My understanding is that an MVP is designed to validate a hypothesis, which is why it should be released quickly, and iterated on quickly. I would also go further and contend that an MVP *should* suck. If it doesn’t then too much time has been invested in it, which is counterproductive, and you would have released an OK-VP. Here is the plan I usually follow for an MVP:
- Create a product that solves one thing and one thing only. No matter if it sucks, as long as it solves that one problem efficiently.
- Release to an early set of adopters who know this is an MVP, designed to solve that one problem.
- Is your hypothesis proven? Was the problem solved efficiently? If yes, you can start refining. If no, back to the drawing board.
Once all the details have been ironed out, then, by all means, make your product delightful, release to your whole audience and their first impression will be a great one with a product that you know works.
My 2 cents of course :)