Applying Survival Analysis for Customer Retention: A U.S. Regional Mobile Service Operator

Customer retention is of importance, as the mobile industry experiences an average of 30-35 percent annual churn rate and it costs 5-10 times more to recruit a new customer than to retain an existing one. In this study, we investigate customer churn in a medium size U.S. mobile service operator (telco) with approximately 1 million subscribers. The operator is experiencing about 5% drop off rate every month, which implied that more than half of the company's customers at the beginning of the year, are gone by the end of the year. The company aims to identify potential churned customers ahead of time and offer them appropriate incentives to entice them to stay. In order to study this, survival analysis was conducted. In this study, nonparametric Kaplan Meier estimator was used to estimated survival probabilities. Key determinants for further investigation were identified using the Walds Test and LogWorth size estimators. Further tests to identify the risk of churn were conducted using the semi-parametric Cox Proportional Hazard model to simultaneously evaluate the effect of covariates on survival and churn risk between the sub-covariates. The study identified the targeting of the top 62% of the customers for optimal benefit, and recommended appropriate actionable insights to implement a loyalty program. Through the understanding of the determinants of customer churning behavior, the prediction of which customers are most likely to leave, and the expected duration of customers who have yet to churn, the company can conduct pertinent marketing and promotional efforts to encourage customers to remain with the company for enhanced profitability. The findings of this study can be applied in supplier convergent triple-or-quad-play services, and form a useful guide to the industry implementation of data analytics survival analysis back-end platform, as part of a larger customer acquisition, development and retention analytics solution in mobile service operators.