Derived vs. stated importance in customer satisfaction surveys

A common problem faced while analysing data from customer satisfaction surveys is the comparison of stated and derived importance for a set of satisfaction dimensions. Stated importance is defined as the straightforward customer preferences for the weight of a satisfaction criterion, while derived importance is estimated by a regression-type quantitative technique using customer judgments for the performance of this set of criteria. This paper presents a methodological framework for the comparative analysis of the derived and stated importance, given that significant differences are usually observed between them. The evaluation of derived importance is based on the MUSA method, while a sorting regression model for ordinal variables is used to estimate stated importance. This proposed model uses linear programming techniques and follows the principles of preference disaggregation approach. The main result of the analysis refers to the development of a dual importance diagram, which combines the derived and stated importance of the criteria. Also, this diagram may introduce the principles of Kano's approach for defining different quality levels and may give the ability to classify customer requirements. The implementation of the proposed approach is illustrated through several real-world applications and the presented results are focused on the development of improvement strategies. Future research is also discussed through this paper, including alternative formulations for modelling this particular problem.