> 1. You already have a confident haskeller or two on your team, ideally senior
There are really only three reasons any particular tech stack is ever selected for any project.
1. It's the standard at the company and you don't have a choice 2. You happen to be working in an extremely specialized field where one language is dominant (e.g. Haskell for compilers, python for AI) 3. Someone picks a language that they like and are confident in
I'd suggest that across the entire industry, (1) dominates by a large margin. It doesn't apply as much to startups though, because by definition startups aren't going to have a lot of existing code and won't have developed restrictive standards.
So yes, the biggest reason to choose Haskell is that you have a confident senior haskeller on the team.
> 2. Your tech needs are vanilla and back-end heavy -- writing UIs in haskell is a nonstarter, and heaven help you if you're not doing vanilla web apis + postgres etc
I agree with you about UIs. It's not a good experience and I wouldn't suggest Haskell for that. There are other specific areas where I wouldn't suggest it outside of a hobbyist project, like games or mobile development. There are still a lot of programs that need to be written and benefit from Haskell's strengths though, and sure, a lot of them are CRUDy web API things, because frankly that's most of all software that gets written.
> 3. You want primarily junior- to mid- talent (great for keeping burn low)
Haskell is perfectly viable if you want to hire a team of junior and mid level developers (so long as you have a more experienced person on the team to help them learn), but not exclusively useful in that situation. There are experienced Haskell developers out there, and there are a lot of highly skilled and experienced developers who are interested in- or at least willing to learn- haskell.
In general, I'd say that Haskell is a great choice for _most_ types of development, and can work for teams with a wide variety of experience levels. It's a real asset for hiring because there are more people who would like to work with Haskell than there are Haskell jobs out there, and in general I think Haskell still benefits from the python paradox (https://paulgraham.com/pypar.html) where people who know or are interested in it tend to be above average.