Modelling customer satisfaction and loyalty on aggregate levels:: Experience from the ECSI pilot study

Aggregate Customer Satisfaction Index (CSI) measures are becoming more and more common as indicators of business performance in different industries, and are used for a multitude of purposes. However, any comparison of such figures from one domain to another calls for tightly harmonized and co-ordinated methodological frameworks. In this paper we study the prerequisites for developing a common model structure useful for devising aggregate CSI results throughout Europe, and comparing also with similar efforts in other parts of the world. Our study is based on an evaluation of the stability and robustness of empirical results from the European Customer Satisfaction Index (ECSI) pilot survey round. In spite of the fact that European customers, product supplies and cultural environments are different, it is found possible to specify a common structural model to be used in such diverse industries as telecoms, banking and supermarkets, throughout 11 European countries. This model is not optimal in the majority of cases (in terms of explanatory power), but neither in any case is it very far away from fulfilling high quality criteria. Thus, it fulfils the prespecified requirements in terms of robustness, and is proposed to be used as the standard in future Pan-European activities of this character. In the second part of the paper the prerequisites for finding adequate manifest (measurable) variables for the structural latent variable model are very briefly considered. A comprehensive factor analysis study of all ECSI pilot corporate models (about 150) is scrutinized. The results of this study are rather comforting in terms of generating adequate factors (similar to the ones predefined as latent variables) and using a common (master) measurement instrument (questionnaire). However, for a few latents (especially perceived value) certain problems of measurement are identified.