Sleep scores explained: what goes into Oura, WHOOP and Garmin numbers

A sleep score is not a measurement. It is a weighted summary of several measurements, some solid (total sleep, timing) and some estimated (stages), compared against targets the company has chosen. That is not a criticism. A single number is easy to act on. But two devices can give the same night different scores, and neither is wrong by its own rules.

What each company says goes in

Oura

Oura's help centre describes the Sleep Score as a 0 to 100 summary of the night and lists seven contributors. Total sleep, efficiency (the percentage of time in bed spent asleep) and latency are the most familiar. Restfulness and timing cover how disturbed the night was and when you slept. REM and deep sleep each have their own contributor, which means the score partly depends on stage estimates, the least validated part of any tracker. Oura measures heart rate, temperature and movement to produce those estimates.

WHOOP

WHOOP's approach starts from need rather than a fixed target. Sleep Need moves each day with how much sleep you usually require, how much Strain you took on, any sleep debt carried over and any naps. Sleep Performance then compares what you got with that need, and WHOOP now also factors in consistency of bed and wake times, sleep efficiency and "sleep stress", a measure of physiological stress during sleep. WHOOP's support material notes that this change can lower scores for shift workers and anyone with an irregular schedule. Stage data (light, slow-wave, REM) are reported but are not a named input to Sleep Performance in WHOOP's description.

Garmin

Garmin describes its Sleep Score as drawing on heart rate, heart rate variability, respiration rate, body movement "and other important data." Excellent scores are uncommon by design. In Garmin's own analysis of users in the last quarter of 2022, only 5 percent averaged an Excellent score, and Excellent nights clustered around 8 to 8.5 hours of sleep. Garmin also reported that users who slept less than five hours had no chance of a Good score.

Why the same night scores differently

• Different targets. Oura and Garmin compare you with fixed or population-informed targets for each component. WHOOP compares you with a need that changes daily.
• Different stage estimates. In a 2022 sleep-lab study, the same 53 people's deep sleep was estimated anywhere from 52 minutes too low to a couple of minutes too high, depending on device. A score that rewards deep sleep inherits that error.
• Different weightings. None of the three companies publishes how much each component counts. That is their proprietary work.

How to use a sleep score well

Use it as a prompt, then look underneath it. If the score drops, check which contributor moved. A low efficiency or long latency points to time awake in bed. A late timing contributor points to schedule. A low deep sleep contributor on one night is the least informative signal, because stage estimates are noisy. Track the 7-day average of the score rather than reacting to single nights, and always read total sleep in hours alongside it. The most consistent finding in sleep research is still the simplest: adults need seven hours or more on a regular basis, as the sleep and health guide explains.

Scores and age

Scores built on fixed targets can mark older adults down for normal ageing. The 2019 meta-analysis of 5,273 healthy adults in the sleep lab found that each decade of age reduced total sleep by 10.1 minutes and sleep efficiency by 2.1 percent, and added 9.7 minutes of wake after sleep onset and 1.1 minutes to the time taken to fall asleep. A healthy 65-year-old will therefore tend to score lower on efficiency and latency than a healthy 25-year-old, with nothing wrong. Trends against your own history are a fairer comparison than the label on the band.

Scores and resting heart rate

Sleep and recovery are linked in the data too. In 92,457 Fitbit users, the lowest average resting heart rate was seen in people sleeping 7 to 7.5 hours a night. That is one reason a run of short nights tends to pull down readiness and recovery scores as well as sleep scores.

What a sleep score cannot do

None of these scores is a medical assessment. Oura states plainly that its ring is not a medical device. A high score does not rule out sleep apnoea, and a low score does not diagnose insomnia. Heavy snoring, gasping, or daytime sleepiness despite a long night need a clinician and, usually, a sleep study.

Logging it

Peptide AI imports sleep scores and the underlying minutes from Oura, WHOOP and Garmin directly, and sleep data from Apple Health and Health Connect, through wearable integrations. The app's daily Peptide Score combines sleep with HRV, weight, resting heart rate and the dose log. The sleep stages page explains which parts of any sleep score rest on firm ground.