A combined QFD and integer programming framework to determine attribute levels for conjoint study

Abstract

Conjoint Analysis (CA) and Quality Function Deployment (QFD) are two popular tools for new product design; marketers frequently use the former and engineers the latter. Typically, in a conjoint study, the attributes and their levels are determined through focus group discussions or market surveys. Sometimes, the market researchers exclude some critical features or include unrealistic attribute levels resulting in infeasible product profiles. Inappropriate selection of attribute levels may render the conjoint study less useful. In QFD, the New Product Development team attempts to identify the technical characteristics (TCs) to be improved (included) to meet the customer requirements (CRs) through a subjective relationship matrix between CRs and TCs. At present there is no methodology that uses the output of QFD to generate feasible product profiles to be used in CA and therefore improve its usefulness. In this paper, QFD is used along with an integer programming (IP) model to determine the appropriate TCs and consequently the right attribute levels. These attribute levels are then used in a conjoint study. It is also proposed to measure the elements of the so-called relationship matrix in QFD in a way so that the right levels of the attributes can be generated from the IP solution. The proposed method is illustrated through a commercial vehicle design problem with hypothetical data.