 Responsiveness ......................................................................................... 144 4.5.6 Engagement .............................................................................................. 148 4.5.7 Reliance .................................................................................................... 153 4.5.8 Maintenance ............................................................................................. 155 4.5.9 Accessibility ............................................................................................. 157 4.6 Conclusion ...................................................................................................... 158 Chapter 5. Discussion ............................................................................ 159 5.1 Introduction ................................................................................................... 159 8 5.2 Discussion of Findings .................................................................................... 159 5.2.1 NC Setting ............................................................................................... 160 5.2.2 AF Setting ................................................................................................ 162 5.2.3 OC Extraction Process ............................................................................. 163 5.3 OCH Model .................................................................................................... 164 5.4 OCH Diagnostic Measurements ...................................................................... 170 5.5 Evaluations of OCs ......................................................................................... 177 5.5.1 OC Health (OCH) .................................................................................... 177 5.5.2 OC Effectiveness (Maturity) .................................................................... 178 5.6 Research Contributions .................................................................................. 180 Chapter 6. Conclusion ........................................................................... 181 6.1 Introduction .................................................................................................... 181 6.2 Findings Summary ......................................................................................... 181 6.3 Research Limitations ...................................................................................... 182 6.4 What is New in this Research ........................................................................ 183 6.5 Future Research ............................................................................................. 183 Bibliography .......................................................................................... 184 Appendix A. CHI Investigation ............................................................ 203 Appendix B. Ethical Approval .............................................................. 214 Appendix C. Perl Extraction Script (NC) ............................................ 242 Appendix D. DB Structure Example (AF) ........................................... 243 9 Appendix E. DB Structure Example (NC) ............................................ 248 Appendix F. Derived Analytics Table ................................................... 249 Appendix G. SQL Example (jive_init_cube) ........................................ 250 10 List of Tables 2.1: Constructs of OC Definitions ............................................................................ 27 2.2: Components & Measures of OC Success ........................................................... 55 2.3: OC Subsystems, OCH Dimensions & Suggested Diagnostic Factors ................ 59 3.1: Contrasting Implications of Positivism, Interpretivism, Critical Research & Pragmatism .............................................................................................................. 67 3.2: The Major Mixed Methods Design Types ......................................................... 72 3.3: Sources of Collected Data ................................................................................. 79 4.1: NC & AF Summary Statistics ......................................................................... 118 4.2: DB Table Sets used in the Analysis ................................................................. 119 4.3: Contributor Ranks’ Construction .................................................................... 121 4.4: User Binning by Roles ..................................................................................... 122 4.5: Table Aggregates used in the Analysis ............................................................ 124 4.6: OC Subsystems, OCH Dimensions & Formulated Diagnostic Factors ............ 125 4.7: NC & AF Maturity .......................................................................................... 126 4.8: Mo. Avg. № of Messages by UGs ..................................................................... 127 4.9: Mo. Avg. Dist. of Messages by UGs ................................................................ 129 4.10: Mo. Avg. № of Threads by UGs (in noms.) ................................................... 131 4.11: Mo. Avg. Dist. of Threads by UGs (in %) ..................................................... 133 4.12: Mo. Avg. of Ignored & Solved Threads (in %) .............................................. 135 4.13: Mo. Avg. № of Solutions by UGs (in noms.) ................................................. 137 4.14: Mo. Avg. Dist. of Sols. by UGs (in %) .......................................................... 138 4.15: Tot. Traffic of Thrds. (№ of Views) .............................................................. 140 11 4.16: Mo. Avg. Dist. of UGs (Unq. Users. in Threads) .......................................... 142 4.17: Mo. Avg. № of Messages in Threads (in noms.) ............................................ 143 4.18: Mo. Avg. Time to 1st Reply (in hrs.) ............................................................. 144 4.19: Mo. Avg. Time to Solution (in hrs.) .............................................................. 146 4.20: Mo. Avg. Dist. of 1st Repliers by UGs (in %) ................................................ 147 4.21: Mo. Avg. № of Active Users (in noms.) ......................................................... 148 4.22: Mo. Avg. Dist. of Active Users by UGs (in %) ............................................. 150 4.23: Mo. Avg. Dist. of Users by Posting Patterns (in %) ..................................... 151 4.24: Avg. № of Boards UGs post in (in noms.) ..................................................... 153 4.25: Mo. Avg. Contributors’ Progression .............................................................. 154 4.26: Mo. Avg. of Most Viewed & Solved Threads (in %) ..................................... 157 5.1: NC & AF OCH ................................................................................................ 178 5.2: NC & AF Maturity Stages ............................................................................... 179 12 List of Figures 2.1: OCs’ Life-Cycle ................................................................................................. 33 2.2: Developmental Progression of Participation in OCs ......................................... 36 2.3: The Reader-to-Leader Framework .................................................................... 37 2.4: A Simplified Life-cycle of a Question in a Typical CQA Site ........................... 41 2.5: A Joint Representation of ISS model ................................................................ 45 2.6: Online Communities as Organisms ................................................................... 50 2.7: Three Dimensions of Community Measurement ............................................... 53 2.8: The Revised “OCs as Organisms” Model .......................................................... 57 2.9: The Community Maturity Model ..................................................................... 61 3.1: The Inductive-Deductive Research Cycle ......................................................... 69 3.2: Research Approach ........................................................................................... 70 3.3: Continuum of QUAL & QUAN Research ......................................................... 71 3.4: Diagram of Research Design ............................................................................. 73 3.5: Modified Multiple Case Study Method ............................................................. 76 3.6: Basic Types of Case Study Design ................................................................... 77 3.7: Overview of Research Design ........................................................................... 78 3.8: Purposeful Sampling ......................................................................................... 81 4.1: AF Points System ............................................................................................ 121 4.2: AF User Recognition Levels ............................................................................ 122 5.1: The Revised “OCs as Organisms” Model ........................................................ 165 5.2: The Resulting “OCs as Organisms” Model ...................................................... 169 13 List of Charts 4.1: NC: № of Messages by UGs (in K.) ................................................................. 128 4.2: AF: № of Messages by UGs (in K.) ................................................................. 129 4.3: NC: Dist. of Tot. Messages by UGs (in %) ...................................................... 130 4.4: AF: Dist. of Tot. Messages. by UGs (in %) ..................................................... 130 4.5: NC: № of Threads by UGs (in noms.) ............................................................. 132 4.6: AF: № of Threads by UGs (in K.) ................................................................... 132 4.7: NC: № of Threads by UGs (in %) ................................................................... 133 4.8: AF: № of Threads by UGs (in %) .................................................................... 133 4.9: NC: Dist. of Ignored, Solved & Unsolved. Threads (in %) .............................. 136 4.10: AF: Dist. of Ignored., Solved & Unsolved Threads (in %) ............................ 136 4.11: NC: № of Solutions by UGs (in noms.) ......................................................... 137 4.12: AF: № of Sols. by UGs (in K.) ...................................................................... 138 4.13: NC: Dist. of Solutions by UGs (in %) ........................................................... 139 4.14: AF: Dist. of Solutions by UGs (in %) ............................................................ 