Estimating Recommendation and Repurchase Thresholds with a Joint Heterogeneity Response Model

Recommendation and repurchase intentions are the two most important dimensions of customer loyalty. Latent satisfaction thresholds at the individual level, if obtained, can function as an additional valid and effective criterion to satisfaction ratings in determining customer behavior. Hence, customer segmentation based on these thresholds can help firms optimize their resources to improve customer retention. This paper proposes a joint heterogeneity response model to simultaneously calibrate the individual-level recommendation and repurchase thresholds through a joint specification over the heterogeneity. We conduct a simulation study to examine the parameter recovery and model comparison between our proposed model and two competing models. We apply our model to the satisfaction survey data of automotive customers. The results outperform the other three alternative models and show more insight into uplifting customer loyalty using a segmentation scheme based on estimated recommendation and repurchase thresholds. In addition to the heterogeneous distribution of thresholds, our results uncover the effects of customer characteristics on the thresholds.