The exercises in this phase force us to stop generating new solutions, converge on the best, and write the test for the prototype.
First, we identify assumptions we're making in the best ideas. We list all assumptions about users' motivations, the business model, our ability to acquire users, and our ability to implement the solution within budget. This helps eliminate some options.
We then look for conflicts in the remaining sticky notes, user flows, and user interfaces: ideas that aim to solve the same problem in different ways. We eliminate solutions that can't be pursued currently.
We then decide whether we're creating one prototype ("best shot") or multiple ("battle royale"). Multiple prototypes are more initial work but may reveal more dead ends and help us dodge more bullets without running follow-on sprints.
We then storyboard each prototype. This is a comic book-style story of our customer moving through the critical path.