Predictors of customer satisfaction: testing the classical paradigms

Abstract

PurposeThe purpose of this study is to ascertain the best predictors of customer satisfaction by analysing and comparing the variables of two classical paradigms: the confirmation/disconfirmation model; and Locke's model of general satisfaction.Design/methodology/approachFollowing a literature review and the development of a conceptual framework in which various problems in the two models are identified, the paper presents a field study that investigates the extent to which the satisfaction of customers of a bank can be explained by these models and by two modified models (regression design).FindingsThe study finds: that the inclusion of “importance” in Locke's model does not provide a better prediction of satisfaction than the variables of “performance” and “expectation” in the confirmation/disconfirmation model; that the absolute level of the expectation‐performance difference is a better predictor of customer satisfaction than the simple relative difference; and that “performance” is a much more reliable predictor of satisfaction than “expectation” and/or “importance”.Research limitations/implicationsThe study is limited to the specific setting of bank customers; it is uncertain whether the findings can be generalised to other fields.Originality/valueThe study provides an original critique and comparison of the classical models and identifies their limitations. The study also demonstrates that the absolute level of the expectation‐performance difference is a better prediction of customer satisfaction than the simple relative difference. The study shows that “performance” is the most powerful predictor of satisfaction and that it is therefore not necessary, in practice, to conduct a differentiated survey of other predictors.