[gvinter]

Start with a really good spec. I started ProjectPulse.io by clearly defining the problem, and what the solution would look like. Then I dove into defining the needs of all the MVP features. From there, it wasn't hard for me to make wireframes and get awesome designs from a designer. Then, I also made a list of features I'm going to want to add, but not for the MVP, so they felt like they were on paper, but not the same as MVP features.

Once you have a list down of things that need to be built because you know the sum of those things is an MVP, put your head down and work until it's done. Don't question anything until you have an MVP. Stop second guessing the product or the market and just build it.

I'm leaving out the whole "research and learn from potential users" - that comes first, but that's not what we're talking about here.

TL;DR: Write a detailed spec for an MVP, then put your head down in code until you have an MVP.