Accounting for dynamics in attribute-importance and for competitor performance to enhance reliability of BPNN-based importance–performance analysis

Importance–performance analysis (IPA) is a decision-support tool used in prioritizing quality improvements of products/services. Recently, back-propagation neural network (BPNN)-based approaches have been proposed to deal with the problem of asymmetric effects in customer satisfaction formation. Though reliability of IPA is increased by the integration of BPNN, shortcomings of the analytical framework remain that (a) it does not provide insight into forms and degrees of these asymmetric effects, (b) it does not account for differences between the relevance and determinance of quality attributes, and (c) it neglects the competitor dimension in attribute-prioritization. Since all these issues have important managerial implications, the authors of this study propose an extended BPNN-based IPA that uses a multidimensional operationalization of attribute-importance, and that considers competitive performance levels. Using data from an airline satisfaction survey, an empirical test reveals that the proposed approach significantly outperforms conventional BPNN-based IPA. In particular, conventional BPNN-IPA would mislead managerial action with regard to 3 out of 8 quality components (37.5%).

► This study proposes an extended BPNN-based importance–performance analysis (IPA).
► The framework accounts for dynamics in attribute-importance and for competitor performance.
► A case study reveals that the extended framework significantly outperforms conventional BPNN-IPA.
► A detailed basis for prioritizing quality improvements in competitive environments is provided.