Service Quality Variability and Termination Behavior

Abstract

We investigate the roles of the level and variability in quality in driving customer retention for a new service. We present model-free evidence that while high average quality helps in retaining customers, high variability leads to higher termination rates. Apart from these main effects, we use model-free evidence to document the presence of (a) an interaction effect between average service quality and its variability on termination rates, (b) customer learning about service quality over time, and (c) slower rate of learning among households that experience high variability. We postulate a mechanism involving risk aversion and learning, which can induce this interaction effect and test this against several alternative explanations. We show that it is important to consider variability in quality while inferring the impact of improvements to average quality - ignoring the interaction effect between average quality and variability leads 18% to 64% (5% to 31%) overestimation (underestimation) of quality improvement elasticities among high-variability (low-variability) households. Given that responsiveness to quality decreases with variability, it is better for the firm to focus quality improvement efforts on customers experiencing low variability; increasing average quality by 1% lowers termination by 1.1% for low-variability households, but only by 0.41% for high-variability households.

Keywords: Service quality, customer retention, Bayesian learning, dynamic models

JEL Classification: M31, C10, C50, D81, D83, D90