Estimation Of The Customer Mean Survival Time In Subscription-based Businesses
Free (open access)
Z. Mohammed, S. Maritz & D. Kotze
Two main components must be estimated in order to estimate customer lifetime value. The first component is customer survival time (time from subscription to cancellation of the service) and the second is the customer monthly margin. While the customer monthly margin could be estimated directly from the accounting model, the challenge will be the estimation of time to cancellation. In this study 30000 customers were selected from a well established subscription-based company. The customers were classified (segmented) according to their demographic and usage related characteristics. The stratified Cox model was used to identify the significant variables and to calculate hazard ratios. Both nonparametric and parametric survival analysis techniques were used to estimate the mean survival time. The results showed that gender, age and direct marketing city are significant in predicting the hazard of cancellation of the service. Young customers (age less than 26 years) have significantly shorter mean survival time than other age groups. A large difference between the restricted and unrestricted mean survival time was found; this may be due to the extreme right censoring of 85% that exists. Key words: subscription-based businesses, customer lifetime value, Kaplan-Meier product limit, Cox proportional hazard model, parametric survival regression. 1 Introduction Customer survival time is defined as time from subscription to cancellation of the service. The study of customer survival time has two important roles to play. The first in identifying important covariates that affect customer relationship
subscription-based businesses, customer lifetime value, Kaplan-Meier product limit, Cox proportional hazard model, parametric survival regression.