Sleep tracking: reading the data without falling into obsession

Consumer devices generally combine movement, heart rate, and sometimes temperature or breathing. An algorithm infers categories such as "awake," "light," "deep," or "REM." The medical reference, polysomnography, instead records brain activity, eye movements, muscle tone, and breathing.

A 2025 meta-analysis compared consumer watches and wristbands with polysomnography across 24 studies and 798 people. On average, the devices poorly estimated sleep time, sleep efficiency, and especially nighttime awakenings, with discrepancies that varied by brand. A 2024 review notes that these devices do better than classic actigraphy, but often confuse being awake and still with being asleep, and perform less well for certain people, for example depending on skin tone or weight.

1. 01How you feel. How do you feel on waking? Does sleepiness interfere with your day?
2. 02Schedule. Bedtime and wake time, consistency from week to week.
3. 03Estimated duration. Useful as a trend over several nights, not as an exact measurement.
4. 04Stages and score. Read these last, as a lead for a question, never as a diagnosis.

Don't compare two brands as if their categories were identical, and be wary of sudden changes after a software update.

Sleep naturally varies: a late meal, travel, a cold, or a child waking up are enough to change the numbers. Instead, observe a defined period of two to three weeks, noting your wake-up time, how you felt, and any unusual events. If the numbers and how you feel contradict each other, that mismatch mainly says something about the device's limits.

In a 2018 trial with 63 adults with insomnia, participants were given fake feedback upon waking about the quality of their night, either positive or negative, unrelated to their actual sleep. Those who received negative feedback felt more tired and less alert in the evening. The displayed label alone was enough to influence the day.

Sleep physicians described in 2017 patients chasing perfect sleep based on their tracker's data, and proposed the term "orthosomnia." It isn't an official diagnosis, but a useful signal: if you check the app several times a day, if a bad score dictates your mood, or if you stay in bed awake to "improve" the numbers, take a break. Hide the stages, switch to a weekly summary, or put the device away for a while. A screen-free evening routine can help without measuring anything.

A tracker can neither confirm nor rule out a diagnosis. Loud snoring, observed breathing pauses, waking up gasping, dangerous drowsiness while driving, or lasting insomnia warrant medical advice, even if the app shows a reassuring score. You can bring a few weeks of data, specifying the model used: the World Sleep Society recommends presenting it as imperfect observations to discuss with your healthcare provider.