139 4.15: NC: Tot. Traffic of Thrds. a Mo. (in M.) ...................................................... 141 4.16: AF: Tot. Traffic of Thrds. a Mo. (in M.) ...................................................... 141 4.17: NC: Dist. of Unq. Users in Threads by UGs (in %) ...................................... 142 4.18: AF: Dist. of Unq. Users in Threads by UGs (in %) ....................................... 142 4.19: NC: Avg. № of Messages in Threads a Mo. (in noms.) .................................. 143 4.20: AF: Avg. № of Messages in Threads a Mo. (in noms.) .................................. 143 4.21: NC: Avg. Time to 1st Reply by UGs (in hrs.) ................................................ 145 4.22: AF: Avg. Time to 1st Reply by UGs (in hrs.) ................................................ 145 14 4.23: NC: Avg. Time to Solution by UGs (in hrs.) ................................................. 146 4.24: AF: Avg. Time to Solution by UGs (in hrs.) ................................................. 146 4.25: NC: Dist. of 1st Repliers in Threads between UGs (in %) ............................. 147 4.26: AF: Dist. of 1st Repliers in Threads between UGs (in %) .............................. 147 4.27: NC: № of Active Users (in noms.) ................................................................. 149 4.28: AF: № of Active Users (in noms.) .................................................................. 149 4.29: NC: Dist. of Unq. Users Posting each Mo. (in %) ......................................... 150 4.30: AF: Dist. of Unq. Users Posting each Mo. (in %) ......................................... 150 4.31: NC: Dist. of Users based on Engagement Types (in %) ................................ 152 4.32: AF: Dist. of Users based on Engagement Types (in %) ................................ 152 4.33: NC: № of Boards UGs post in (in noms.) ...................................................... 153 4.34: AF: № of Boards UGs post in (in noms.) ...................................................... 153 4.35: NC: Contributors’ Progression (in noms.) ..................................................... 155 4.36: AF: Contributors’ Progression (in noms.) ..................................................... 155 4.37: NC: Removed Threads. (in noms.) ................................................................ 156 4.38: AF: Removed Threads (in noms.) ................................................................. 156 4.39: NC: Dist. of Solved & Unsolved 10 Most-Viewed Threads (in %) ................. 158 4.40: AF: Dist. of Solved & Unsolved 10 Most-Viewed Threads (in %) ................. 158 15 Abbreviations ACP Adobe Community Professional AEL Adobe Education Leader API Application Programming Interface AUG Adobe User Group CC Creative Cloud (Adobe) CHI Community Health Index CL Community Leader CM Community Manager/Moderator CMC Computer-Mediated Communication CQA Community Question Answering CRM Customer Relationship Management IS Information Systems IT Information Technology KB Knowledge Base LU Lead User 16 MVP Most Valuable Participant NDA Non-Disclosure Agreement OC Online Community OCH Online Community Health Q&A Question and Answer ROI Return on Investment SoC Sense of Community SQL Structured Query Language TS Technical Support UGC User-Generated Content XML Extensible Markup Language Chapter 1 Introduction and Overview 1.1 Introduction This thesis utilises an exploratory case study approach to assess online community (OC) health of technical support (TS) communities that are owned by enterprises. The purpose of this study is to support and enhance existing research on online community health (OCH), investigate practical methods of measuring health and explore ways of improving OCH by ensuring a balanced representation of various user groups in online forums. 1.2 Background and Context Online communities refer to virtual spaces, where individuals come together to interact, share information and ideas, collaborate and engage in common activities. Since the 1980s, OCs have captured the attention of many internet users and they have evolved to become widely adopted by businesses (Preece, Maloney-Krichmar & Abras, 2003; Bughin, 2015). Enterprises use and deploy OCs both internally and externally to interact with their customers, employees and partners. According to Margolis (2016), it was predicted that by the end of 2017, almost 70% of companies worldwide would have incorporated some form of Enterprise 2.0 technologies, including corporate OCs, leading to a projected $23 billion industry by 2019 Chapter 1. Introduction and Overview 18 (Hinchcliffe, 2016). The widespread adoption of OCs by businesses can be attributed to various social, technological and commercial reasons, such as the development of new internet-based technologies, like cloud and big data, the popularity of social networking sites (SNS), the growth of e-commerce and the monetisation of online transactions. Recent interest by enterprises in online interactions with customers has led to a development of social customer relationship management (CRM) offerings, which include an OC component/platform. At the time of this research, system vendors such as Jive, Lithium and Higher Logic (The Community Roundtable, 2017) provided OC hosting solutions with built-in analytic tools. These tools enable community managers/moderators (CMs) to measure basic activity metrics, such as the number of members, posts/threads, clicks and views (traffic), among other criteria. While these numbers are typically included in business reports, they do not provide valuable insights into how well an OC serves its purpose and delivers on its promises. To address this challenge, new paradigms of community analytics have emerged, such as online community health (OCH) and return-on-investment (ROI or social ROI) of OCs (The Community Roundtable, 2017; Wu, 2012; Cothrel & Schustler, 2013; Oracle Corporation, 2012; Lithium Technologies, 2009). ROI helps assess OC’s effectiveness in generating business value, while OCH allows companies to not only evaluate how well their OCs are performing, but also ensure that their social platforms continue to thrive and grow amidst future challenges. Community health is similar to how humans monitor their own health (Carillo, 2017), where using sophisticated diagnostic and preventive measures help address issues related to physical well-being. In OCs, health diagnostics are aimed at measuring its current activity and performance levels, where CMs can formulate a plan for improving OCH and predict how well the community will thrive in the future. Chapter 1. Introduction and Overview 19 This research employs an exploratory case study methodology to examine a pair of enterprise-owned tech support OCs, that belong to large and established software vendors. The research findings will contribute to the understanding of the concept of OCH, how it can be measured and improved through changing the presence of various user groups in online forums. At the time of writing, there were no studies, that bridged previously derived theoretical models with practical diagnostic factors of OC health and comprehensively showed how OCH is analysed in large TS corporate communities. 1.3 Statement of the Problem The management of online communities is a complex issue that involves ensuring long- term prosperity through active performance assessments. Despite a growing number of academic publications on measuring OC dynamics, the concept of OCH remains poorly defined (Lithium Technologies, 2009; Rowe & Alani, 2012; Wagner et al., 2014; Wang & Lantzy, 2011; Carillo, 2017). The theoretical model currently available for researching OC health is still in its infancy (Carillo, 2017) and many proposed diagnostic factors have yet to be validated by both researchers and practitioners (Lithium Technologies, 2009; Rowe & Alani, 2012; Wagner et al., 2014; Wang & Lantzy, 2011; Aumayr & Hayes, 2017b). Additionally, there is limited understanding of how user group activity patterns influence OC ‘healthiness’ (Rowe & Hayes, 2012; Angeletou, Rowe & Alani, 2011; Rowe & Alani, 2012; Ridings & Wasko, 2010). This presents a challenge for business specialists, as engagement with crowds carries high risks of failure and unpredictable outcomes, if OCH is not sustained. Consequently, CMs require deeper assessment and forecasting of OC health, that goes beyond basic metrics, such as total activity, number of members, active and new Chapter 1. Introduction and Overview 20 members, number of posts/threads and views, time to first response and others (The Community Roundtable, 2017; Wagner et al., 2014). An important current focus is on developing more comprehensive methods of diagnosing OC health, which are tailored to specific community types. Such methods are likely to provide a better diagnosis of OC health for particular community types, such as TS OCs, which represent almost 40% of all external enterprise-owned communities (The Community Roundtable, 2017). 1.4 Research Questions The research problem has directed the formulation of the following primary research aim: To explore the health of enterprise-owned tech support online communities using an exploratory mixed methods case study methodology. This research is exploratory due to the emergent nature of the OCH concept and the need to understand OC health, rather than focusing on verifying the previously- suggested diagnostic factors empirically. A structured methodology was used together with instructed data collection and analysis procedures in order to guide the exploratory research aim and attain the following research questions: • RQ1: To what extent can OC health can be measured in an enterprise-owned Tech Support community? • RQ2: How can OC health be measured in an enterprise-owned tech support community? Chapter 1. Introduction and Overview 21 1.5 The Boundaries of this Study This research is constrained by the existing literature on the success of OCs and their impact on OCH (refer to Section 2.5-2.7 for more information). Furthermore, it is contextualised within enterprise-owned communities, where the strategic objectives of host organisations, Symantec and Adobe, define the purpose of the two communities under investigation – Norton Community (NC) and Adobe Forums (AF). These communities were chosen for this study, since they aim to address similar customer support issues, they are centred around software products/services and have dedicated OC management teams, who were interviewed for this study. The wealth of data collected, including interviews with key OC personnel and extracted OC data in the form of relational databases, enabled both qualitative and quantitative analysis of the OCs in question. The study was able to overcome various technical challenges, such as differences in hosting platforms, database schemes, built-in analytical metrics and rendered results, which were suitable for cross-examination. 