For thirty nights, one very patient member of our team went to bed wearing a smart ring, a smartwatch, a fitness band, and an EEG headband — simultaneously. The goal wasn't to crown a winner. It was to answer a simpler question: when four devices measure the same night, do they agree?
Short answer: on some things, remarkably well. On the things they advertise hardest, not even close.
What wearables measure well
Heart rate and heart-rate variability. Overnight resting heart rate tracked within 1–2 bpm across all four devices, night after night. HRV trends moved together too — when one device showed a rough night after late meals or alcohol, they all did. Validation studies back this up: optical heart-rate sensing during sleep, when movement is minimal, is genuinely good.1
Sleep duration and timing. Time asleep agreed within about 15–20 minutes across devices on most nights. If what you want to know is “when do I actually fall asleep and how long do I sleep,” a wearable answers that honestly.
Where they fall apart
Sleep stages. This is the feature on every box — deep sleep, REM, light sleep, drawn in confident colored bars. Compared against the EEG headband (the closest consumer proxy to how sleep is actually staged), the wrist and finger devices frequently disagreed by 40–60 minutes on deep sleep alone — and disagreed with each other even more. The published literature says the same thing: wearables reach roughly 60–80% epoch-level agreement with polysomnography, with deep sleep the weakest link.2 A wearable infers stages from heart rate and movement; it isn't measuring your brain.
The sleep score. Each morning, our four devices produced four different scores for the same night — sometimes 15 points apart. The score is a proprietary blend of real measurements and assumptions, which makes it the least comparable, least scientific number on the screen. Researchers have even coined a term for chasing it: orthosomnia — anxiety about sleep scores that measurably worsens sleep.3
How to actually use one
- Trust trends, not absolutes. “My deep sleep dropped this week” is signal. “I got exactly 1h42 of deep sleep” is not.
- Use overnight heart rate and HRV as your recovery dashboard — it's the most accurate data the device produces.
- Ignore night-to-night score fluctuations entirely. Compare weeks, not days.
- If the tracker makes you anxious at bedtime, take it off. The measurement is not worth the effect it's measuring.
A tracker is a decent thermometer and a terrible doctor. It can tell you something changed; it can't tell you why — and it was never going to.
3 sources
- Sargent C, et al. The validity of activity monitors for measuring sleep in athletes and healthy populations — systematic evaluations of consumer wearables. Sleep Med Rev (multiple validation cohorts).
- Chinoy ED, et al. Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep. 2021.
- Baron KG, et al. Orthosomnia: are some patients taking the quantified self too far? J Clin Sleep Med. 2017.