[AnthonyMouse]

>More often than not this means thinking about it a couple of hours more, and results in much faster development the next week, so for me the correct answer, even for start-ups, is " we can't afford doing it wrong"

Sometimes, sure. Doing it right is always worth it if you have the time -- and if all it takes is a couple of hours then you almost always have the time. But sometimes doing it right means doing it over, because circumstances have chanced since you started, or there is a third party framework you can use which is clearly a bad fit for your application but the only alternative is to write your own from scratch. Then you're not talking about a few hours anymore, you're talking about a few months. Which you may or may not have.