1.6 Research Contributions In this thesis, the focus is on examining techniques and approaches for measuring and enhancing the health of enterprise-owned technical support OCs. The research makes the following contributions to the existing body of knowledge: • Providing a unified view of OC success and health. Earlier research has viewed OC success and health as similar concepts without defining how they differ. This thesis examines both concepts and proposes ways of establishing a success-health relationship for future research, thus ensuring the separation of these concepts in subsequent studies. Chapter 1. Introduction and Overview 22 • Bridging the current theory of OCH with computable diagnostic metrics. The OCH factors suggested in previous research are mainly based on practice and lack theoretical justification. The research aims to connect the theoretical OCH framework with the previously proposed factors to derive diagnostic methods that are grounded in both theory and practice. • Expanding the methods of OCH diagnosis in enterprise-owned TS communities. Managers of OCs require new tools to carry out OCH diagnosis, specifically in communities created for customer support. This study captures multiple dimensions of how OCH can be diagnosed in TS communities and provides suggestions for CMs on how they can apply similar methods in their daily routines. 1.7 Thesis Structure The structure of this thesis is outlined as follows: Literature Review. Chapter 2 provides an explanation of OCs, including the different types and the nature of user participation. Additionally, it discusses technical support OCs, that are owned by enterprises. This chapter also reviews the concepts of OC success and health, including the diagnostic factors from previous studies. Furthermore, it establishes the relationship between OC success and health and presents an updated model (“OCs as Organisms”), which reflects this relationship. The remainder of the chapter focuses on expanding the OCH framework by linking OC health subsystems with appropriate dimensions and examples of measurable diagnostic factors. Chapter 1. Introduction and Overview 23 Methodology. Chapter 3 describes the philosophical, epistemological rationale and methods required to answer the research questions. The chapter explains research elements, such as research design, sampling, data collection and data analysis. Additionally, ethical considerations, applicable in the context of OC research, are discussed in this chapter. Analysis. Chapter 4 presents the results of the exploratory case studies, including the qualitative (interviews) and quantitative (OC datasets) analysis for both OCs. The interviews describe the history, context and specifics of each community, together with the suggested OCH factors. The analysis of the OC datasets looks at each separate OCH factor by reviewing community dynamics, user group composition charts and descriptive statistics, where applicable. Discussion. Chapter 5 evaluates the results of the previously presented analysis. It describes each OCH dimension and evaluates the level of OC health in each case study. Furthermore, it establishes the results of the cross-case study. Conclusions. Chapter 6 summarises the study conclusions and proposes ideas for future work. Chapter 2 Literature Review 2.1 Introduction This section provides a comprehensive analysis of the health of OCs and explains how the main research objective of this thesis was formulated. Firstly, the concept of online communities is reviewed along with an investigation of different types of OCs and user engagement levels. Subsequently, TS OCs owned by enterprises are examined together with the crucial themes, that are most pertinent in measuring OC success and health. Next, the novel concept of OCH is scrutinised and studied in-depth, including an evaluation of the diverse health metrics employed thus far in OC research. 2.2 Definition of an OC Over the past three decades, online communities have been the subject of extensive research in various fields and contexts, including sociology, psychology, business, management and information systems (IS) (Hew, 2009; Preece, 2000; Wenger, White & Smith, 2009; Rheingold, 1993; Kozinets, 2010; Kim, 2000). Despite the significant amount of literature available, OCs are still considered to be an emerging, continuously evolving and complex phenomenon (Marquois-Ogez & Bothorel, 2006; Hercheui, 2011; Leyton Escobar, Kommers & Beldad, 2014). The term “online community” itself is a topic of debate among academics (Preece, 2001; Leimeister, Chapter 2. Literature Review 25 Sidiras & Krcmar, 2006; Malinen, 2015). Alternative terms used include “virtual community” (Rheingold, 1993) and “community of practice” (sometimes seen with an “online” or “virtual” prefix) (Wenger, 1998; Johnson, 2001; Bourhis, Dubé & Jacob, 2005). However, the term “online community” remains the most prevalent in IS research (Lin & Lee, 2006; Iriberri & Leroy, 2009; Rowe & Alani, 2012; Petrič, 2014; Aumayr & Hayes, 2017b). Similarly, the terms “online” or “online communications” are used interchangeably in this study to represent any computer-mediated communications (CMC) conducted over the Internet. Howard Rheingold (1993), one of the most cited authors in the literature on online communities, proposed the first working definition of an OC from a sociological perspective: “Social aggregations that emerge from the Net when enough people carry on public discussions long enough, with sufficient human feeling to form webs of personal relationships in cyberspace” (Rheingold, 1993:p.5). Jenny Preece (2005) presented an alternative definition that emphasised specific components of OCs, including people, common interests, shared purposes, policies and CMC: “A group of people with a common interest or a shared purpose whose interactions are governed by policies in the form of tacit assumptions, rituals, protocols, rules, and laws and who use computer systems to support and mediate social interaction and facilitate a sense of togetherness” (Maloney-Krichmar & Preece, 2005:p.203). Chapter 2. Literature Review 26 Various OC definitions have been proposed in recent times, using constructs that align with Preece’s definition: “Online communities (OCs) are open collectives of dispersed individuals with members who are not necessarily known or identifiable and who share common interests, and these communities attend to both their individual and their collective welfare” (Faraj, Jarvenpaa & Majchrzak, 2011:p.1224). “Online communities (OCs) are virtual social groups with a set of members who contribute to a varying extent to a common activity and/or good according to behavioural scripts” (Schneider, Krogh & Jäger, 2013:p.1). Some researchers have suggested focusing on the salient characteristics of OCs instead of seeking a canonical definition. Velasquez et al. (2014) describes OCs as a “super- set” of systems that allows: • expression through information and communication technology (ICT) • interaction between users of the site; • dependence on user-generated content (UGC). An analysis of the presented notions of online communities revealed shared constructs used by researchers in defining their OCs, as presented in Table 2.1: The Constructs of OC Definitions. Chapter 2. Literature Review 27 Rheingold 1993 Maloney- Krichmar & Preece 2005 Faraj et al. 2011 Schneider et al. 2013 Velasquez et al. 2014 Collective Social aggregations A group of people with a common interest or a shared purpose Open collectives of dispersed individuals who share common interest Social groups – Technology Emerge from the Net Use of computer systems – Virtual Expression through information and communication technology Members Enough people – Members who are not necessarily known or identifiable A set of members – Interactions Public discussions long enough Interactions – – Interaction between users of that site Sense of Community Sufficient human feeling A sense of togetherness – – – Purpose To from webs of personal relationships Support and mediate social interaction Attend to their individual and collective welfare Contributions of a varying extent to a common activity and/or good Dependence on user-generated content Policies – Governed by policies (tacit assumptions, rituals, protocols, rules and laws) – According to behavioural scripts – Table 2.1: Constructs of OC Definitions • Collective construct represents groupings, gatherings or collections of people that form the social system of an online community. People within OCs share a common interest based around a specific domain and they are not limited by geographical boundaries and time zones, according to Faraj et al. (2011) and Johnson (2001). Chapter 2. Literature Review 28 • Technology construct represents CMC that enables the development and maintenance of online or virtual communities using the Internet. The unique aspects of CMC-aided social interactions include alteration and archiving of any user-generated content (UGC), as well as anonymity and accessibility, that additionally empowers individuals to actively participate in an OC (Kozinets, 2010). • A minimum number of active Members is necessary for OC sustainability, while identities of members are not necessarily exposed publicly and can remain anonymous. • In order to provide a collective experience, foster a sense of belonging and allow for UGC, Interactions between members of a community are necessary (Kang, Tang & Fiore, 2014; Kim & Shyam Sundar, 2014; Leyton Escobar, Kommers & Beldad, 2014). According to Rheingold (1993), interactions in OCs are public and ongoing. Consequently, a “minimum number of interactions and exposure over time” is necessary to develop a sense of community (Kozinets, 2010: p.9). OCs have unique characteristics of anonymity and accessibility that provide “distinctive styles of interaction” (Kozinets, 2010: p.25) not found in offline groups. • The notion of a Sense of Community (Peterson, Speer & McMillan, 2008:p.62; Blanchard & Markus, 2004:p.67) or Sense of Togetherness (Preece, 2000:p.10), refers to the feeling of belonging, security and attachment, that individuals experience towards a group in OCs. This sense of community (SoC) is characterised by four primary components, as defined by McMillan and Chavis (1986): Chapter 2. Literature Review 29 • Membership: the feeling of belonging and personal connectedness; • Influence: the sense of making a difference to the community and its members; • Integration and fulfilment of needs: the expectation of having one’s needs met by the community’s resources; • Shared emotional connection: the belief and commitment, that members share and will continue to share, forming a history together. While researchers acknowledge the importance of SoC in OCs, it can be argued, that not all social computing tools or sites are conducive to building a sense of togetherness. For instance, some OC types, such as question and answer (Q&A) forums (Anderson et