Best-Worst Case 2 is Jordan's twist on Paul Green traditional card-sort conjoint designs (where we show one concept at a time to respondents). But, with BW2, respondents don't give an overall evaluation of the suitability of that concept. Rather, we ask them which level within that concept makes them most want to purchase the product and which level within that concept makes them least want to purchase.
I can think of three good ways to generate traditional card-sort conjoint designs that would be suitable for this using Sawtooth Software:
1. Use our MaxDiff software, but specify the prohibitions between levels within the same attribute, so that levels within the same attribute don't show up within the same concept. This is the same as the tutorial study that ships with MaxDiff software (the Navy reenlistment study). We also describe this approach and test out the prohibitions within an appendix of the MaxDiff manual.
2. Use our CVA software, which has a very nice algorithm for searching for d-optimal designs.
3. Use our CBC software, where you specify just one product concept per task vs. the None alternative.
All three methods will produce high quality designs. CVA is limited to generating up to 10 different blocks (versions) at a time, but the d-efficiency will be slightly higher for its designs. MaxDiff and CBC will both allow you to generate hundreds of different blocks within the same design generation run.