Abstract:Low engagement rates and high attrition rates have been formidable challenges to mobile apps and their long-term success, especially for those whose revenues derive mainly from in-app purchases. To date, little is known about how companies can scientifically detect user engagement stages and optimize corresponding personalized-targeting promotion strategies so as to improve business revenues. This paper proposes a new structural forward-looking Hidden Markov Model (FHMM) as combined with a randomized field experiment on app notification promotions. Our model can recover consumer latent engagement stages by accounting for both the time-varying nature of users' engagement and their forward-looking consumption behavior. Although app users in most of the engagement stages are likely to become less dynamically engaged, this slippery slope of user engagement can be alleviated by randomized treatments of app promotions. The structural estimates from the FHMM with the field-experimental data also enable us to identify heterogeneity in the treatment effects, specifically in terms of the causal impact of app promotions on continuous app consumption behavior across different hidden engagement stages. Additionally, we simulate and optimize the revenues of different personalized-targeting promotion strategies with the structural estimates. Personalized dynamic engagement-based targeting based on the FHMM can, compared with non-personalized mass promotion, generate 101.84% more revenue for the price promotion and 72.46% more revenue for the free content promotion. It also can generate substantially higher revenues than the experience-based targeting strategy applied by current industry practices and targeting strategies based on alternative customer segmentation models such as k-means or myopic HMM. Overall, the novel feature of our paper is its proposal of a new personalized-targeting approach combining the FHMM with a field experiment to tackle the challenge of low engagement with mobile apps.

Bayesian Non-parametric model to Target Gamification Notifications Using Big Data

A Nonparametric Bayes Approach to Online Activity Prediction

Modeling User Engagement in Mobile Content Consumption with Tapstream Data.

Bayesian Nonparametrics: An Alternative to Deep Learning

Personalized Mobile Targeting with User Engagement Stages: Combining a Structural Hidden Markov Model and Field Experiment

Deep Bayesian Multi-Target Learning for Recommender Systems

Big Learning with Bayesian Methods

A Bayesian Approach to Online Learning for Contextual Restless Bandits with Applications to Public Health

A Bayesian Approach to Data Point Selection

A new online learning algorithm for streaming data and decision support with a Bayesian approach

Games and Big Data: A Scalable Multi-Dimensional Churn Prediction Model

Should I send this notification? Optimizing push notifications decision making by modeling the future

Bayesian Exploration with Heterogeneous Agents

Bayesian nonparametric modeling of user activities

Bayesian Non-stationary Linear Bandits for Large-Scale Recommender Systems

Exploring Bayesian Deep Learning for Urgent Instructor Intervention Need in MOOC Forums

Model-Free Approximate Bayesian Learning for Large-Scale Conversion Funnel Optimization

BayesCNS: A Unified Bayesian Approach to Address Cold Start and Non-Stationarity in Search Systems at Scale

BayesGaze: A Bayesian Approach to Eye-Gaze Based Target Selection.

Bayesian modeling of networks in complex business intelligence problems

Bayesian Learning of Play Styles in Multiplayer Video Games