al., 2012) and ‘weak tie’ communities (Preece & Maloney- Krichmar, 2003), are not explicitly designed for relationship-building. Velasquez et al. (2014) excluded SoC from their definition of OC to encompass social media, social network sites and social computing platforms under a single umbrella. This allowed freeing “a canonical definition of the differences between formats” (Velasquez et al., 2014:p.23). However, Robert Kozinets (2010) posits, that there is no type of OC, where “deep and meaningful personal relationships cannot be built” (Kozinets, 2010:p.32). • According to researchers, the Purpose of OCs involves the development of personal relationships, provision of support to its members, creation of social capital on an individual and collective level and sharing of knowledge through the production and consumption of user-generated content (UGC). Velasquez et al. (2014) identifies a strong dependence on UGC as a defining characteristic of OCs, while Rheingold’s (1993) original definition places greater emphasis on the social Chapter 2. Literature Review 30 engagement between members. Kozinets (2010) notes, that OCs display a variety of usage patterns. • Policies of OCs refer to the “tacit assumptions, rituals, protocols, rules and laws”, that each community displays (Maloney-Krichmar & Preece, 2005: p.203). While communities can have different governance and moderation systems, there are implicit norms and behavioural scripts, that members must adhere to in order to be accepted. After conducting a thorough literature review and adopting a pragmatic approach, this study proposes a definition that comprehensively summarises the various OC constructs previously presented. According to this definition, an OC must possess the following minimum characteristics: • An open collective comprising dispersed, active and mostly anonymous individuals; • Members are expected to interact online according to accepted behavioural scripts and protocols; • Members share a common interest and have the ability to form social connections, foster a SoC and attend to their personal and collective well- being, while creating and consuming UGC. The presented definition of OCs encompasses a wide range of groups, from small, tight-knit networks to larger, more dispersed communities with millions of members. Regardless of their size, OCs are places, where people come together to converse, exchange resources, build knowledge, learn and collaborate. Members join OCs primarily because of other active members and their UGC. Participation is key to the success of any OC, whether through sharing resources, creating discussions or engaging with others at any level of skill or knowledge. Joining an OC is typically low cost, Chapter 2. Literature Review 31 with online users having the option to lurk, participate, leave at any time or even engage in a one-off transaction (Andrews, 2002). Active participation in an OC can provide significant benefits and serve a variety of purposes for users, including social support and networking, sharing and building knowledge, medical and emotional support, recreational and creative pursuits, professional development and learning, technical assistance and customer support, engaging with brands and making transactions (Malinen, 2015; Petrič, 2014; Ridings & Gefen, 2004; Butler et al., 2014). Kozinets (2010) proposed four general categories of OCs, that satisfy users basic needs: • Cruising: Examples include massively multiplayer online game (MMOG) spaces like League of Legends, traditional and webcam-based chats, such as Chatroulette and similar. These are weak-tie OCs, that satisfy entertainment and informational exchange needs. • Bonding: These include social networking sites (SNS) like Facebook, that fulfil members social needs and promote long-lasting relationships. • Geeking: Such OCs include newsgroups, website forums, social content sites and blog platforms similar to Reddit and Tumblr. Such OCs provide information about a particular set of activities, but do not necessarily engage users in a meaningful social relationship. • Building: Represented by interest groups on SNS and wiki-type platforms, such as Wikipedia and IMDb (Internet Movie Database). These mostly satisfy informational needs and allow in-depth discussions, evaluations, creation of content/products and building relationships. Robert V. Kozinets’ four-pillar categorisation is a useful tool for classifying OCs. However, enterprise-owned OCs (further discussed in Section 2.4) have different goals Chapter 2. Literature Review 32 and therefore require a separate typology. A study of IBM Connections Communities by Muller et al. (2012) examined 188 enterprise communities and placed each OC into one of the following categories based on Wenger’s earlier work (Wenger, McDermott & Snyder, 2002): • Community of Practice: A group of people who share information about events and content related to a shared interest or practice. • Team: A group of people working towards a shared goal for a particular client, project or business function. • Technical Support: Support boards aimed at answering technical questions. • Idea Lab: Communities in which users brainstorm around a set of questions and issues with a tight focus on a topic and with a short timeframe. • Recreation: Communities devoted to recreational talks and activities. Both typologies show similarities in terms of OC type definitions. Recreation OCs can be both Bonding and Cruising communities, that focus on social ties or entertainment. Team and Idea Labs are Building communities are centred around deliverables, where both informational and relational interactions are possible. Technical Support and Community of Practice OCs can be viewed as Geeking communities, where members can get in-depth information about a particular set of activities. Distinguishing and assuming the community types help to understand what drives user participation, which is crucial for assessing OC’s viability, nurturing existing communities and building new ones (Iriberri & Leroy, 2009; Muller et al., 2012; Leimeister, Sidiras & Krcmar, 2004). Knowing the current development stage of each community is equally important, since OCs are never static. Evaluations should account not only for the type, but also the structure and age of the community. Iriberri and Leroy (2009) proposed a life-cycle Chapter 2. Literature Review 33 model (Figure 2.1), that encapsulates four stages of online community development: inception, creation, growth, maturity and death. As OCs go through each stage, members’ needs change, requiring a different set of tools, features, mechanisms, technologies and management techniques. CMs and OC architects should identify these needs at each stage and adapt accordingly, to better support the members and the community. Fig. 2.1: OCs’ Life-Cycle (Iriberri & Leroy, 2009:p.14) The Inception stage encapsulates the future OC vision, where a certain need for one is identified. Creation of an OC begins with the technological components, that are put in place, as a support system for future members. When enough members have joined, they continue with defining their identity and selecting their community roles, as well as the common vocabulary. This resembles the Growth stage, that requires introduction of new rules, roles and establishing a common identity among OC members. When an OC gains Maturity, more formal moderation and policies are put in place, such as the introduction of rewards for creators of UGC and support for subgroup formations. The authors note, that many communities stay in this state for long periods of time. If an OC sees a decrease in member participation together with Chapter 2. Literature Review 34 an increase of user turnover rates, it will have a negative effect on the SoC and member satisfaction rates (Malinen, 2015). When a community cannot satisfy its users and provide enough value, it will eventually enter its Death stage. While forming understanding of OCs, their types and development stages are particularly relevant in relation to making OC evaluations, as these cannot be carried out without understanding the nature of user participation. 2.3 Participation in OC The involvement of users is the primary step towards establishing an OC. Such communities rely heavily on their members to provide mutual benefits and serve a larger audience at the community level (Butler, 2001; Wasko & Faraj, 2005; Hung & Chien, 2015). A participant is defined as anyone, who has engaged with an online community in any capacity (Malinen, 2015). Member participation in OCs is primarily motivated by various factors, such as informational needs, the desire to help others, the ability to identify with a group, the propensity to form friendships, the fulfilment of intrinsic needs, as well as the achievement of reputational, career and other extrinsic objectives (Ridings & Gefen, 2004; Roberts, Hann & Slaughter, 2006; Wasko & Faraj, 2005). The ability to maintain member engagement and satisfaction is referred to as OC resilience (Butler et al., 2014). Community resilience is only possible with a “critical mass” of members present in an OC (Markus, 1987), who continue to participate in discussions, even if the topics do not pertain directly to their personal interests (Butler, 2001; Wasko & Faraj, 2005). OCs with a niche, that is too narrow or a size that is too small, may struggle to attract enough members to sustain the community (Cummings, Butler & Kraut, 2002) and may provide limited value to their members Chapter 2. Literature Review 35 (Blanchard & Markus, 2004; Ridings & Gefen, 2004). Since OCs are fluid in nature (Faraj et al., 2011), participation is an ever-evolving process, that must be ongoing. Otherwise, the OC will cease to exist (Butler, 2001; Preece & Maloney-Krichmar, 2003) In current discourse, participation in OCs is typically classified as active or non-active. A lack of engaged members can hinder the growth and sustainability of an OC, while an increase in non-active members may limit its potential to attract new ones (Ren et al., 2012; Koh et al., 2007; Wasko & Faraj, 2005). On the other hand, OCs with a large pool of active and committed members are better equipped to thrive. Such members generate and share high-quality UGC, engage in discussions, support peers, uphold community values and policies, take on administrative roles (Butler et al., 2003; Ackerman & Palen, 1996; Moon & Sproull, 2008; Ren et al., 2012). Long-term active participation by community members serves as an example and encourages others to follow suit, thereby building their expertise and fostering SoC (Lee, Park & Han, 2014; Ren et al., 2012). The concept of Legitimate Peripheral Participation (LPP) explains how users progress from new members, who learn community norms, to those, who actively contribute and establish broader social relations (Lave & Wenger, 1991). As members increase their expertise, their overall