Yeah, I've had this problem. It's a clash of models generally and I find that rather than (re)explaining it helps to kick up to a meta-level and talk about the general concepts for a bit to see where the model mismatch is. Often I find that the other person doesn't even agree on the general form of the problem to be solved, which is why they don't think my solution is relevant or don't understand my solution.