That's what I asked previously here:
http://news.ycombinator.com/item?id=1514218Anyway, my input is, based on what you know best and your experiences, come up with something that makes sense. Keep developing the hypotheses until you cannot poke holes in them or have honest doubts about them, then go test your hypotheses against anyone who knows what they're talking about other than yourself. If your hypotheses are still intact, then expend the cost to reach out to a potential customer. Saves you a lot of hard work.