activity in the OC also increases. This progression from silent UGC consumers (“lurkers”) to core community users (“gurus”) is captured in Kozinets’ participant progression model for OCs, as depicted in Figure 2.2. Over time, as members engage more frequently and meaningfully, they develop task-oriented and goal-directed informational knowledge and strengthen their social and cultural ties within the community. This model provides insights into the mechanisms behind the development of expertise and status in OCs. (Kozinets, 2010). Chapter 2. Literature Review 36 Fig. 2.2: Developmental Progression of Participation in OCs (Kozinets, 2010:p.28) Robert Kozinets’ work bears resemblance to the Reader-to-Leader Framework proposed by Preece and Schneiderman (2009), which outlines how OC members undergo a process of socialisation and transition further, as their membership status matures. Each stage, that a user passes through, is characterised by a set of motivations and user behaviours. The first user type is the Reader, which pertains to returning users, registered or otherwise, who primarily engage in content consumption and tend to be passive in their interactions with other users. Such users are typically motivated by informational and recreational interests. As they become more involved in the community and consume more content, they may progress to become Contributors. Contributors are characterised by self-interest motivations, as they engage in creating UGC and seek recognition within the OC, as producers of high-quality content. Although Contributors focus on UGC creation, they tend not to interact significantly with other users. In contrast, Collaborators seek out contact with other OC users, with a shared sense of belonging and identity, acting as a significant motivator. These users collaborate, share and interact with others to the point of developing strong personal ties. Finally, gaming-based community, the learning of the social norms or game rules, or the mastering of the online environment itself, might be the primary objective. In those cases, topics about which people share information might be more personal in nature or more about the characteristics of the social or constructed environment. From there, the participant might follow a path to the learning of cultural norms, and the broadening and spreading of social relations throughout the various extended arms of the online community. Regardless of the medium or exact pathway to participation, the theory suggests that, over time and with increasingly frequent communications, the sharing of personal identity information and clarification of power relations and new social norms transpires in the online community – that social and cultural information permeates every exchange, effecting a type of gravitational pull that causes every exchange to become coloured with emotional, affiliative, and meaning-rich elements. This emotional, affiliative element – its social psychological origins and its social values – has been recognized repeatedly in research. Research using survey responses and structural equation modelling by McKenna and Bargh (1998) revealed that many respondents had, as a result of their online community participation, come out to their families and friends about a stigmatized aspect of their identity for the first time in their lives. Because of their online community experiences with others who shared their own stigmatic status, they considered themselves less different, benefited from the increase in self-acceptance, and felt less socially isolated. Similarly, another study of online support groups for people with serious and often stigmatized illnesses such as alcoholism, AIDS, and forms of cancer, established the benefits of online communities (Davison et al. 2000). For those who sought out similar others under conditions of great anxiety and uncertainty, the anonymity and accessibility of these communities has been a virtual godsend.A range of studies also suggest that online communities have considerable stress reduction, self-acceptance, and informational value, even for people who have illnesses and conditions that are not stigmatized, such as diabetes or hearing impairments (see, e.g.,McKay et al. 2002). Ethnographically studying the phenomenon in a subcultural context from a symbolic interactionist frame, the findings ofWilliams and Copes also reinforce the utility of the online communal forum for those who feel disenfranchised or marginalized. Using NETNOGRAPHY28 FIGURE 2.1 DEVELOPMENTAL PROGRESSION OF PARTICIPATION IN ONLINE COMMUNITIES (ADAPTED FROM KOZINETS 1999) Kozinets-3963-Ch-02:Kozinets Sample.qxp 08/09/2009 11:25 AM Page 28 Chapter 2. Literature Review 37 the Leader user type has the fewest number of participants, as it represents the final stage of the user socialisation model. Leaders take voluntary responsibility for defining and enforcing community norms and policies, seek solutions to any problems that may arise in the OC regarding its users or content and help in building the overall community identity. Fig. 2.3: The Reader-to-Leader Framework: Motivating Technology-Mediated Social Participation (Preece & Schneiderman, 2009:p.16) Figure 2.3 of the Reader-to-Leader Framework shows, that users do not progress in a linear fashion, as they have the option to move backwards, skip steps or disengage completely. This principle is also applicable to a similar framework developed by Turner and Fisher (2006), which identifies four typical roles in technical support forums and newsgroups (Turner & Fisher, 2006). The first category, known as Questioners, consists of participants who may be passive consumers of UGC (similar to Readers) or active members, who ask technical questions in OCs. The Answer Person category is equivalent to Contributors, as they are active members, who assist others with technical queries in the hope of being recognised for their contributions. Chapter 2. Literature Review 38 Community Managers represent a third category of participants, who are responsible for governing and nurturing an online community. Unlike the Reader-to- Leader Framework, this role has no equivalent category. In enterprise-owned OCs, that serve specific organisational and commercial interests, CMs play a critical role in ensuring, that the interests of the host organisation are met. They engage in significantly more community building work than any other member type (Butler et al., 2003; Lazar & Preece, 2002). The final category, Moguls, refers to community leaders (CLs) and lead users (LUs), who are experts, that possess the knowledge and skills to answer the most complex TS queries. According to Michael Wu (2012), Q&A Person categories make up 90% of all participants (Audience), while CMs represent 9% of OC members (Editors) and Moguls are the unique 1% (Creators), who influence community dynamics and content generation levels (Wu, 2012). 2.4 Enterprise-owned TS OC Enterprise-owned OCs are established with the aim of generating revenue for its owners (Hunter & Stockdale, 2009). Such OCs offer a range of benefits, including improved customer service, product feedback, new product development based on customer input, building a stronger brand reputation and customer awareness (Wasko & Faraj, 2005; Carillo, 2017). Enterprise-owned OCs attract customers, who exchange information about the products and services provided by host organisations (Hunter & Stockdale, 2009). A well-designed enterprise-owned OC can help frustrated customers, enhance customer experience and knowledge, provide the necessary tools, receive feedback and address important product issues (Wu, 2012). The use of enterprise-owned OCs is not limited to business-to-consumer (B2C) Chapter 2. Literature Review 39 applications only. Corporate communities are also adopted to establish relationships with business partners in business-to-business (B2B) contexts, but also connect employees in business-to-employees (B2E) OCs to promote collaboration and increase operational performance (Dubé, Bourhis & Jacob, 2005). This trend has driven large software corporations, like IBM and SAP, to establish dedicated CM teams and to allow researchers to analyse data directly from their corporate OCs (Wagner et al., 2014; Rowe & Alani, 2012; Muller et al., 2012). A noteworthy approach for leveraging enterprise-owned OCs is by providing a platform for peer-to-peer support, where customers can collaborate and assist each other in resolving queries related to products and services offered by the host organisation. This kind of corporate OC is referred to as Customer-to-Customer communities (Evans & Cothrel, 2014). It is worth mentioning, that enterprise-owned OCs can serve multiple purposes simultaneously by having designated areas or forums that facilitate Business-to-Customer (B2C), Customer-to-Customer (C2C) and Business-to-Employee (B2E) functions (Muller et al., 2012; Yardi & Poole, 2009). An example of such a dedicated area is TS or Q&A forums, which primarily aim to enable customers to generate and consume technical support knowledge (Gu & Jarvenpaa, 2003). In enterprise-owned TS or Q&A OCs, customers can ask questions, receive prompt and accurate responses from fellow customers and/or company employees (Aumayr & Hayes, 2017a). Based on the reviewed literature sources, TS OCs are one of the most popular and effective means of acquiring information on the internet (Liu, Bian & Agichtein, 2008; Nam et al., 2009). As an OC becomes populated with UGC, such as answers provided by other customers or the company’s technical staff, the overall knowledge increases. The generated knowledge can then be filtered, archived and classified on the firm’s Chapter 2. Literature Review 40 servers for future reference (Gu & Jarvenpaa, 2003) and become part of a knowledge base (KB). TS forums are mainly structured around technical topics and consumer products to enhance user logistics and usability. These forums can be easily found through popular search engines and provide a low-cost avenue for customers seeking support (Hung & Chien, 2015). Discussions in such OCs are typically more factual and less discussion-oriented than in other types of online communities (Yardi & Poole, 2009). TS or Q&A communities also capture the entire problem-solving process, along with the results of each query: the questions, answers and whether a suitable solution has been found (Liu, Bian & Agichtein, 2008). This enables researchers to analyse TS or Q&A OCs successfully, since the outcome of each thread is documented and known (Moon & Sproull, 2008; Aumayr & Hayes, 2017a; Anderson et al., 2012). However, feedback mechanisms differ between TS and Q&A communities. TS OCs are primarily designed as discussion boards with a classic thread structure, where all posts are arranged chronologically. The feedback is presented in the form of member comments or through a limited feedback system, which indicates whether