84
85
also argue that in the case of experience goods consumers have to rely more on indirect cues from their social environment, such as the product-related behaviors of friends, because they cannot directly assess the quality of the product prior to purchase or usage.
Moreover, even within the category of experience goods there are subtle, yet important, differences that may affect the potential impact of eWOB. Blockbuster and Spotify differ, for example, with respect to their model of revenue (pay per view versus all you can eat
subscription, respectively). Consequently, we should expect that users in the subscription-based model are more likely to try out new content because there is no monetary risk of doing so, and even the potential cost of wasted time is rather small (a couple of minutes to listen to a song). On the contrary, movies in the pay per view model carry an actual monetary risk, as well as a larger cost in terms of wasted time (a couple of hours to watch an obliged movie now paid for). Furthermore, music is listened to again and again, whereas the majority of movies are only watched once. Such subtle differences can affect how eWOB plays out. For example, based on the above we should expect the volume of eWOB to be higher in a subscription-based music services such as Spotify compared to a pay per view movie service. However, since the risk involved in making a wrong choice is larger in movie streaming, we should expect users to do more extensive research and be quite dedicated to the movies they watch. As such, having
‘watched’ a particular movie might carry more weight when observed by other users than having ‘listened to’ a particular song. This reasoning illustrates the need for further research that compares the impact of eWOB across different product categories and revenue models, such as experience versus search goods, products with repetitive use versus single use, goods of mundane character versus self-expressive character, goods with sensitive or controversial content versus uncontroversial, and the like. Finally, also relevant in terms of generalization is the issue of cultural differences. Both studies were carried out in a Danish context – a country known for its high level of individualism 10. Consequently, it is reasonable to believe that eWOB can have different, and perhaps more significant, effects in more collectivistically orientated cultures where there is less willingness to stand out than in Denmark.
Moreover, it should be stressed that the conceptual framework for eWOB does not necessarily represent an exhaustive framework. Nor is it fully empirically validated. Rather, it seeks to provide scholars with a framework for understanding eWOB – its main components, processes, design dimensions, and how it conceptually differs from eWOM. It is the role of future research
10https://www.hofstede-insights.com/country-comparison/denmark/
86
to empirically assess the connections in this framework. This also ties to the importance of design in making eWOB come to life, and the testing of additional design dimensions. For example, what is the role of geographic immediacy – of being able to observe the behaviors of other users in geographical vicinity compared to long distance? Or what is the role of temporal immediacy – the disclosure of recent behaviors versus all behaviors? These are important topics for further scholarly exploration. Related to the conceptual framework, I identified the
‘dual sender’, the platform as well as user, as a key factor distinguishing eWOB from eWOM.
This deserves further research across product types and different ways of presenting eWOB.
Most likely, the aggregated type of eWOB would mean a stronger emphasis on the platform as sender, whereas the friend-specific type of eWOB would appear more to come directly from users. But such differences have yet to be tested empirically.
Finally, both of the empirical studies raised issues that warrant further investigation. Firstly, future research should explore the design space related to social information, including eWOB.
For example, investigating what the threshold is for the number of friends displayed to have an impact. Our study only investigated two fixed representations (two versus eight friends), but future research could benefit from viewing effects on a continuum. Further, we investigated the impact of influential friends determined by users’ own pre-selection. What might be the effect if the users were not involved in identifying the influentials? Such issues of design are
important to further explore.
Secondly, both the experimental study and the qualitative study point towards the user need be able to easily control eWOB. In the experiment, users expressed how they would like to have the option to control from whom they received information, and in turn who they could observe their behaviors. Similar needs were expressed in the interview with Spotify users. Here, the main concern for users was options to implement audience control. Such controls are easy to implement, however their use can potentially affect the trustworthiness of behavior-based information, as described earlier in the discussion. On the other hand, such controls might mean that users will only receive information that is relevant to them. Accordingly, future research should address these potential pitfalls and opportunities of user-control by assessing whether the impact of eWOB decreases or increases in user-controlled conditions. Moreover, the qualitative study suggests that the disclosure of one’s own behaviors to oneself can be an important factor in the satisfaction or thwarting of users’ need for competence. As described above, digitization of products, and hence the product-related behaviors, have profound
87
implications for the individual, and future research should explore to which degree digital behavior-based information can replace the once physical artefacts as a means for self-expression and identity formation and reinforcement. Finally, while the extant literature generally paints a picture of users, more or less blindly following others, the interviews with Spotify users also gave indications of how eWOB can also be used inversely. In effect,
providing a means for users to avoid certain, popular, music. Accordingly, the use of reference groups by niche segments should be further explored in future research.
6 Conclusion
This dissertation aims at theorizing an emerging empirical phenomenon and empirically investigating its potential as a communicative design element in two content-based digital services. I arrive at a conceptualization of the phenomenon of interest as ‘Electronic Word of Behavior’ (eWOB) defined as published accounts of behavior, based on the unobservable digital traces of consumers’ behaviors. I have related eWOB to the concepts of social interactions and eWOM and identified unique properties and three design dimensions of eWOB. eWOB distinguishes itself by a) being based on the unobservable digital traces of behavior, b) being plentiful and requiring little, if any, effort from users to generate c) being design-driven d) possessing an inherit duality in terms of who the communicator is e) being more neutral in nature compared to eWOM and finally f) being concerned with not only sales, but also on continued usage and engagement.
To assist the future design of eWOB in manners that leverage the strengths and minimizes the weaknesses, I moreover conducted two empirical studies. The first study, an online
experiment, departed from a business perspective and investigated the use of social
information in product design, consisting of opinions and behaviors of friends. The findings from this study demonstrated that online content-based services can benefit from incorporating social information into their interface in that it positively increases potential users’ attitude towards the service, albeit not intention to use the service. Moreover, the results suggest that behaviors can be designed to look like opinions and serve as endorsements. This means that eWOM, which requires user effort and suffers from an underreporting, can in some cases be substituted with eWOB, which requires little, if any, effort from the user and is plentiful. The experiment did not produce statistically significant results regarding the relative effectiveness of different design configurations and hence, further research in this direction is needed.
Finally, the second empirical study, an interview study, departed from a user-perspective and
88
demonstrated that consumers apply interpretations and ascribe use-opportunities to eWOB that go beyond the notion of ‘popularity information’. Specifically, eWOB can satisfy the basic psychological needs of relatedness and competence but can also potentially thwart autonomy and competence. I posit that this need satisfaction can be further increased (and need thwarting reduced) if product designers become aware of this potential, presumably resulting in an increase in users’ general well-being. To this effect, the PhD offers a set of implications for design.