Conjoint analysis with textual external information

In market researche we often need statistical methods able to jointly deal with numerical and textual variables. Particularly in market segmentation, several tools have been proposed for profiling customers, by surveying a sample of potential buyers. One of the most common statistical analysis aiming at identifying the “ideal” product for a target market is Conjoint Analysis (Green et Srinivasan, 1990). This technique considers a good/service as a proper combination of jointly evaluated characteristics. Each customer has to compare those combinations, i.e. the different potential (or actual) goods. The so-called full profile method often used in Conjoint Analysis sometimes seems too rigid in its experimental structure. Moreover, we are not always sure about the descriptions of the different goods, as we can consider only a small number of variables. We propose a procedure aimed at integrating Conjoint Analysis with textual information achieved by answers to open-ended questions. From a methodological viewpoint, this means introducing external information in the Takane and Shibayama (1991) sense. The proposed procedure improves the understanding of the preference structure and the ideal product definition. An empirical analysis on the watch market shows the effectiveness of this proposal.