a thread was resolved or not, with the shortest path to a solution or helpful answer (Moon & Sproull, 2008). On the other hand, Q&A communities have more advanced feedback systems, where voting and reputation mechanisms are central to their design (Anderson et al., 2012). Members can rate each answer, answers receive approval or disapproval from the community and solutions are displayed in non-chronological order, starting with the most popular ones (Liu, Bian & Agichtein, 2008). The complete set of answers then creates an in-depth examination of issues, providing added value to consumers, compared to viewing a single best answer in a classic thread (Anderson et al., 2012). To demonstrate the process of asking a typical question in a Q&A community, researchers Liu, Bian and Agichtein (2008) defined a Community Question Answering (CQA) Life-cycle, shown in Figure 2.4. Chapter 2. Literature Review 41 Fig. 2.4: A Simplified Life-cycle of a Question in a Typical CQA Site (Liu, Bian & Agichtein, 2008:p.3) The same life-cycle can also be applied to TS OCs, with the exception, that members usually have limited or no voting abilities to rank replies to the original question. Instead, a moderator typically selects a solution or the most helpful answer. Community managers/moderators (CMs) play a crucial role in TS and Q&A OCs, often tasked with both technical maintenance and community building activities. CMs are typically OC experts with the ability to direct members to existing resources, appropriate discussions and arrange the environment to suit the needs of a conversation (Gurzick et al., 2009). They facilitate interaction in a community by promoting member discourse, facilitating ongoing discussions and ensuring the discourse is conducted appropriately and according to OC policies. CMs can reduce the impact of negative behaviour, such as teasing, abusing and information overload, by manually or automatically filtering and moderating discussions based on their content (Ren & Kraut, 2014). OC managers/moderators should maintain a healthy balance between the quality and quantity of information. Over-managing an OC can lead to trust issues with the rest of OC members and turn participants away from contributing (Moon & Sproull, 2008; Ren & Kraut, 2014). Losing participants, especially those recognised as CLs and LUs, can negatively affect OC operations and sustainability. Chapter 2. Literature Review 42 CLs are usually formally-recognised CMs, which provide guidance and management to build and maintain the relevance and strategic importance of the community within an organisation (Bourhis, Dubé & Jacob, 2005:p.26). They can be either employee curators or users approved by critical stakeholders to moderate the community within an enterprise-owned OC. On the other hand, lead users (LUs), also known as superusers or gurus, are looked upon for guidance and leadership in developing the community’s mission and purpose (Bourhis, Dubé & Jacob, 2005:p.25). LUs can receive special badges such as Most Valuable Participant (MVP) to differentiate themselves from other users within an OC (Shih, Hu & Farn, 2010; Leupp, 2016). Both CLs and LUs in enterprise-owned OCs can help in reducing support costs, scaling community operations, serving as evangelists, participating in product development and testing, taking on moderation tasks and providing other various benefits to the OC and the host organisation (Wu, 2012). They also contribute to UGC formation and discussions, help in organising and curating UGC, maintain a positive and friendly environment, foster connections, nurture new members, advertise externally and minimise disruptive behaviour and maintain OC infrastructure (Matthews et al., 2013). Frequent participation by CLs and ULs results in quick response times and comprehensive solutions provided by experts, which is desirable in TS and Q&A communities (Aumayr & Hayes, 2017a; Wang & Lantzy, 2011). It is recommended that some replies and solutions in TS OCs come directly from technical support staff, while participating employees and top-contributing CLs are rewarded by OC owners for their voluntary work (Gu & Jarvenpaa, 2003). Although CMs and LUs often work closely together, their participation nature differs. Formal employees participate in an OC based on economic incentives, formal training, contracts, established corporate productivity measurements and formal supervision Chapter 2. Literature Review 43 procedures (Moon & Sproull, 2008). Conversely, LUs and other OC contributors, including non-employee CLs, rely on non-economic incentives, peer reviews and community norms to maintain volunteer commitment and OC’s UGC quality (Moon & Sproull, 2008). Given how crucial enterprise-owned OCs are in establishing relationships online with various parties (Business-to-Consumer, Consumer-to-Consumer, Business-to- Employee and Business-to-Business), companies aim to ensure the long-term success, sustainability and survival of their OCs on an ongoing basis (Bourhis & Dubé, 2010). 2.5 OC Success In the late 1990s, the concept of OC success became popular among researchers from various disciplines, who tried to identify its key determinants and indicators. However, after almost three decades of research, there is still no scientific definition or unified vision of what OC success represents. This is partly because the meaning of success varies among different stakeholders involved in any type of OC and depends on the community’s type, size and age (DeLone & McLean, 2003; Preece, 2001; Iriberri & Leroy, 2009). Moreover, communities do not operate in isolation and OC success depends not only on internal processes, such as strategy, policies and members behaviour, but also on many external forces, such as competitive communities, viral trends, spam and Distributed Denial of Service (DDoS) attacks (Ridings & Wasko, 2010; Wang, Butler & Ren, 2013; Rowe & Alani, 2012). The evaluation of OC success is not directly comparable to some of the previously proposed performance measures used in socio- economic or organisational contexts, thus the development of new measures is required (Wagner et al., 2014). This ongoing process receives contributions from both Chapter 2. Literature Review 44 practitioners (Lithium Technologies, 2015; Evans & Cothrel, 2014; Wu, 2012) and researchers (Lili & Rong, 2013; Aumayr & Hayes, 2017b; Matthews et al., 2013) indicating, that theory development in the area is still incomplete. Some of the widely-adopted models used by researchers in the context of OC success came from the information systems (DeLone & McLean, 2003) and socio-technical (Preece, 2001) perspectives. DeLone and McLeans’ Information System Success (ISS) model, originally formulated in 1993 (DeLone & McLean, 1992) and revised ten years later (DeLone & McLean, 2003), allowed OC researchers to recognise OCs as web- based information systems (Wachter, Gupta & Quaddus, 2000). According to ISS, OCs are technical entities, that should fulfil six factors to be perceived as a successful information system (IS): system quality, information quality, user satisfaction, service quality, usage intention and net benefits. The model and its separate constructs, such as system and information quality, were adopted and extended by OC researchers to include loyalty, leadership, usefulness of content, member’s satisfaction and SoC (Lin & Lee, 2006; Zhang, 2010; Lili & Rong, 2013; Koh et al., 2007). However, critics argue, that OCs are not limited to being a CMC-enabled network of individuals that fulfils the reviewed ISS factors. System quality, service quality or usage intentions are not directly comparable between IS systems and OCs. These factors do not address the complex interdependencies, that exist in adaptive and open systems, such as OCs (Wagner et al., 2014; Carillo, 2017). Viewing OCs purely through these IS prisms neglects the importance of human interactions, which are integral to its nature and define OCs’ existence. While an enterprise-owned OC can provide quality service as a CMC platform, users upset by the actions of OC sponsors may choose to leave the community (Thurrott & Nguyen, 1999). This can have a negative impact on a community and damage the Chapter 2. Literature Review 45 sponsors reputation, as an intense focus on external expectations by OC owners can potentially contradict and undermine social interactions between members (Wagner et al., 2014). Some measures used in organisational research, such as performance, are not always relevant in the context of OCs either. Users participate voluntarily and are not tied to an OC in a way, that is common in traditional organisations with more formal structures and entry/exit barriers (Wang & Lantzy, 2011). Fig. 2.5: A Joint Representation of ISS model (DeLone & McLean, 2003) [blue] and Sociability & Usability Framework (Preece, 2001) [green] explaining OC success According to Jenny Preece’s (2001) Sociability and Usability Framework, the success of OCs depends on two key factors: technical factors related to Usability and social factors related to Sociability. The framework considers interactivity, reciprocity, quality of contributions, number of OC members, participant types, flaming and uncivil behaviour, information validity, as suitable measures of a successful community from a sociability perspective. In addition, usability measures, such as user interface (UI) adoption speed, UGC consumption/generation rates, UI user satisfaction levels, UI stickiness and error rates contribute to the success of OCs. Chapter 2. Literature Review 46 Hew (2009) extended Preece’s measures by including willingness to share knowledge and UGC, diversity of views, relevant discussions, a friendly environment and fast response times. Iriberri and Leroy (2009) identified member contribution and quality of relationships among users, as the most common determinants of OC success. Their proposed measurements include: community size, participation rates, volume of contributions, the extent of contacts between members, member satisfaction and quality of relationships. Other success factors found in the literature on OC success include responsiveness, individual-community interactions, member turnover, user acceptance, member participation and engagement, survivability. For instance, responsiveness refers to the percentage of new threads, that receive at least one reply, while individual-community interactions entail the ability to reply to members queries and keep them satisfied (Arguello et al., 2006; Wagner et al., 2014). Similarly, member turnover refers to the percentage of an OC’s userbase, that leaves in a given period, while user acceptance relates to the perceived usefulness of OCs to users (Lin, 2007). Member participation and engagement levels, as well as the ability of OCs to continually exist, also play crucial roles in their success (Lu, Phang & Yu, 2011; Raban, Moldovan & Jones, 2010). In the context of