Predicting customer churn from valuable B2B customers in the logistics industry: a case study

Abstract

This study uncovers the effect of the length, recency, frequency, monetary, and profit (LRFMP) customer value model in a logistics company to predict customer churn. This unique context has useful business implications compared to the main stream customer churn studies where individual customers (rather than business customers) are the main focus. Our results show the five LRFMP variables had a varying effect on customer churn. Specifically length, recency and monetary variables had a significant effect on churn, while the frequency variable only became a top predictor when the variability of the first three variables was limited. The profit variable had never become a significant predictor. Certain other behavioral variables (such as time between transactions) also had an effect on churn. The resulting set of predictors of churn expands the original LRFMP and RFM models with additional insights. Managerial implications were provided.