You wouldn't necessarily notice. A lot of the use of functional languages is finance-industry where companies prefer to keep the technology they're using secret.
This makes me uncomfortable. Use in a particular niche sector does not preclude use in a different sector. Finance may well secretly use all of these weird high-level languages (though how would you falsify this?), but that still doesn't explain why you don't see them anywhere else.
So either finance is amazingly brave or well-suited to niche functional languages (in which case use in finance proves nothing), or it's a conveniently reputable-sounding use that doesn't actually require any proof. I, for one, will continue treating visible, published software (and libraries) as a useful metric of programming language usefulness and popularity.
I'd say it's the opposite: use in visible, published software and libraries is a very specific niche (or rather two specific niches: OSS and webdev), and not representative of software as a whole. The open-source community is a Unix- and therefore C-oriented community; a lot of Haskell work happens at Microsoft, and a lot of JVM work happens at Oracle, and a lot of OCaml work happens in Europe, which puts all those things somewhat at odds with RedHat. Webdev is uniquely ill-suited to typed languages because of the history of browsers and HTML in particular. If you just looked at published software/libraries you'd think no-one was using Java or C++, when those are likely still the most popular languages in software development as a whole.
facepalm
Perhaps some Haskellers should take the effort and develop a really sophisticated Haskell IDE which Haskell beginners could follow and use to learn and advance in Haskell. Theoretic knowledge about Haskell is good but practical application is another thing.