Q&A OCs, some success factors are specific to this type of community. These include question-solving performance, number of responses, number of unique thread participants and the presence of LUs. However, certain practices can have negative effects on Q&A OCs, such as the use of URLs in responses, long answers and long response times, which can be detrimental to health (Aumayr & Hayes, 2017a). On the other hand, Aumayr & Hayes (2017a) found, that well-structured questions and well-written topic titles can have a positive effect on question-solving performance. Chapter 2. Literature Review 47 The factors, that have been discussed regarding OCs are mainly focused on the members, such as the amount and quality of content, responsiveness and user satisfaction. However, success for OCs is multidimensional and varies depending on the specific goals of community stakeholders, such as sponsors, moderators and participants, as well as the type of OC (e.g., Q&A, social support, knowledge creation) (Larson & Watson, 2011; Preece, 2001; Aumayr & Hayes, 2017b). From a business perspective, OC success is related to the fulfilment of the strategic intents of a host organisation. These include productivity, cost-effectiveness, quality of outputs, amount and quality of inputs, the presence of a critical mass, matching community and business needs (Qin, Cunningham & Salter-Townshend, 2016; Bughin & Hagel, 2000; Ransbotham & Kane, 2011; Spaulding, 2010). In addition, OC moderators’ success is tied to the integration of new participants, member retention, adherence to rules and policies (Iriberri & Leroy, 2009; Wang & Lantzy, 2011; Sangwan, 2005). It is important to note, that the reviewed success factors do not provide a comprehensive list of OC success determinants and measures. Many of these factors were developed theoretically and have not been evaluated against objectively measurable OC success, which limits their usefulness in capturing the current state of a given OC by community managers and owners (Aumayr & Hayes, 2017b). Furthermore, different types of communities may define success based on their own desired outcomes. For example, Q&A community members may prioritise quick and quality solutions to their questions (weak ties), while social support OC members may seek longer and more personal discussions to gain empathy and positive emotions (stronger ties) (Aumayr & Hayes, 2017b). Overall, it is possible to conclude, that Chapter 2. Literature Review 48 defining OC success is a complex problem that is context-dependent, where evaluations would vary based on various internal and external factors. 2.6 OC Health OC managers and owners have expressed the need for a more comprehensive approach to assessing OCH (Lithium Technologies, 2009; Rowe & Alani, 2012; Wagner et al., 2014; Wang & Lantzy, 2011; Carillo, 2017). OC health places emphasis on the vitality and performance of an OCs systems at any given time (Wang & Lantzy, 2011). This concept draws upon the metaphor of an OC as a living organism composed of organs and subsystems that evolve, age and eventually cease to function (Wagner et al., 2014; Morgan, 2006). This metaphor suggests, that OC health can be directly likened to human health, where an individual’s physical and mental wellbeing is evaluated based on internal metrics such as blood pressure, temperature and other factors, including the ability to cope with various stressors. Thus, OCH can be viewed as “the ability to function effectively, cope adequately and adapt appropriately in response to internal and external stimuli” (Carillo, 2017: p.7). Unlike the static success-failure dichotomy, that describes online community success as either successful or not, OCH represents the degree of physiological wellness of an OC, with both current state and future prognosis (Wagner et al., 2014). This enables OCH diagnosis, in theory, to inform proactive measures to address current issues and prevent future ones. Despite the emerging nature of online OCH (Wagner et al., 2014), there are still gaps that need to be addressed by researchers. These gaps include the interplay between structural and social components, the impact this interplay has on the functioning of OCs (Carillo, 2017) and how OCs interact with their external environment (Wang, Butler & Ren, 2013). To tackle these issues, Carillo (2017) examined the theoretical Chapter 2. Literature Review 49 principles of living organisms and compared them to an OC, which represents a unique form of an open system. The author found, that when viewed through an organismic metaphor, an OC should meet the following conditions (Carillo, 2017: p.9): • Exist within an ecosystem comprised of other OCs, individuals, organisations and entities; • Continuously produce outputs from throughputs and inputs; • Be composed of evolving and coexisting subsystems that operate within the boundaries defined by the OC type; • Permit the free exchange of information and energy beyond their perimeter; • Constantly adapt to positive and negative impulses through self-regulation; • Require a consistent inflow of building materials, energy and information for internal sustainability; • Achieve sustainability in diverse conditions and through various means; • The internal complexity of an OC corresponds to the impact of the external environment. The exchange of metabolites with the external environment involves several inputs, such as active participants, informational and knowledge resources, technical assets and financial investments coming from the hosting organisation. Once inputs are processed, throughputs can take forms of an improved/optimised solution to the problem (Q&A), code additions (Open-Source Projects), evolution of ideas (Idea Labs) and other artefacts. These co-created artefacts may include support KB documents, software artefacts and ideas awaiting implementation. In addition to the described components, Carillo contends, that an OC can be represented by one Environmental suprasystem composed of five subsystems (see Figure 2.6): Strategic, Production, Human, Technological and Managerial. Chapter 2. Literature Review 50 Fig. 2.6: Online Communities as Organisms (Carillo, 2017:p.13) The Environmental suprasystem determines the evolutionary processes happening in an OC. As OCs are not operating in isolation to the rest of the Internet, their life and operations are impacted directly or indirectly by OCs in competition, organisations, governments, other various parties and the society as a whole (Hung & Chien, 2015). In addition, an enterprise-owned OC is impacted by both sponsors and their competitors. The topic of environmental pressures, applied to OCs, is largely ignored in IS research (Wang, Butler & Ren, 2013). The Strategic subsystem defines strategic decisions, which are made in response to various stimuli, coming from the environment or changes in the environment itself. The purpose of the OC has a direct effect on how this subsystem is structured, how it functions and the level of its complexity. The Production subsystem is centred on converting inputs into throughputs, outputs and encapsulates all structural aspects, processes, procedures, Chapter 2. Literature Review 51 coordination, information flow, rules, policies and other elements, that make the conversion possible. Individuals, involved in an OC, make up the Human subsystem, which is tightly tangled with the production subsystem, as OC members play an integral part in the process of generating value (outputs). The Human subsystem relates to member attitudes, motivations, values, status, norms, experience, enrolment, interactions, culture, friendship and other user characteristics and aspects. Researchers have previously noted, that willingness of members to participate and enact roles within an OC, affects productivity and longevity of the group, i.e., adds to OC health and subsequently success (Gleave & Welser, 2009; Cothrel & Williams, 1999). An OC, as a CMC system, is not possible without the Technological subsystem, which consists of a platform, that supports synchronous or asynchronous interactions, intake and distribution of UGC (images, video, posts, etc.), provides a user-friendly interface and uncluttered visualisations. The level of sophistication of the Technological subsystem depends on the complexity of the Production subsystem, which it supports. The last element of the environmental suprasystem is Managerial subsystem, which is responsible for linking and balancing out the rest of the subsystems through governance, which includes recruitment, conflict resolution and leadership style aspects. Carillo (2017) notes, that overall OCH depends on the balance of the internal ecosystem, where misalignment between two or more individually healthy subsystems can lead to a dysfunction. Thus, the ability to cope with changes inside and outside of the OC’s boundary, is the main indicator of a healthy community and its ability to evolve in the future. The majority of studies on success and health of OCs look at communities in isolation (Lithium Technologies, 2009; Rowe & Alani, 2012; Wagner et al., 2014; Wang & Lantzy, 2011; Aumayr & Hayes, 2017b), thus, ignoring external influences on an OC, its context and history, when proposing new constructs and measures of success/health. (Carillo, 2017). Chapter 2. Literature Review 52 2.7 OC Success-Health Relationship Previous studies have largely ignored the link between the concepts of OC success and health. Some authors have used these terms interchangeably (Xu et al., 2013; Rowe & Alani, 2012), while others have not specifically established relationships between them (Wang & Lantzy, 2011; Aumayr & Hayes, 2017b). Although a limited number of publications have stated, that health is a necessary condition for success in OCs, but not a sufficient one (Wagner et al., 2014; Lithium Technologies, 2015), enterprise- owned OCs further simplify this link. Lithium Technologies proposed the following definitions: • Healthy OCs fulfilling the needs of its members (i.e., consumers); • Successful OCs fulfilling the needs of the business (i.e., the host organisation). (Lithium Technologies, 2015). Meeting the needs of OC members is not always a guarantee of a positive impact on the host organisation’s performance. An OC can be both healthy and successful, when customers are satisfied, which in turn, allows an OC to align with its sponsor’s strategic goals (Lithium Technologies, 2009). However, there have only been a few attempts to establish the relationship between OC health and success, which have led to the works of Joseph Cothrel (1999, 2000), the lead Chief Community Officer at Lithium Technologies (Bloomberg.com, 2018) and the publications of Anne Bourhis and Line Dubé (2005, 2010), that bear similarities to the Cothrel’s elements. Cothrel (2000) supports the vision, that OC success should be judged based on how it supports the host organisation’s business goals. The author