Seth Godin: Creating a useful spec

1. What is the problem to be solved?
2. How are we going to solve it?
3. How can we test that the thing we built matches what we set out to build?
4. How will we know if it’s working?

Source: Creating a useful spec | Seth’s Blog