Customer Lifetime Value in Streaming : a Markov Chain Approach

Abstract

Customer lifetime value is an essential metric for driving marketing decisions. The present work focuses on its estimation for active users of MTN TV application in the streaming industry following an approach based on Markov chains. The models implemented vary according to the churn scenario (lost forever or semi-lost), the segmentation used as a basis for the states, and the combination of the previous models. A total of seven different models are implemented and evaluated on data for the year 2022, according to the root mean square error, mean absolute error and relative error metrics. The best-performing model results from an equally-weighted combination of chains built on a segmentation of the recency and the monetary components of the RFM (Recency, Frequency, Monetary) framework. This customer lifetime value model is strategically important, as it takes into account the dynamics between different user segments in the decision-making process.