Multistate Belief Probabilities-Based Prioritization Framework for Customer Satisfaction Attributes in Product Development

The traditional approach to prioritization of customer satisfaction (CS) attributes includes methods such as analytic hierarchical process (AHP) that do not consider the correlation among CS attributes. This paper presents an analytic network process based framework that allows decision makers to prioritize the CS attributes by considering not only the correlation among the attributes themselves, but also the cross correlation among attributes and factors. Furthermore, the proposed framework employs Bayesian Belief Network methodology to deal with uncertainty in prioritization process due to subjectivity of the information during the early stages of product development. The belief probabilities are expressed in terms of conditional probabilities that reflect the contribution of an attribute toward a given prioritization criterion. In addition, we propose a novel approach to estimate the belief probabilities by considering different states of an attribute such as strong, average, and null. This approach improves the precision of the belief probabilities that are usually estimated through expert interviews. The framework is illustrated with an automotive industry case study with results presented for two disparate vehicle types. The model results are also compared with those of a traditional AHP model to show the benefits of the proposed framework. Lastly, sensitivity analysis is performed to understand the impact of network structural uncertainty on the prioritization decisions thereby demonstrating the robustness of the framework.