Iterating to a correct solution to a problem works well as a strategy for projects, but very poorly for smaller units of work like a feature or a bug fix. Often, doing the quick hacky solution for a feature won't show the problems it has for a while, by which time you've built things on top of it. Having to go back and fix that impacts far more than if you'd spent a little more time in the first place.

It's particularly important when you're designing data structure changes more than than code. Getting an API or a database table design right first time (or even just 'more right') is worth the time invested.