Conjoint and Cluster Analysis


Conjoint analysis is a technique for analysing preferences in a way that represents as realistically as possible the buying decision. The buyer no doubt will have a "wish list" eg low price, high quality, reliable delivery etc. However, in the real world he/she has to choose between products/suppliers, none of whom may conform to his ideal. In effect he has to trade off price against quality, against delivery etc. Conjoint analysis presents choices to the respondent in exactly this way. Respondents are asked to choose between different combinations of features - the full concept method. The results are analysed through a sophisticated mathematical programme to establish the relative ranking of features that lay behind the choices. The respondents would probably find it very difficult to articulate their thought process. It is then possible to simulate how a new combination of features - on a new product or service - will be ranked by customers.


The analysis package will look for clusters of similar answers and flag these for further scrutiny. On a practical basis any of the sets of answers which shows a wide standard deviation, or standard error invites a further look. This usually shows that the average answer is covering up a wide range of replies. Bunches of customers, when discussed with the client, are found to have common characteristics which can be addressed in market or sales activity.