The three studies in this dissertation offer several contributions, both to marketing theory and to marketing practice.
3.1. Theoretical contributions
From a theoretical perspective, we have studied the relatively under-researched area of social media marketing, and contributed to several aspects of this domain in the different chapters. In chapter 2, we introduced the notions of leading and lagging information for sentiment prediction models, which are promising paths to optimize sentiment prediction, next to research focusing on text elements. Relating to these variables, we laid out the fundamentals for more in-depth research into the formation of online sentiment, its antecedents and consequences. Although we do not formally test the proposed model, and especially the proposed middle-layer of unobserved concepts, we show the value of the observable characteristics in providing more accurate predictions of user sentiment. One option for future research would be to disentangle the effects and relationships of the unobserved concepts.
Chapter 3 makes significant contributions to the marketing literature, in several ways. First, we argue that online created content (UGC and MGC) can be linked to identifiable, actual customer experience encounters, instead of aggregating these measures over a particular period of time. This has important implications for our understanding of customer sentiment. We can link objective performance characteristics of the identified customer experience encounters to customer sentiment, and we can investigate the moderating role of MGC on the link between the experience encounter and customer sentiment. Second, we further link customer sentiment to direct engagement (CLV), thereby establishing the link between the experience encounters, MGC, customer sentiment and direct engagement in one model. We thus contribute to the literature on customer engagement (Pansari and Kumar, 2017) by demonstrating potential firm influences beyond more traditional marketing activities aimed at creating awareness. By doing so, we might link the theories of customer engagement (Pansari and Kumar, 2017) and customer engagement marketing (as conceptualized by Harmeling et al. (2017)). Whereas these latter authors focus on the direct influence of firm communications, our results support its moderating impact based on actual brand experiences. Third, we argue to include different measures of UGC and MGC in one comprehensive model, with control variables, in order to understand the influence of SM content on direct engagement, while previous literature has focused on individual measures. This allows researchers to better identify the real value of these social
media measures in relation to direct engagement. Finally, to the best of our knowledge we were among the first to introduce social media network metrics into direct engagement models, in addition to the other relevant social media variables. While previous research has focused mainly on networks via e-mailing or calling behavior (e.g., Nitzan and Libai, 2011), or has used social networks to set up viral marketing campaigns (Kumar et al., 2013), we show that social network information obtained via (online) social media also offer additional insights for modeling direct engagement with the firm. Thus, in spite of evidence stating that social media networks cannot readily be compared with offline networks because of the potentially large number of unrelated ‘friends’ (Dunbar, 2016), our research shows that the social media network is useful for modeling direct customer engagement.
Chapter 4 addresses the call for more (social media) marketing analytics research in B2B (Lilien, 2016). We are the first to quantitatively analyze the use of social media in the B2B acquisition process, instead of taking a qualitative approach. Moreover, from a modeling perspective, we have demonstrated that the acquisition model development is iterative in nature, and that it can benefit from including updated information into the model. With this research, we hope to spur academic interest in B2B applications in social media, since this is still an major untapped research topic.
3.2. Managerial contributions
From a managerial perspective, we have demonstrated in chapter 2 the ability to better predict customer valence related to Facebook posts. Since valence has been shown to be related to sales, it is important to correctly measure valence. Specifically in marketing, customer sentiment or satisfaction about a brand can be deduced from social media (e.g., Go et al., 2009; Schweidel and Moe, 2014; Tirunillai and Tellis, 2014). Making customer sentiment predictions more accurate also increases the applicability of these methods in comparison to previous methods (e.g., satisfaction surveys).
Chapter 3 offers insights for social media managers by investigating the role of MGC on social media. Our results imply that MGC can be effective to change customer sentiment, and ultimately customer engagement, but that its effectiveness is limited and dependent on the objective performance related to actual customer experience encounters. Positive customer experience encounters do not benefit as much from changes in MGC behavior as do more negative and neutral encounters. This is not surprising, since, within a service context, these latter encounters can be seen as service failures, and previous literature has already identified
151 that company-initiated recovery actions, such as MGC, can help to obtain service recovery (Smith et al., 1999). Moreover, the interactive nature of social media may further help to lead to positive service recoveries (Dong et al., 2008). Finally, we have shown that marketers’ interest should go beyond merely measuring and influencing ‘likes’ on social media, to include (at least) customer sentiment.
Chapter 4 offers direction to B2B marketing managers in how social media can be used in a quantitative way. While we acknowledge that these models may be adapted to specific environments, we delineate a standard procedure to perform acquisition modeling, we show that social media is a valuable source of information in the context of prospect to customer conversion, and that this approach can be highly profitable.