Abstract

This research sets out to investigate consumer choices in a world-renowned music festival. We conceptualize consumers as actors whose action is to adopt a festival mobile companion application—an event-based information technology (IT). Their mass adoption creates artifacts in the form of large-scale anonymized digital trace data, which shed light on activities, chiefly, event attendance. Our main objective is market segmentation by which consumers with similar tastes are grouped together. To this end, we concretely operationalize mainstream, novel, and diverse tastes in content consumption. These tastes widely apply to consumers and can be generalized to various social commerce settings. Integrating taste-aware recommendations into event-based IT services would support consumer decision-making against the backdrop of the attention economy characterized by information overload (e.g., overabundant leisure activities) and opportunity costs (e.g., schedule conflicts). Our clustering-based market segmentation methodology comprises motif recognition and novel visualizations. The first step harnesses recent advances in natural language processing to build contextual representations of participating content creators by generating text embeddings (i.e., vectors of numerical values) from their content. Drawing on the recommender systems and innovation literature, the second step ranks each creator’s mainstreamness, novelty, and genre diversity according to text embeddings and/or numerical Spotify audio features (e.g., danceability, instrumentality). Our mainstream ranking reflects the closeness of a creator to the “average” and is more context-specific than popularity. Our comparative novelty ranking gauges the differentiation of creators, whereas the combinatorial novelty ranking measures the degree to which creators combine keywords with disparate meanings. Our genre diversity ranking quantifies the divergence between creators’ affiliated genres (akin to product tags). Leveraging event attendance, the third step aggregates these rankings per consumer to obtain consumer tastes. The final step clusters consumers along their mainstream, novel, and diverse tastes, followed by robustness checks. Our cluster analysis finds basic consumer archetypes, including mainstreamers who join mainstream events and novelty-seekers who participate in niche events. We uncover an “anti-mainstream” archetype regardless of novelty-seeking, and a “ambidextrous” archetype consuming mainstream yet niche events. Contrary to the conventional expectation, this suggests mainstream and novel tastes are not mutually exclusive and complement each other. Furthermore, we identify a sizable minority of consumers who explore several homogeneous events. By contrast, only a small consumer segment picks highly heterogeneous events, which may be a proxy for boundary-spanning. This indicates diversity and exploration capture different aspects of content consumption–the former is the inverse of content similarity while the latter is primarily based on frequency or quantity. Lastly, we discuss how event management could instantiate tailored responses to different consumer tastes instead of naïvely suggesting the most popular or similar events. One practical implication is to afford consumers serendipity by recommending niche or diverse events. We position this research at the intersection of information systems and computational social sciences. One major contribution is the development of theory-driven, taste-grounded explanations for event attendance and potential consumer base overlap or lack thereof, constituting a new account of exploration vis-à-vis exploitation.

Comments

tpp1426

Download

Share

COinS