Parmandur
Book-Friend
Oh, so you have internal information about WotC market research...? I'll take Crawford's word that is what their data suggests, as he has said it more than the one time.A 2018 Crawford tweet based on a convenience sample is not very good evidence for a fact.
All WOTC has are convenience samples. By contrast, the majority of online discussion about 5e involves feats. I can make just as strong a claim about the representativeness of that anecdata as WotC can for their convenience samples.
If only a minority used them, why would they keep designing more?
A large enough minority uses Feats to make it worth their time, unlike some of the modules mentioned in this thread, particularly since they have the balance worked out with ASIs pretty tightly. A large amount of online chatter represents a tiny fraction of the D&D player base, and is less reliable than WotC own info.