Another one that started as a Quora question:
Absolutely worry first about whether your product will be popular! (And likewise for a service, and whether it’s software or something else.)
If your software isn’t popular, you won’t ever need scalability. All the time you spend worrying about scalability first will just delay you from finding out whether it’s popular. That delay might be enough for some competitor to beat you to the market and establish enough of a foothold that your entry will flop, even if it would have succeeded if it got there first.
Put out an MVP ASAP, even if it’s horribly unmaintainable and unscalable! Chances are very high that you’re going to pivot to another idea, rewrite the software, or totally fail anyway. And even if your product is popular, it’s extremely difficult to figure out how much scalability you’ll actually need.
This is one of the reasons why I describe the ideal clients for my software process/tools/etc. advisory service as “established” (among other factors), by which I mean either they’ve been in business for at least a year or they’re turning a profit, preferably both. By this time they’ve figured out what people really want, and are ready to get serious about the “ilities” like scalability, maintainability, and so on. Before then, the focus needs to be on speed to market, and other ways of determining “product/market fit”.