What sleep trackers can and cannot tell you

Most sleep trackers work out whether you are asleep from how much you move, and often from your heart rate. That is a sensible guess, and for total sleep time it is often a decent one. For anything finer, treat the numbers as estimates.

How a tracker decides you are asleep

A wrist or finger device has no way to see sleep directly. In a sleep lab, sleep is measured with polysomnography: sensors on the scalp record brain waves, and others track eye movements, muscle tone and breathing. That is how sleep stages are defined in the first place.

A wearable instead combines two signals. A motion sensor notices when you stop moving, which is the same idea clinicians have used for decades in a research tool called actigraphy. An optical sensor reads your pulse through the skin, and changes in heart rate and its variability help the device guess which stage you are in. The final picture is a prediction made by the manufacturer’s software, trained on people who wore both the device and lab sensors.

What they are fairly good at

• When you went still, and when you started moving again.
• A rough total for the night, especially over a week rather than a single night.
• Trends: whether you are generally sleeping more or less than you were a month ago.
• Regularity: whether your bedtime and wake time drift around or stay steady.

When researchers have put consumer devices side by side with polysomnography, the better ones match the lab well at telling sleep from wake across a night, and often do as well as or better than research-grade actigraphy. That is genuinely useful.

Where they go wrong

The same studies show a consistent weak spot. Trackers are good at recognising sleep and poor at recognising quiet wakefulness. If you lie very still for forty minutes, awake and thinking, a device that mostly relies on movement may log that time as light sleep.

That cuts both ways. For a good sleeper it slightly flatters the night. For someone with insomnia, who often lies still and awake for long stretches, the tracker can show more sleep than they feel they had. Neither the feeling nor the number is the full truth: people who sleep badly also tend to underestimate how much they slept, which we explain in what normal sleep looks like.

What they guess at

Sleep stages (light, deep and REM) are properly measured in a sleep lab with sensors on the scalp. A wrist device infers them from movement and heart rhythm. The result can be close, or it can be quite far off, and the device will not tell you which. Agreement with the lab is noticeably weaker for stages than for total sleep, and varies between brands and between people.

The single “sleep score” many apps show is a further step removed. Each company builds its own formula from these estimates, weighs them differently, and rarely publishes how. Two devices can give the same night quite different scores.

A score of 71 is not a diagnosis.

One night tells you very little

Sleep varies a lot from night to night, even in people who sleep well. Brief waking several times a night is normal, and most of it is never remembered. A short night after a stressful day is not a trend. If you use a tracker at all, look at the weekly average and the general shape, and ignore any single morning’s result.

It also helps to decide in advance what you would do differently with the information. If the honest answer is nothing, the number is only something to worry about.

When the tracker becomes the problem

Some people start to sleep worse because they are trying to improve a number. Sleep researchers have a name for it: orthosomnia. If you find yourself checking the score the moment you wake, or lying awake worrying about tonight’s result, the kindest thing to do is to put the tracker in a drawer for a week and see how you feel.

Trying hard to sleep well is one of the things that keeps people awake. A device that grades every night can turn rest into a performance, which is exactly the pressure that techniques like paradoxical intention are designed to remove.

Using one well

• Wear it for the long view: weeks, not nights.
• Use it to keep a regular wake time, which matters more than almost anything else it measures.
• Ignore the stage breakdown unless a clinician asks for it.
• Turn off notifications and scores if they make you anxious. The device still records in the background.

If the numbers worry you

American sleep medicine specialists are clear that consumer trackers are not diagnostic tools, though they can be a useful way into a conversation. If a tracker suggests something is genuinely wrong, such as very short sleep night after night, or loud snoring and pauses in breathing that someone else has noticed, take it to a doctor. They can arrange proper tests. The tracker has done its job if it got you there, and we have written about what to ask when you go.