defines success as a three-dimensional entity consisting of Return on Investments (ROI), Insight and Health, as shown in Figure 2.7. Chapter 2. Literature Review 53 Fig. 2.7: Three Dimensions of Community Measurement (Cothrel, 2000:p.19) Return on Investment (ROI) is a conventional financial evaluation tool, while Insight and Health are non-financial measures of success, that relate to knowledge and OC functioning. The ROI of OCs primarily utilises quantitative financial indicators, such as generating additional sales, reducing customer churn and saving on advertisements, as well as qualitative measures, like the quality of the knowledgebase, customer satisfaction and brand affinity (Cothrel & Schustler, 2013; Cothrel, 2000; The Community Roundtable, 2017; Oracle Corporation, 2012). Health factors are internal metrics, that pertain to OC activity, such as membership size changes, user churn rates and traffic inflow. Cothrel suggests, that an additional factor for health is active user fulfilment of various informal roles, such as experts, mentors, information sharers and even critics (Cothrel & Williams, 1999). CMs use these factors to monitor their OCs daily. Insight is the third success dimension, which examines how much knowledge can be derived from OC customer discussions and processes, including early issue identification, new product ideas, product and service improvement, generation of KB articles (Cothrel & Schustler, 2013). It is worth noting that some of Cothrel’s concepts were realised and implemented in practice by his employer, Lithium Technologies. For instance, the development of the Social ROI concept further expanded to include both the ROI and Insight dimensions Chapter 2. Literature Review 54 (Cothrel & Schustler, 2013). The company also designed a Community Health Index (CHI), which is a commercial diagnostic tool available to its customers (Lithium Technologies, 2009). Many authors cite Lithium’s white paper on CHI (2009), as the earliest source that introduced the concept of OC health (Wang & Lantzy, 2011; Rowe & Alani, 2012; Wagner et al., 2014; Carillo, 2017), overlooking some of Cothrel’s theoretical work from the pre-Lithium era. Bourhis and Dubé (2005, 2010) proposed, that OC success consists of two dimensions: Effectiveness and Health. They define Effectiveness, as the extent to which the organisation has met its objectives in establishing an OC, the value it created and the benefits it provided to its members (Bourhis & Dubé, 2010). The authors define Health, as the factors, that can be measured within an OC in computable terms, such as the level of online activity and member satisfaction. Both concepts, initially proposed by Cothrel and later by Bourhis and Dubé, describe success-health relationships in a similar manner. Cothrel’s original success concept (2000) evolved and transformed into two measurements: Social ROI (Cothrel & Schustler, 2013) and CHI (Lithium Technologies, 2009). Jive’s Measurement Framework (Jive Software, 2017) proposes a similar approach to measuring OC success, suggesting three dimensions: Community Adoption & Health, Perceived Value for members and Business Value for organisations. Table 2.2 provides a summary of all the approaches and how they relate to the success dimensions defined by Bourhis and Dubé (2010). Chapter 2. Literature Review 55 Components (Bourhis & Dubé, 2010) Measures used by Lithium (Lithium Technologies, 2009; Cothrel & Schustler, 2013) Measures used by Jive (Jive Software, 2017) Health Community Health Index (quantitative) Adoption & Health (quantitative) Effectiveness Social ROI (quantitative + qualitative) Perceived Value (qualitative), Business Value (quantitative + qualitative) Table 2.2: Components & Measures of OC Success To summarise, the Health dimension, as measured by CHI and other OCH models, refers to the physical wellness of an online community (OC), while the Effectiveness dimension, as measured by ROI and similar tools, refers to how well the OC contributes to the success of its various stakeholders, including OC owners, sponsors, members and other parties. Therefore, the relationship between OC success and health can be expressed as: SUCCESS = HEALTH + EFFECTIVENESS This approach differs from viewing health and success as being specifically member- focused and sponsor-focused entities respectively, which provides a more holistic perspective. The above formula can be used to expand upon the “Online Communities as Organisms” model developed by Carillo (2017) and discussed in the previous section. This model provides a shared OC taxonomy derived from the perspectives of OCs, IS, live organisms and serves as a “theoretical hub” (Carillo, 2017:p.15) for studying the mechanisms that govern the life and death of OCs. Chapter 2. Literature Review 56 The updated model, shown in Figure 2.8, introduces 3 new dimensions of OC Health, Effectiveness and Success. These dimensions draw inspiration from the Physical, Emotional and Social health dimensions found in humans (Carillo, 2017): • Physical health refers to how well the OC is capable of maintaining physiological homeostasis. • Emotional health refers to an individual’s capacity to cope through preserving a sense of mental coherence. • Social health relates to the ability of an OC to succeed in social life. The updated “OCs as Organisms” model (Figure 2.8) demonstrates three new dimensions, where: • Health corresponds to the well-being of the “organism” (OC); • Effectiveness encapsulates how well the inputs are converted into outputs by an OC; • Success relates to the overall ability to satisfy OC stakeholders’ needs and can be viewed from individual or combined perspectives of members and a host organisation. Chapter 2. Literature Review 57 Fig. 2.8: The Revised “OCs as Organisms” Model (Carillo, 2017) with Health, Effectiveness & Success 2.8 Measuring OCH At present, there are no publicly available and widely accepted instruments for diagnosing OCH. However, scholars have suggested various health indicators, that can be applied to a wide range of OCs (Hew, 2009; Lin & Lee, 2006; Lili & Rong, 2013; Preece, 2001; Wang & Lantzy, 2011; Rowe & Alani, 2012; Wagner et al., 2014; Aumayr & Hayes, 2017b). The Community Health Index (CHI), developed by Lithium Technologies, is the closest commercial attempt to creating a model for measuring OCH. The logic behind the model and the formula for calculations are presented in Lithium’s white paper (Lithium Technologies, 2009). The CHI has introduced the concept of OCH to the industry and has attracted the interest of multiple academics Chapter 2. Literature Review 58 (Rowe & Alani, 2012; Wang & Lantzy, 2011; Wagner et al., 2014; Aumayr & Hayes, 2017b; Carillo, 2017). Lithium suggests that a healthy OC should have positive monthly registrations, a critical mass of content, high traffic, thread depth, long-term member motivation, a friendly atmosphere, trust, and civil behaviour. During the CHI development, Lithium analysed various communities and aggregated 15 billion actions from 6 million users to identify diagnostic health factors. These factors include Responsiveness, Interaction, Liveliness, Members, Content and Traffic. The health function is calculated as the square root of the product of all diagnostic and predictive factors. The indicators are multiplied together and adjusted to create a CHI score ranging from 1 to 1000. However, if some factors have significantly small or large numbers, it can diminish the importance of other factors, raising concerns about the reliability of the final CHI score. A more comprehensive review of Lithium’s CHI is available in Appendix A. Despite many suggested success and health measures (reviewed in Sections 2.5-2.6 and CHI factors discussed above), only a few metrics have been tested empirically (Iriberri & Leroy, 2009; Leimeister, Sidiras & Krcmar, 2006). Furthermore, Carillo’s (2017) OCH model contains no information on how the health of each of the subsystems can be measured in empirical terms. Using Carillo’s (2017) “OC as Organisms” model as a guide, Table 2.3 examines the measures, that have been used in previous studies and are applicable in the context of enterprise-owned TS OCs. Each subsystem is matched with the appropriate health dimension and examples of suitable metrics proposed by various authors. Chapter 2. Literature Review 59 Subsystem Dimension Description Examples References Strategic Maturity Host organisation’s social and OC efforts Strategy, Leadership, Culture; OC Management, Content & Programming, Policies & Governance, etc. The Community Roundtable (2017) Production Generation UGC creation, including new threads and messages Number of new messages, threads, etc. Lithium Technologies (2009), Preece (2001), Wang & Lantzy (2011), Wagner et al. (2014), Aumayr & Hayes (2017b) Consumption Information quality and how much attention it generates Number of solved/closed threads, ignored threads, views (traffic), etc. Hew (2009), Lin & Lee (2006), Lili & Rong (2013), Preece (2001), Wang & Lantzy (2011), Rowe & Alani (2012), Aumayr & Hayes (2017b) Human Interactivity How many unique members participate in discussions across OC Number of unique users in threads, number of messages per thread, etc. Lithium Technologies (2009), Wang & Lantzy (2011), Wagner et al. (2014) Responsivene ss How long a user should wait for the OC to respond Time to first response, time to solution, etc. Lithium Technologies (2009), Wang & Lantzy (2011), Wagner et al. (2014), Aumayr & Hayes (2017b) Engagement The number of OC members registering and staying active Number of new registrants, active users, etc. Lithium Technologies (2009), Preece (2001), Schneider et al. (2013), Wang & Lantzy (2011), Wasko & Faraj (2005), Rowe & Alani (2012) Reliance The ability to trust people’s actions or what they say User cumulative ratings, registration age, etc. Lin & Lee (2006), Lithium Technologies (2009), Preece (2001), Wagner et al. (2014) Management Maintenance The general atmosphere of the OC, e.g., flaming and constructiveness of exchanges Number of moved/removed threads, banned users, etc. Hew (2009), Lili & Rong (2013), Lithium Technologies (2009), Preece (2001), Wagner et al. (2014) Technological Accessibility The ability to get the information required with ease Usability characteristics, number of created KB articles, etc. Preece (2001), Lin & Lee (2006), Wagner et al. (2014) Table 2.3: OC Subsystems, OCH Dimensions & Suggested Diagnostic Factors Table 2.3 presents examples of diagnostic factors, that are commonly used in OCH research and represent fundamental metrics integrated into popular OC hosting platforms. However, the Maturity dimension and its diagnostic factors, which are employed to analyse the Strategic subsystem, are an exception to this. In this research, Chapter 2. Literature Review 60 we will use The Community Maturity Model (The Community Roundtable, 2017) to determine the level of effort inve