ANALYTICS
An example of a project which requires conjoint could be one in which we are trying to determine what set of product features is most advantageous. For example you may be looking to launch a new mobile tariff and want to ascertain what the optimum feature set would be given certain constraints i.e. the best handset is unlikely to offered at the lowest price.
An example of the feature set you might want to test for a new contract is given below.
- Handset (5 different models – low spec through to high spec)
- Battery life – talk time ( 3.5 hours, 4 hours, 6.5 hours)
- Handset price (4 levels)
- Text bundle/month (250, 500, 1000)
- Talk time (100, 200, 300,400)
- Contract Price (5 levels)
- Network(Brand A, Brand B, Brand C, Brand D)
From this list there are in total 14,400 product permutations but only a small subset of all these possible combinations needs to be presented to be able to make accurate predictions. In fact on average any one respondent would only complete 16 exercises. The gaps are filled via a conjoint model, which infers the missing values from the choices respondents make in the exercises they do complete.
