Three essays in modeling customer retention
The objective of this research was to improve the prediction of customer retention and customer churn and to establish empirically the impact on these rates of different factors in the context of the customer-firm relationship. In this dissertation I focus on three research problems. In the first problem, I model the prediction of customer retention in the context of repeat visits to a website. I incorporate heterogeneity in customer retention over time by allowing customer retention probabilities to vary across first and subsequent return visits. I show the effects of customer fit (how well the product meets the customer's requirement), switching costs and customer interactions on customer retention at an Internet recommendation site. In the second problem, I model the prediction of customer churn in the context of a continuous subscription product i.e., direct-to-home satellite television. I highlight the importance of incorporating heterogeneity in the customer churn rates across customers. I empirically link customer churn to customer service experience, failure recovery (how well the firm performs in its recovery efforts after a service failure) and payment equity. In the third problem, I model the prediction of different types of customer churn for a continuous subscription product. I show the importance of accounting for all types of customer churn---voluntary as well as involuntary when estimating the churn probabilities. Ignoring the existence of involuntary churn significantly impacts the prediction and diagnosis of voluntary churn.
What people are saying - Write a review
We haven't found any reviews in the usual places.
Customer Fit and Customer Retention at an Internet
Hazards of Ignoring Involuntary Customer Churn
1 other sections not shown
algorithm ARREAR baseline hazard function Bayesian Information Criterion benchmark model bivariate model bivariate Weibull survival Bolton capture change in hazard churn probabilities churners competing event conditional probability Control Variables covariates covariates on voluntary credit card customer churn customer fit customer interactions customer lifetime value customer retention probabilities customer satisfaction customer service experience customer-firm relationship customer's customers in segment data sample DCASH defined dependent competing risk duration observations Exponential firm firm's hazard rate holdout sample homogeneous model incorporating heterogeneity increase indicator variable Internet recommendation Journal of Marketing likelihood function main model managers Marketing Science Model Selection number of customers number of ratings observed duration payment equity prediction of customer proportional hazard model recommendation engine response parameters retention rate revisit RMSE significant stratum subsequent return visits survival probability switching costs t-statistic value types of churn univariate Weibull model voluntary and involuntary voluntary churn rates Weibull distribution Weibull survival model