I know a few people who have a startup that's based on the biggest mess in the world, that took off like a rocket when it was launched, and pulled in a big pile

I know a few people who have a startup that's based on the biggest mess in the world, that took off like a rocket when it was launched, and pulled in a big pile of funding. Problem is, now they need to change direction slightly, and they can't, because their codebase and underlaying data model were built without anybody asking the question, 'So, can we fix this later?' I'm all for throwing a rough prototype together that's buggy as all hell and lacks features, but don't build bits of it that you can't easily change later. This isn't really as hard as it sounds, but it takes foresight from somebody who's built a lot of crap over their career and seen how their bad decisions made things fall apart. :)