Modeling Customer Satisfaction’s Impact on Loyalty: Insights for Customer-Centric Resource Allocation

This study explores the nonlinear modeling of customer satisfaction (CS) and loyalty for resource allocation strategies that promote efficient loyalty programs. We first introduce global models to reflect several important nonlinear characteristics, particularly (a) a saturation-attainable limit of effectiveness, (b) nonconstant marginal returns, and (c) asymmetric responses between satisfied and dissatisfied customers. In the proposed models, we put forth the joint use of two types of mixture models to deal with different levels of heterogeneity. Based on the model best supported among alternatives in the literature, we targeted customers using expected incremental loyalty, which is derived from the joint use of an estimated response curve of loyalty to satisfaction and an empirical distribution of CS scores. Then, we evaluate the efficiency of loyalty programs under the assumption of full and limited access to customers and subsequently derive managerial implications. For instance, through a counterfactual simulation of the models for the three industries, we find that improving the perceived quality in the mobile communication industry may have a greater effect on loyalty than in the hotel and convenience store industries because of its higher switching cost. Funding: N. Terui acknowledges funding by JSPS KAKENHI [Grants (A)17H01001 and (A)22H00056]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/serv.2022.0313 .