Variety Seeking and OTT Subscription Behavior: An Analysis of Duration and Win-Back Using Transaction Data*

Purpose: As the over-the-top (OTT) market matures and content substitutability across platforms increases, understanding platform-level subscription dynamics—duration, churn, and win-back—has become increasingly important for platform sustainability. Prior OTT research has mainly focused on platform attributes such as content quality, system quality, satisfaction, and continuance intention, relying largely on survey-based data. Less is known about whether consumers’ intrinsic behavioral tendencies, observed in real-world spending patterns, help explain actual subscription adjustment behavior. This study addresses that gap by examining the role of variety seeking (VS)—a tendency to pursue novelty and avoid satiation—in shaping subscription duration and post-churn resubscription behavior in OTT services. Research design, data, and methodology: This study uses 36 months of actual credit and debit card transaction data (January 2022 to December 2024) from 47,050 users of Tenqube’s mobile household accounting application. Individual-level VS indices were constructed from observed brand diversity in franchise restaurant and café purchases, thereby operationalizing VS as a relatively stable behavioral tendency manifested in everyday consumption. These measures were then linked to subscription payment histories for five major OTT platforms in South Korea—Netflix, TVING, Coupang Play, Wavve, and Disney+. To capture platform-level subscription behavior, user-platform subscription sessions were constructed separately for each OTT service, including cases of concurrent subscriptions. A Cox proportional hazards model was employed to estimate the effect of VS on subscription duration, and a negative binomial regression model was used to examine the effect of VS on win-back frequency after churn. The models controlled for age, gender, platform-level MAU, baseline subscription price, and integrated service quality, while also incorporating time-varying covariates for concurrent subscriptions and actual payment amount. Results: VS was systematically associated with both shorter subscription duration and lower same-platform win-back frequency. In the Cox proportional hazards model, higher VS significantly increased churn risk, indicating shorter subscription duration on a given OTT platform (Hazard Ratio=1.51, p<.001). In the negative binomial model, VS was negatively associated with win-back frequency after cancellation (IRR= 0.606, p<.001), suggesting that a one-unit increase in VS corresponds to approximately a 40% reduction in the likelihood of returning to the same platform after cancellation. Additional robustness checks using a Weibull AFT model, a zero-inflated negative binomial model, and a Copula framework produced similar results. A price-segment moderation analysis further showed that the adverse effect of VS was amplified among premium-tier subscribers, with a greater churn hazard and a lower win-back rate. Conclusions: This study extends the concept of variety seeking to the digital subscription economy by showing that behaviorally measured VS predicts platform-level OTT subscription dynamics. By using transaction-based behavioral measures rather than self-reported survey responses, it provides a behaviorally grounded explanation of how consumer tendencies translate into observable subscription duration, churn, and win-back patterns. Managerially, the findings suggest that uniform customer relationship management (CRM) strategies may be insufficient in heterogeneous subscription markets. For lower-VS users, relationship-reinforcing tactics may help extend subscription duration, whereas for higher-VS users, flexible subscription plans, pause-subscription options, novelty-oriented content curation, and prediction-based win-back campaigns may be more effective. Overall, consumer heterogeneity in variety seeking should be considered in designing strategies to extend subscription duration and encourage reacquisition in subscription-based media markets.