I just checked the
wowhead dps spec guides pages (which have API scrapes from 15 june till july 1) from that those scrapes suggests that:
- In 17/24 specs the most played covenant is what the wowhead guide suggests is the 'best';
- In 5/24 specs the most played covenant is
not what wowhead suggests is the 'best'; and
- I classified 2/24 as special cases because Enhance's most played was venthyr which was the strongest aoe but not the strongest st and Shadow where nightfae was the strongest but most played venthyr by a margin of 0.4%.
Like everything, there is obviously more going on here than raw numbers. While for example, Nf is the best druid covenant and one would imagine druids would be drawn to it, the same can't be said for locks and NF, where every single lock spec overwhelming played NF. Further, fully 3/5 guide opposed specs where hunters and the guide seemed to imply that before 9.1 NF was the superior hunter convenant (replaced by kyrian in this patch) so there may in fact be some lag time there (or hunters just don't give a shit).
And while all that is very interesting it seems to me this narrative that there is no evidence that
at least a large minority of players are playing what's strong based on the stats available is unsustainable.