Why you’re in zone 5 on your easy runs
This comes up at run club constantly. Someone finishes a conversational jog, opens Strava, and the whole run is red - 40 minutes in zone 5, average 178. They conclude they're desperately unfit. Usually what they've got is heart rate cadence lock.
It's cadence lock: the wrist sensor loses your pulse and latches onto your arm swing instead, so it reports your steps per minute as a heart rate. If the number is flat, high from minute one, and suspiciously close to your cadence, that's what you're looking at. Get an armband or a chest strap before concluding anything about your fitness - and if you can hold a conversation, you're in zone 2 whatever the watch says.
Cadence lock, specifically
Optical sensors shine a light into your wrist and count how the reflection changes as blood pulses past, roughly speaking. It works fine sitting still. Running, it has to pick your pulse out of a wrist that's swinging, a strap that's shifting, tendons moving underneath, and skin that's cold and short of blood in the first mile.
When the signal gets bad enough the algorithm latches onto the strongest rhythm it can find, and while running that's your arm swing. Cadence is typically somewhere around 170 to 185 steps per minute, which is also a completely plausible heart rate - so the watch reports a number that looks real, holds steady, and is entirely made up. That's cadence lock, and once you know the name you'll spot it in about four seconds.
It also explains the specific thing that upsets people, which is a whole easy run coming back as zone 5. A genuine 178 for forty minutes would have you on the floor. A cadence-locked 178 feels like a jog because it was a jog.
The quickest check costs nothing: on your next easy run, glance at your cadence and your heart rate at the same time. If they're within a few beats of each other, you're looking at your own footfalls.
The three options, having bought all three
Genuinely good at rest, at sleep, and on steady efforts once you're warm. Bad in the cold, bad in the first mile, bad during intervals, and bad at the top end - it undercounts near max, which quietly wrecks your zones in the other direction. Tighten it a notch and move it above the wrist bone before giving up on it; that alone fixes a lot of cases.
Also optical, but on a part of you that doesn't swing, so most of the problem goes away. Mine is the Coros armband ↗, the Jakob Ingebrigtsen edition, because why not. This was my third purchase and it took me a lot of fiddling to get right - the position that finally worked was the outer muscle belly of the tricep. Not the forearm, not the front of the bicep. If yours reads badly, move it before you return it; a couple of inches changes everything, which nobody tells you.
Electrical rather than optical, so it's measuring the actual signal and it is the reference. The cheap ones off Amazon are reliable - mine is a Cospoo ↗, about as Amazon-special as it gets, and it does the job. This is not a place you need to spend. My problem with it is practical: mine slides down and needs pulling up, and having to pull it up during intervals or hard runs is a dealbreaker for me, so I stopped wearing it. Wet it before you put it on, which helps, and accept a minute of nonsense at the start while it settles.
Before you buy anything
The talk test probably outranks all three of them. If you can speak in full sentences you're running easy, and no number is allowed to overrule that. I'd rather someone ran the right effort with no data than the wrong effort with a confident readout, and a cadence-locked watch is the second one - it tells people their easy run was hard and they either speed up out of spite or feel bad about themselves.
And if you're going to trust the number, make sure your zones are yours first. Mine run 43 asleep to 205 running, so a chart built on 220-minus-age was never going to describe me.