Assessing User Retention of Updatable Home Appliances: Survival Analysis

Abstract

The advancement of smart home appliances has introduced updatable features, necessitating a reevaluation of traditional evaluation methods. While retention metrics have been widely used in marketing, they lack statistical validation and fail to capture personalized user interactions with these devices. This paper leverages data from downloadable and updatable content to develop a survival analysis-based engagement metric, addressing these limitations. This paper reframes engagement measurement as a time-to-event problem at the individual user level: for each user, we define the event as the onset of a substantively long period of inactivity and estimate the resulting survival function using standard nonparametric methods. We formally demonstrate that our survival framework extends conventional retention measures, with retention as a special case in the absence of censoring. Empirical validation on real-world data establishes a foundation for continuous engagement measurement, offering practical insights for improving user assessment and informing product development and marketing strategies.