Customer defections analysis: an examination of online bookstores

Abstract

Purpose

The cost of retaining a customer is lower than that of obtaining a new one, so potential customer defection is an important issue in the fiercely competitive environment of electronic commerce. Accordingly, this paper aims to present a new way for gauging customer loyalty and predicting their possibility of defection reference to a set of quality attributes satisfaction and three types of belief in the theory of planned behavior (TPB).

Design/methodology/approach

The performance of the classification utilization artificial neural networks (ANNs) was compared to that of traditional analytic tools, such as multiple discriminant analysis (MDA) and classificatory data mining technique – decision tree.

Findings

The analytical result represented that the predicted accuracy of ANNs is better then MDA and decision tree in both training and testing phases. Degree of repurchase intention has been classified correctly with a success rate of 83 percent using neural networks.

Research limitations/implications

Like all research, this study has its limitations. One such limitation is that the predictive model was designed for application to online bookstores. A further limitation of this survey is that it reflects intentions instead of actual behavior. Finally, despite of ANNs has been applied to numerous areas and have demonstrated a degree of classification success, it is difficult to extract rules for explanation. Therefore, enhancing ability of model explanation would be a valuable work in the future.

Originality/value

The contribution of this paper is to predict how marketing practitioners can tactically market to customers with weak repurchase intentions to prevent defections.

Citation

Shih, Y. and Fang, K. (2005), "Customer defections analysis: an examination of online bookstores", The TQM Magazine, Vol. 17 No. 5, pp. 425-439. https://doi.org/10.1108/09544780510615933