Did you ask respondents to evaluate the product concepts on a purchase likelihood scale? For example, with traditional ratings-based conjoint (in our software product line, it is called CVA), you can ask respondents to rate the purchase likelihood for each concept. Or, with our ACA system, you could have included Calibration Concepts, wherein respondents rate a handful of product concepts on a 100-point likelihood of purchase scale.
For CBC, it is possible to ask binary buy/no buy questions for product concepts. With that information, you can simulate likelihood of choice (purchase) vs. the no buy option.
All three of these cases lead to an opportunity to simulate (model) purchase likelihood for individual products. Of course, we recognize that in many/most cases, respondents tend to overstate their purchase intent...so we interpret the results cautiously.