Are Sleep Trackers Accurate? What Your Phone Can and Cannot Measure

Your phone told you that you got 42 minutes of deep sleep and scored 71. It sounds precise. It is not a measurement in the sense that a thermometer is a measurement, and knowing the difference makes the data far more useful rather than less.

The gold standard, and why nothing you own is it

Sleep stages are defined by brain activity. The clinical standard, polysomnography, records electrical activity from the scalp along with eye movement, muscle tone, breathing and heart rhythm, and a trained scorer assigns stages in 30-second epochs. That is what "deep sleep" and "REM" actually mean.

Your phone has none of that. A watch adds heart rate and movement. Neither has any access to your brain. So every consumer stage breakdown you have ever seen is an estimate produced by an algorithm from indirect signals — and any product that presents it as a measurement is overselling.

What trackers genuinely get right

This is the part the sceptics oversimplify. Movement-based sleep estimation — actigraphy — has been used in sleep research for decades and it is good at several things:

• Sleep versus wake. The core judgement is reliable. You are strikingly still when asleep and not when awake.
• Duration and timing. When you fell asleep, when you woke, how long you were down. Solid, and far better than self-report — people are poor at estimating their own sleep, usually underestimating it after a bad night.
• Consistency. Bedtime regularity across weeks is one of the most useful things you can know about your sleep, and a tracker captures it effortlessly.
• Disruption. Restless nights show up clearly as movement, and ambient sound adds a second signal for what caused them.

The known weakness of actigraphy is the opposite case: lying awake very still. To a movement sensor that looks like sleep, so trackers tend to overestimate sleep in people with insomnia.

What is inferred rather than measured

Stage breakdowns. Deep, light and REM percentages come from pattern-matching the movement and heart-rate signal against what those stages usually look like. The inference is reasonable in aggregate and unreliable night by night. Independent comparisons against polysomnography consistently find consumer devices good at sleep-versus-wake and considerably weaker at four-stage classification.

Sleep scores are a further step removed: a vendor-specific formula combining duration, timing, continuity and stage estimates into one number. A 71 on one app and a 71 on another are not the same quantity, and neither is a clinical result.

Phone, watch or headband?

• Phone on the mattress. Detects whole-body movement through the bed. Good for duration, timing and disruption. Confused by a partner or a pet in the bed. Costs nothing and needs no charging.
• Wrist wearable. Adds heart rate and heart-rate variability, which improves stage estimates. Still an estimate. Has to be worn and charged.
• EEG headband. Actually measures brain activity, so this is a real step up in validity — and most people will not wear one for more than a week.

The best tracker is the one you will still be using in three months, because the value is almost entirely in the trend.

How to use a tracker well

Read trends, not nights. A single night's numbers are noise. Two weeks compared against what you actually did — alcohol, late meals, room temperature, bedtime — will tell you something you can act on.

Trust duration over stages. Duration and timing are the reliable outputs. Treat the stage split as a rough shape rather than a figure to optimise.

Do not chase the score. There is a well-documented failure mode, sometimes called orthosomnia, where anxiety about sleep data makes sleep worse. If a bad score ruins your morning, or you find yourself lying in bed worrying about what the app will say, the tracker has become the problem.

Believe how you feel over what it says. If you slept badly and the app says 88, you slept badly.

What MathWake does, stated plainly

MathWake estimates your sleep from your phone's accelerometer and ambient sound, sampled through the night. It produces total sleep time, a stage hypnogram, a score, and a count of noisy minutes as a rough snoring proxy. The stage classification uses established movement-scoring logic — neighbour smoothing and rescoring rules — rather than raw thresholds, and the score is weighted toward duration and efficiency, the parts the sensors are actually good at.

It does not measure brain activity, and it will never tell you it does. All of it is computed and stored on your device, and you can delete the entire history in one tap.

Frequently asked questions

How accurate are phone sleep trackers?

Good for duration, timing and disruption; weaker for sleep stages. Movement-based estimation reliably distinguishes sleep from wake, which is why it has been used in sleep research for decades, but stage breakdowns are inferred from indirect signals rather than measured.

Can a phone really detect sleep stages?

It can estimate them. Stages are defined by brain activity, and a phone has no access to that. Algorithms infer stages from movement patterns and, on a wearable, heart rate. The result is reasonable in aggregate and unreliable on any single night.

Why does my sleep tracker say I slept when I was awake?

This is the known weak spot of movement-based tracking. Lying awake without moving looks like sleep to a motion sensor, so trackers tend to overestimate sleep for people with insomnia. If you regularly lie awake still, expect the numbers to flatter you.

Should I stop using a sleep tracker if it makes me anxious?

Yes, or at least stop looking at the score every morning. Anxiety about sleep data measurably worsens sleep, a pattern sometimes called orthosomnia. A tracker is useful for spotting trends over weeks; it is not a daily verdict on how you slept.