Background: Customers make multiple frequent purchases that prior research indicates fall into three broad themes (delivery speed, price and reliability) which differ for each purchase. The categories are the main attributes (there is also brand) for the first stage of the decision. There are only 120 possible option combinations (ignoring brand).
I plan to approach this using a volumetric CBC so that respondents can indicate what percentage/volume to allocate to each of three options (with 'None' being positioned as 'these three not suitable'). Each option can be of course randomly generated, or with a bit of manual tweaking themed by the broad response types (still allowing for limited overlap). I'm thinking that theming the left/middle/right options would:
Be easier on respondents - they only need to make decisions at the margin and nudge prior responses.
Lose some discrimination - they would have fewer opportunities to choose between three roughly equal options (although they would still see two equal speed or price or reliability levels, next to each other for easy comparison
Theming may, however, lose a little information as true randomization doesn't occur, however, this can be tested for to ensure nothing bad is likely to happen.
The other possible advantage of theming is that I have a number of additional, optional, services. It may be easier to analyse these using MBC where the conditional choice is made as a result of choosing a particular theme.