USING EXTREMES TO DESIGN PRODUCTS AND SEGMENT MARKETS

Current marketing methodologies used to study consumers are inadequate for identifying and understanding respondents whose preferences for a product offering are most extreme. These ''extreme respondents'' have important implications for product design and market segmentation decisions. The authors develop a hierarchical Bayes random-effects model and apply it to a conjoint study of credit card attributes. Their proposed model facilitates an in-depth study of respondent heterogeneity, especially of extreme respondents. The authors demonstrate the importance of characterizing extremes in identifying product attributes and predicting the success of potential products.