In today’s mobile ecosystem, new apps emerge constantly while many quickly disappear, driving users to revise and reconfigure their app portfolios. Yet little research examines how individuals manage adoption and churn across the full lifecycle. Prior studies often treat apps in isolation, overlooking interdependencies, complementarities, and multi-homing. We address this gap by applying portfolio theory, which views users as managing “app portfolios” by balancing utility (returns) against satiation or fatigue (risks). Using Multiple Discrete-Continuous Extreme Value (MDCEV) models and machine learning, we forecast adoption and discontinuation. Empirical analyses show that portfolio-based models substantially outperform baseline models, underscoring the importance of portfolio-level dynamics. Notably, performance gains are strongest among high-income and highly educated users, suggesting that portfolio-theoretic representations align with more rational decision-making tendencies.
