Body · lesson 7 of 10
Sleep tracking: what to trust
Wearables are good at when and how long you sleep and much shakier on stages. Plus: when tracking backfires.
7 min read · reviewed October 2026
What's actually being measured
The lab reference for sleep is polysomnography (PSG), which records brain waves, eye movements, muscle tone, breathing and oxygen. Only brain waves can truly tell light, deep and REM sleep apart.
Rings and watches don't record brain waves. They infer sleep from movement, heart rate, HRV, and sometimes temperature and breathing rate, using algorithms trained against PSG. The sleep stages in your app are a prediction, not a measurement.
How far to trust each tracker number
Good. Timing is what movement and heart-rate sensors capture best.
Fairly good on a typical night; average errors in validation studies are often minutes to tens of minutes. Worse on disrupted nights.
Weak. Lying still while awake often gets scored as sleep.
Inconsistent across devices. Fine for curiosity, not for decisions.
A company's own blend of inputs. Not a validated clinical measure, and not comparable between brands.
Some devices flag possible signs of sleep apnea. A notification doesn't diagnose apnea, and no notification doesn't rule it out.
In a 2021 lab study of seven consumer sleep trackers against polysomnography, every device correctly caught at least 93% of the time people were asleep. What share of the time people were actually AWAKE did the best device correctly detect?
Somewhere between 0 and 100 %.
Reveal the answer →
~54% (worst device: ~18%)
Specificity for wake ranged from about 18% to 54% (Chinoy et al., Sleep 2021). Trackers are good at recognizing sleep and poor at recognizing quiet wakefulness, and they performed worse on nights of poorer, more disrupted sleep, which are exactly the nights you most want accurate data. Stage estimates were inconsistent.
Lean on what trackers do well
Regular timing is itself a strong health signal. In 60,977 UK Biobank participants wearing wrist accelerometers, consistent sleep-wake timing predicted lower mortality more strongly than sleep duration did (Windred et al., Sleep 2024). The study is observational, but regularity and duration happen to be what trackers measure best.
Orthosomnia: when the data makes sleep worse
Some people get so focused on perfecting their sleep score that the worry itself disrupts their sleep. Sleep researchers call this orthosomnia.
How it works →
Insomnia is driven partly by effort and arousal around sleep. Checking a score every morning, spending longer in bed to 'fix' a number, or trusting the device over how you feel can all feed that cycle. And because stage estimates are unreliable, people can end up chasing a 'low deep sleep' problem that doesn't exist.
What the studies show →
The term comes from a 2017 case report in the Journal of Clinical Sleep Medicine (Baron et al.) describing patients seeking treatment for sleep problems they had inferred from tracker data, who trusted the device more than validated sleep measurements. It's a descriptive concept, not something tested in large trials; researchers have since developed questionnaires to measure it. The first-line treatment for chronic insomnia remains cognitive behavioral therapy for insomnia (CBT-I).
Tracking without harm
- Review weekly averages for duration and timing. Ignore single-night stage breakdowns.
- If a score contradicts how you feel, believe how you feel.
- Don't stay in bed longer just to raise a number; extra time in bed can worsen insomnia.
- If checking the app makes you anxious, hide the score or leave the device off at night for a few weeks.
The Sleep track covers the levers (light, caffeine, temperature, timing) that change the numbers you're tracking.