Sleep score: what the number means, what goes into it, and why brands disagree

What is a sleep score, in plain terms?

A sleep score is a single grade for a night. Oura presents its score on a scale of 0 to 100.³ The number is not something the device measures directly. It is a summary, worked out from several other measures such as how long you slept, how often you woke and when you went to bed.

That makes it different from a stage chart. A stage chart describes the night, with light, deep and REM sleep laid out in order. A score tries to turn the whole night into one verdict. If you want the description rather than the grade, start with the guide to sleep stages.

There is no shared formula. Each brand picks its own ingredients, its own weights and its own bands. Oura says that scores of 100 are designed to be rare rather than regular.³ Google Health tells people to aim for the higher end of their own healthy range rather than chase the global average.⁵ Both pieces of advice point the same way: the number is for comparing you with yourself.

What goes into a sleep score?

Oura says its score is made up of seven “Sleep Contributors”: total sleep, efficiency, restfulness, REM sleep, deep sleep, latency and timing.³ Google Health describes six: sleep duration, time to sound sleep, sound sleep, restlessness, full awakenings and interruptions.⁵ Garmin says its score is calculated from a combination of sleep duration and sleep quality factors, and lists sleep architecture, stress data and interruptions among them.⁴

Apple is the most open about weights. Its Apple Watch score combines sleep duration (50 points), bedtime consistency (30 points) and interruptions (20 points).⁶ Those add up to 100, which is the whole scale. Apple looks at your bedtime consistency over the last 13 nights, and its interruptions measure takes in both how often and for how long you were awake.⁶

Two practical points follow. First, some scores need a certain kind of hardware. Google Health says a device must track sleep stages to produce a score.⁵ Garmin says models without a wrist heart rate sensor cannot detect sleep stages.⁴ Second, Garmin files heart rate variability under its stress factor, so heart rate variability can feed into the score through its stress element.⁴ For what HRV itself means, see the guide to heart rate variability.

What counts as a good sleep score?

It depends on whose number it is. The bands below come from the brands’ own pages. None of them is a clinical standard.

A few examples show how much the label moves. A score of 82 is “good” at Oura, “Good” at Google Health and “High” at Apple.³⁵⁶ A score of 85 is “optimal” at Oura, “Good” at Google Health and “High” at Apple. A score of 78 is “good” at Oura, “Fair” at Google Health and “OK” at Apple.

So there is no single threshold such as “80 and above is good”. Pick the bands of the app you actually use, and treat them as that brand’s labels rather than a verdict on your health.

Why do two devices give two different scores?

Three things differ: the sensors, the ingredients and the weights. Garmin says it works out sleep time and stages from heart rate, heart rate variability and movement.⁴ Apple’s published score leans on duration, bedtime consistency and interruptions.⁶ Oura’s lists efficiency, latency and timing.³ Feed the same night into those recipes and the result will rarely match.

There is a second layer underneath. Independent research has looked at how well consumer trackers measure the raw inputs, and the answers are mixed. A meta-analysis of 24 studies with 798 participants found significant differences between consumer wrist-worn trackers and polysomnography on total sleep time, sleep efficiency, sleep latency and waking after sleep onset.⁷ Its authors concluded that the devices are not as reliable as polysomnography for key sleep parameters, but can still be useful for following broad trends.⁷

A rapid review of 29 real-world studies rated precision as moderate for total sleep time and time in bed, and lower for sleep efficiency, waking after sleep onset and stage classification. It called REM and deep sleep estimates particularly unreliable.⁸ An older review from 2016 found that consumer devices tended to overestimate total sleep time and sleep efficiency in people without sleep problems, and noted that the algorithms are proprietary.¹⁰

None of the studies read for this guide validated a composite sleep score itself. They tested the ingredients. A score built from shaky ingredients will inherit their wobble, and no independent test of the final number was found.

A worked example: same time in bed, different sleep

Efficiency is time asleep divided by time in bed, and Oura counts it as one contributor.³ Take two nights of 7 hours 30 minutes in bed. This is arithmetic only, and does not copy any brand’s formula.

Both nights look identical on a bedside clock. Night B has just over an hour less sleep in it, and Oura names 85% as the optimal efficiency for an adult.³ Night B falls short of that, and Night A clears it. That is one way two nights of the same length in bed can score differently.

The NHS gives a useful anchor for the duration part: adults need 7 to 9 hours on average, and everyone’s needs differ.¹ Your score and your need for sleep are separate questions. A high score does not prove you slept enough for you, and a low one does not prove you slept too little. For the longer discussion, see how much sleep you need.

Sleep score versus how you actually feel

The NHS suggests a simple test of your sleep that needs no device: do you wake up feeling refreshed, have energy and be able to concentrate?² That is a different kind of evidence from a score, and it is the one that matters day to day.

The NHS also says a few nights of bad sleep are usually nothing to worry about.² It separates short-term insomnia, lasting under 3 months, from long-term insomnia, which lasts 3 months or longer.¹ A single poor score is a data point, not a condition.

There is a real caution about checking too closely. Baron and colleagues described “orthosomnia”, in which patients chase perfect sleep and seek treatment for problems they have diagnosed themselves from tracker data.⁹ They noted that the tracker’s numbers often feel more consistent with a patient’s experience than validated tests do.⁹ In an observational study of 943 London Marathon participants, tracker users reported worse sleep satisfaction. The authors suggested a possible harmful influence and said orthosomnia may be relevant, but a survey of this kind cannot show cause.¹²

The evidence does not say trackers cause anxiety. In a four-week trial with 26 people with insomnia, 14 using a wearable and 12 keeping a handwritten diary, wearing the device did not necessarily make sleep worries worse.¹¹ If you notice that a morning score ruins your mood before you have even got up, that is useful information about how to use it, and hiding the number for a week is a reasonable experiment.

What to do about a low score

Start by splitting the number into parts. Look at the contributors your app shows: was it duration, interruptions, timing or something else? Fixing the part that dragged the score down is more useful than chasing the total.

Then use the habits the NHS lists for better sleep. These include getting up at the same time every day, avoiding alcohol, tea and coffee in the 6 hours before bed, and not exercising in the 4 hours before bed.¹ Google Health’s page adds that improvement is personal, and suggests aiming for the higher end of your own healthy range.⁵ The wider list is in the guide to how to improve sleep quality, and if you are running short over several nights, sleep debt explains what that means. Two related reads are readiness score explained, which covers a different daily number, and heart rate recovery.

A 14-night check you can do with a pen

1. Each morning, before looking at the app, write down how rested you feel on a scale of 1 to 5.
2. Then write down the score, and what you did the evening before: caffeine, alcohol, a late meal or session, bedtime.
3. After 14 nights, put the two columns side by side. Do low feelings and low scores line up?
4. If they do not, trust the feeling. If both are low for weeks, take it to a GP rather than to another app.

The 14 nights is our suggestion, not a figure from a source. The idea is to learn how your own score relates to your own mornings.

How to use a sleep score sensibly

Notice what the routine leaves out: a target score. Without a standard that holds across brands, the most defensible use of the number is the trend in your own readings, set beside how you feel.

Frequently asked questions