Product design using consumer and designer preferences: The analysis of multisource conjoint data

Date of Completion

January 1997


Business Administration, Management|Operations Research




Product design decisions are of strategic importance to any firm. Conjoint analysis procedures are often used to collect data used in the product design process. But while many methods have been developed to take this data and produce optimal or near-optimal product designs, conjoint analysis designs a product based only on input from the consumer. Designer preferences can be as important as consumer preferences in the product design problem, and there is a need to integrate the two perspectives.^ Models for product design using preference data from both the consumer and the designer are developed. Since the way that a consumer looks at product attributes is usually quite different from the way a designer looks at product characteristics, two distinct sets of conjoint data are collected. These sets of data are related to each other through a "house of quality" type matrix. Because the product design problem is NP-hard, heuristics incorporating Genetic Algorithms and pruning techniques are developed to take this data and determine an optimal product design. Optimality is based either on maximizing the total market share, or on minimizing the Taguchi concept of a total loss to society, considering both consumer and designer requirements in each case.^ Computational results using test data are presented for the heuristics. Structural and sensitivity results are also presented for the models. The models and heuristics are then applied to the real-world problem of redesigning a course in business information systems. A second application looks at the problem of multiperiod sales promotion design. Results of these applications support the need, usefulness, and significance of integrating consumer and designer preferences. ^