Sleep tracking metrics: what to pay attention to (and what to ignore)

Consumer sleep trackers are useful, within limits. The metrics they report vary widely in how well they're validated against polysomnography (PSG), the lab gold standard. If you're choosing which specific device to buy, our smart ring comparison covers UK prices, battery life and validation evidence for the current major rings.

Metrics ranked by how trustworthy they are#

Total sleep time. Generally good agreement with PSG on modern wrist-worn devices when validated in healthy adults. The 2024 SleepScore-vs-PSG and Oura-vs-PSG comparison studies show errors in the range of 10–30 minutes per night, with systematic under- or over-estimation depending on the device.

Sleep efficiency (% of time in bed spent asleep). Reasonable correlation with PSG, similar accuracy to total sleep time on the same hardware.

Sleep latency. Less reliable. Wearables often misclassify quiet wakefulness as sleep, so reported latency tends to underestimate true latency.

Wake After Sleep Onset (WASO). Moderately reliable on devices that combine accelerometry with HRV.

Stage classification: light/deep/REM. Substantially less accurate. Comparisons with PSG show four-stage classification typically right 50–70% of epochs in good devices; REM and deep are most often confused. Trends within an individual are more useful than absolute values.

HRV during sleep. Now standard on Oura, Whoop, Garmin, Eight Sleep. Within-device night-to-night reliability is reasonable; absolute values vary by device, so don't compare a friend's number to yours.

What to actually pay attention to#

For most users:

• Total sleep time: are you averaging enough across a week?
• Consistency: is your wake-time variance under 30 minutes?
• Resting heart rate trend: week-on-week trends can flag overtraining, illness onset, or alcohol's effect.
• HRV trend — same idea; treat it as a chronic load signal, not a single-night score.

What to ignore#

• The exact "deep sleep" number on any single night
• Any "sleep score" composite that mixes incomparable metrics — most are vendor-specific marketing constructs
• Comparisons between different brands of wearable

When tracking becomes the problem#

A small fraction of users develop what researchers have nicknamed orthosomnia — anxiety driven by sleep-tracking data itself. If checking your sleep score in the morning is making your sleep worse, hide the data or pause tracking. The original paper (Baron et al., 2017) makes the point that no clinically validated treatment for orthosomnia exists because we created the problem.

Treat the wearable as a diary, not an exam result. The most useful question isn't "what was my score?" but "did this week trend in the right direction?"