Upon close inspection, Statify was over-counting in two obvious ways. First, relative to Google Search Console stats (by about ~40%), and second it was picking up things like “page 2 of tag archive X” when it didn’t obviously follow from the articles that had been visited before. This may have to do with my page caching (WP Super Cache, advanced caching enabled, preload enabled), albeit that wouldn’t have explained the strange archive hits (tag and category archives are not cached). In any event, I saw there is a new Koko Analytics version, so I am giving it another go and not trying to re-import my old stats, which I salvaged in a Koko Analytics export in any event.
I briefly returned to Koko Analytics but noticed two things that made me leave it for the second time for Statify. Firstly, Koko was also recording hits for random tags which did not follow from any of the posts it recorded, which suggests it has a similar issue to Statify (Koko Analytics did not show taxonomy hits prior to version 2.0.0). Secondly, speed testing shows that NLJ is a little bit faster with Statify. Thirdly, Statify has one database table while Koko Analytics has six (granting Statify’s database grows at a quick clip due to how it saves data). I’ll look into seeing if there are some ways to make Statify a little bit more accurate (note that those ways do not include enabling async JS with ful page caching, that makes it record very little).