Sleep Trackers: What They Get Right and Wrong

A ring or a watch cannot see your brain. That one fact explains most of what sleep trackers get right and most of what they get wrong. Sleep stages are defined by brain wave patterns measured in a lab with electrodes on the scalp. A wearable sees movement, heart rate, the spacing between heartbeats and skin temperature, and it uses those signals to guess what the brain was doing.

What they get right

The guess is good at the coarse question: were you asleep or awake, and for how long. Lying still with a slow, steady heart rate looks like sleep, and moving around with a faster heart rate looks like being awake. Against lab recordings, the better consumer devices land within roughly half an hour of total sleep time on average, though any single night can be further off, and bedtime and wake time are usually close.

Timing is the other strength. A tracker is very good at showing that you went to bed at 11:40 on Tuesday, 1:15 on Friday and 10:30 on Sunday.

The overnight vitals are also real measurements rather than guesses. Resting heart rate, heart rate variability and skin temperature come straight from the sensors, breathing rate is estimated from the pulse signal fairly well, and their night to night pattern tells us something about recovery, illness and alcohol.

What they get wrong

Stages are where the accuracy falls off. When a wearable calls a given stretch of the night deep, light or REM, it agrees with the lab roughly 60 to 80 percent of the time on a minute by minute basis, depending on the device. The device is inferring the stage from heart rhythm and stillness, and those signals overlap across stages. Deep sleep is the usual casualty. People read "42 minutes of deep sleep" and worry, when the real number could be well above or below that.

The other blind spot is quiet wakefulness. Someone who lies awake but still at 3 a.m. is often scored as asleep, so the tracker overstates sleep in exactly the people who sleep worst.

How to use the data

Use the device for the questions it can answer. Watch total sleep over a week, not one night. Watch how consistent bedtime is. Watch resting heart rate and HRV for drift that lines up with a hard training block, a few drinks or the start of a cold. Treat the stage breakdown as a rough sketch and treat a single bad score as noise.

If you wake up feeling rested and the device says you slept poorly, trust how you feel over the score. If you feel terrible most mornings and the device shows a decent score, that is still worth looking into. A repeated pattern of low oxygen or disturbed breathing overnight is a flag we take seriously regardless of the score.

We read members' wearable data alongside their labs and training, because the trend in the overnight numbers is what carries the information.

This article is educational and is not medical advice.

See what the HYBRD Membership includes