Consumer Engagement Pattern Analysis Leading to Improved Churn Analytics: An Approach for Telecom Industry

Abstract

Telecom industry is ever evolving not only in terms of technology and services but also with new players entering the market, and providing consumers with attractive options. The churn in telecom has always been a concern for service providers. Churn may relate to prepaid or postpaid services. With postpaid services, the customer may voluntarily churn by raising a cancelation request, hence enabling the service provider to take retention action. However, in case of prepaid services, the customer may stop using the services abruptly without prior intimation to the service provider. In such cases, there is no way to retain, as the customer has already churned. The only way to handle such problem is by predictive modeling where the target variable is defined as customers who have churned in past, to train the model. However, the problem with such model is that they use “decline in usage” as independent variables while modeling. These variables are generally a weak indicator of churn. The main goal of this research is to provide a new mechanism to develop churn model by taking into account the shifts in activity patterns of service usages. In particular, we measured the customer activity by measuring the average length of inactive days and frequency of inactivity. By introducing such measurements of inactivity in defining the churn variable, we could: (1) Quantify churn risk for customers. (2) Attempt to retain the customers while they are still active.