A little data is better than no data since you would then be left with only your business knowledge to lean on.   It is not uncommon to use surrogate target variables that might be related to the new target variable.   You might also profile your new population (if possible) and study their behavior compares to your previous population.  You might be able to weight your sampling based to make the distribution of the observations similar in the sample on the previously known population and the new population.   You are really left with a judgement call to try and identify a meaningful surrogate target and to choose a comparable sample.   Hope this helps! Doug ... View more