I recently sat in on a Talk by the author of this article. It's a really interesting approach to making sure the problem is well defined before leaping into the work of a new feature, idea, or product. In a nutshell the questions are:
Question 2 (the people problem) is probably the most important: how would people on the street describe the problem this is addressing and how does this solve it?
Do you have frameworks like this for your work? What do you do differently?