A startup company is much like a multi-variable search. It's a search for a viable product, a viable business model, a viable company culture, a viable sales model, a viable team, a viable set of processes.... And all of this is being done at break-neck speed, in a rush against time and money.
As with all other things, a startup is likely to begin from scratch when it comes to process work. That is a good thing - it's a part of a strategy to remain mobile and flexible, as Pierre has said.
A process is a proven, repeatable set of steps, which bring about a reliable, predictable result - when the company is just starting, there are few places where they belong. There isn't much that's still reliably repeatable. And it's going to take some time before you can be certain that a given set of repeatable processes is the correct one, or incorrect one, or required but not discovered yet.
So yes, you'll see lots of best practices ignored/not practiced at startups, as a trade-off for flexibility and the ability to discover. That client running in and requiring something be finished tonight is not normally appropriate when you're fairly certain your SCRUM process works in this industry, and can produce good results for most people. But maybe a SCRUM process is not appropriate. Or maybe the tradeoff between burning out developers tonight and losing your only customer isn't something you can afford - maybe this is a Hail-Mary moment.
That said, it's generally accepted in startup culture that being enthusiastic about doing urgent work on the spot, and avoiding experimenting with processes, are both signs of a properly working startup. The savvy entrepreneur and early company builder, will have his wits about him and know when the tradeoffs are worth it, and when they're not. Either that, or he will fail, and learn in the process.