Modeling User Engagement in Mobile Content Consumption with Tapstream Data and Field Experiment

Abstract

Low engagement rates and high attrition rates have been formidable challenges for mobile apps and their long-term success, especially for those whose revenues come mainly from in-app purchases. To date, still little is known about how companies can comprehensively identify user engagement stages so as to improve business revenues. This paper proposes a structural econometric framework for modeling of consumer latent engagement stages that accounts for both the time-varying nature of engagement and consumer forward-looking consumption behavior. The present study analyzed a fine-grained mobile tapstream dataset on mobile users' continuous content consumption behavior in a popular mobile reading app. Our policy simulation enabled us to tailor, based on the model-detected engagement stages, an optimal pricing strategy to each consumer. Interestingly, we found that such an engagement-specific pricing strategy leads, simultaneously, to lower average prices for consumers and higher overall business revenues for the app. To further evaluate the effectiveness of our method, we conducted a randomized field experiment on a mobile app platform. Our experimental results provide more causal evidence that a personalized promotion strategy targeting user engagement stages can both decrease costs to app users and enhance overall business performance. Our structural-model- and field-experimentation-based findings are nontrivial and suggest, with respect to the crucial role of modeling user engagement, potential overall welfare improvements in the mobile app market.