Validating your idea. That is one of the best but most overlooked advise. Validate, validate again.
Polished product > Happy path demo > Mockups (mechanical, digital, etc) > User stories > Deck
At the very least iterate on things like the user stories to make sure your customers can see your product fitting in to their day to day life. Without any form of iteration, you're leaving the most important thing -- product market fit -- somewhat to chance.
In our case, we can iterate easily based upon real market information post-launch. It's literally illegal to do market research otherwise, so we can't really get a foot in the door hyper-analyzing potential segments ahead of schedule.