What your sleep score actually measures

You wake up, reach for your phone, and a number has already made a judgement about your night. Seventy-two. Eighty-four. Fifty-one, and suddenly the whole day feels heavier than it did ten seconds ago.

That number is doing something genuinely difficult: compressing several hours of a brain cycling through stages into one figure, using only the signals available at your wrist. It is useful. It is also often misunderstood, including by people who check it every morning for years.

So here is the honest version — what a sleep score is built from, what the published validation research says about the parts it is built from, and how to read it without handing it more authority than it has earned.

A score is a summary, not a measurement

Your heart rate is a measurement. Your step count is a measurement. Your sleep score is neither: it is an opinion expressed as a number, assembled from measurements by a formula somebody inside a company chose.

Most sleep scores blend some combination of how long you slept, how continuous it was, when you slept relative to your usual timing, how your heart behaved overnight, and how much you moved. Those are reasonable ingredients. But the weighting — how much a short night should cost you, how much a late bedtime should cost you — is a product decision, not a clinical standard.

This has one consequence worth remembering above all others: scores are not comparable between brands. An 80 on one device and an 80 on another are not the same claim about your night. They are two companies' separate opinions, expressed in the same borrowed vocabulary.

A score is only ever meaningful against your own history, on your own device.

What your wrist actually sees well

Underneath the score sit real signals, and some of them are good. A wrist-worn band measures movement through an accelerometer and blood flow through an optical pulse sensor. From those two streams it can infer, with decent reliability:

• When you fell asleep and when you woke up — the broad shape of the night.
• Roughly how long you slept — useful at the weekly level, less exact on a single night.
• Your overnight heart rate, including the resting low point, which is one of the more stable and genuinely informative numbers a band produces.
• How restless the night was — movement is the thing an accelerometer is actually built to detect.

The validation literature supports this. In a 2025 laboratory study, researchers put six consumer wrist devices — including the Fitbit Charge 5, Garmin Vivosmart 4, WHOOP 4.0 and Apple Watch Series 8 — against polysomnography, the full clinical sleep study with electrodes on the scalp. Every device correctly identified more than 90% of the moments when participants were asleep.

That is a real result. Your band is not guessing.

Where the research says it drifts

The same study found the mirror image of that strength. While the devices caught sleep well, they were considerably weaker at catching wake.

The reason is almost mundane. Lying still in a dark room with a slow heart rate, waiting to fall asleep, looks nearly identical at the wrist to light sleep. There is no movement to detect and the pulse has already settled. So the device makes the more probable call and scores it as sleep.

The practical effect is that most trackers slightly overestimate how long you slept, and systematically underestimate the time you spent awake in bed. If you have ever lain awake for forty minutes and then been told you slept eight hours, this is why — and you were right, not the band.

The same study also found that most of the devices differed significantly from the lab on total sleep time, sleep efficiency and time spent awake after first falling asleep. A 2024 systematic review reached a similar place from a different direction, rating consumer accuracy for stages and total sleep time as moderate, and noting that the published evidence base remains thin.

The deep sleep problem

Of everything a sleep tracker shows you, the stage breakdown is the part to hold most loosely.

Light Deep REM Awake

Clinical sleep staging reads electrical activity in the brain. A wrist band cannot do that, so it infers stages from heart rate patterns, the variation between beats, and movement. That inference works better than chance — the Apple Watch Series 8 reached the highest agreement in the 2025 study, and the authors noted that the better devices could be useful for spotting sustained changes in sleep architecture over time. But agreement across the six devices still topped out at moderate, and the authors' own conclusion was that all of them need improvement for multi-stage classification.

Which means: if your deep sleep reads 42 minutes one night and 71 the next, a meaningful part of that swing may be the estimate moving rather than your biology. Chasing it is usually wasted attention.

Four numbers worth more than the score

If you want to actually use a sleep tracker rather than be managed by one, read these instead.

1. Duration, averaged over a week

The best-measured thing your band produces, and the thing with the clearest link to how you feel. One short night tells you little. Seven nights averaging five and a half hours tells you a great deal.

2. Consistency of timing

When you go to bed and when you get up, night after night. This requires no sensor accuracy at all — a clock is enough — and it is one of the most reliable levers you have over sleep quality.

3. Your overnight resting heart rate, as a trend

Measured directly rather than inferred. A resting heart rate sitting several beats above your normal range for a few nights often shows up before you consciously notice you are getting ill, under-recovered or over-stressed.

4. How often you woke up

Fragmentation matters, and movement-based detection is comparatively good at catching the real awakenings — the ones where you actually moved. Six hours in one piece is not the same as six hours in nine pieces.

How to use a sleep score well

None of this means you should ignore the number. It means you should treat it as a prompt rather than a verdict.

• Read the direction, not the digit. Three declining nights is information. A drop from 84 to 79 is noise.
• Check it against how you feel. If the score says 62 and you feel excellent, you feel excellent. Your own experience is the ground truth the score is trying to approximate.
• Don't let it set your day. There is good evidence that anxiety about sleep makes sleep worse. A number that makes you nervous at 7am has stopped being useful.
• Never compare across brands. Or across people. It is a private index.

The point of measuring your sleep is to change something about it. If a number isn't doing that, it is decoration.

Questions people actually ask