Piss poor? Measurements done by The Quantified Scientist on YouTube seem to disagree with that. Also what I've checked with my Polar H10, the AW sensor has been really close if not spot on. At least I would like blood pressure measurement for the same reason I use it on my Galaxy Watch, i.e. to see if there's a reason to check more closely with my Omron.
Piss poor in my usage, I can't comment on results others are getting. Any activity with rapidly changing HR is awful, it takes at least 5-10 seconds, sometimes more, for HR to catch up with HIIT type activities (and often it's just searching) and makes it useless, I have to resort to using a much less comfortable chest strap. I also regularly check my HR on my AW against my high end SOLE treadmills hand grip HR sensors as well as a HR strap directly on my chest and I can't say I ever found them matching, and often wildly different. There are even times when I get no HR measurement for several minutes! This is pretty random and has happened with different workout types, MMA, HIIT, steady state, weight lifting, etc. I've noticed these deficiencies across several generations starting from an AW3 to my current AW Ultra.
For steady state where HR is averaged, and possibly for health overall I'm sure the HR in the AW is fine, but for HIIT it's beyond terrible, even Apple advises
this. From what I understand the AW measures HR every 5 seconds then averages this. This makes sense as a few
studies found that it loses accuracy the more strenuous the activity, with good accuracy while walking but getting worse going to jogging and then running, and that was steady state not HIIT. With that said it's not that I'm particularly blaming Apple, maybe the technology just isn't there yet. But I do know that a few years ago I was using a forearm mounted Scosche bluetooth band that was just as accurate as a chest band so I'm not sure if the wrist placement is what makes such measurement difficult. It's interesting to note that the cited study thought possibly decreasing accuracy was because increasing arm movements might have caused movement artifacts, but that wouldn't jive with my own experience with the Scosche band. They also speculate the possibility of Apple's algorithm possibly throwing out "physiologically implausible" readings, interesting.