While the advice of doing simple-that-works when starting is not bad, I'd add the caveat of "keep options open" when there's little to no cost in doing so. For

While the advice of doing simple-that-works when starting is not bad, I'd add the caveat of "keep options open" when there's little to no cost in doing so. For example, good database design will pay off in spades later on. For example 23 years ago I used a natural primary key, rather than a surrogate primary key, in a table. I've grown to regret that choice a LOT in the last decade. At the time employee numbers were unique and immutable, so perfect natural key. Since then we've grown, and they are often not unique (companies merge), not immutable (companies merge) and in some cases they're not even employees. So yes, absolutely, keep it as sinple as you can, and ship to paying customers as fast as you can. But make choices that keep as many options open as possible. Sure, most options will not be realised, but the ones that are, which can easily be bolted on because of good basic foundations, are really valuable.