Analysis of the value added from improving contact-center response time in a customer service application

Abstract

Targeted maximum likelihood estimation (TMLE) procedure is proposed to evaluate the causal impact of the waiting time and abandonment of a phone call in customer profits. Based on different spending behaviors and Customer Service (CS) information, this work finds the causal effect for the Average Treatment Effect (ATE) of the waiting time. This is done for customers with a single contact, two contacts and single contact with abandonment. The results are supported by showing the corresponding 95% confidence intervals as well as clearly stating all the assumptions involved and the limitations of the TMLE methodology. Additionally, the document illustrates the methodology for the identification of the effect by providing a comprehensive expla-nation of the use of Directed acyclic causal graphs.