Sleep and Recovery Scores, Honestly Assessed

A sleep tracker measures two things: how much you moved and what your heart was doing. Everything above that, the stage graph, the sleep score, the readiness number in the morning, is an algorithm's estimate built from those signals. Duration and consistency are the parts worth trusting and acting on. The stage breakdown deserves a glance and no more, and a recovery score is a proprietary model rather than a measurement of your body.

None of this is medical advice. A watch cannot diagnose a sleep disorder, and persistent sleep problems, daytime exhaustion or breathing trouble at night need a doctor, not a better tracker.

What the device actually records

Overnight, a wrist device samples accelerometer data and heart rate, and on some models heart rate variability, skin temperature and blood oxygen. Movement tells it whether you were still. Heart rate and its variability tell it something about how settled your nervous system was. Those are real signals, and the overnight window is one of the better conditions for an optical sensor: you are warm, still, and the watch is not being yanked around, which is why resting heart rate from a night's sleep is more dependable than heart rate during a hard interval.

What no wrist device records is brain activity. Sleep stages are defined in a lab by brain waves, eye movement and muscle tone, measured with electrodes. Your watch has none of that, so it cannot see stages. It infers them.

Staging is a model, not a measurement

The stage graph comes from a classifier: movement and heart rate patterns go in, a guess at light, deep and REM comes out. Each brand trains its own model, weights the inputs differently and smooths the output differently. That is why two devices worn on the same person on the same night routinely disagree, sometimes by a large margin on deep sleep, and why the same night can look restorative on one app and poor on another.

Both cannot be right, and there is no way for you to tell which is closer. This is the single most useful fact about sleep tracking. Once you accept that the stage split is a model's opinion, the temptation to chase more deep sleep as though it were a dial you could turn evaporates.

Total sleep time is more solid, because the boundary between lying still with a low heart rate and being awake and moving is a much easier thing to detect than the boundary between one stage and another. It is still an estimate, and trackers tend to be generous, scoring quiet wakefulness as sleep. But it is an estimate of something the device can nearly see.

The parts worth acting on

Two numbers survive scrutiny, and they happen to be the two that matter most.

Duration. How long you slept, tracked as a weekly average rather than a nightly figure. One short night tells you nothing you did not already know. Four in a row is a pattern, and it is actionable in an obvious way.

Consistency. When you went to bed and when you got up, and how much those times wander across the week. Regular timings are one of the few sleep behaviours you can genuinely change by decision, and a tracker's bedtime chart shows drift far better than memory does.

If you only ever looked at those two charts and ignored every score on the device, you would lose almost nothing. Most of the honest value in sleep tracking is a record of the boring things you would otherwise misremember.

How readiness and recovery scores are built

A recovery, readiness or body battery score blends several inputs, typically overnight heart rate variability, resting heart rate, recent sleep, recent training load and sometimes skin temperature, and compresses them into one number. The inputs are real. The weighting is a business decision, and the exact formula is proprietary and undisclosed.

Three consequences follow. Scores are not comparable between brands, so a 74 on one watch means nothing on another, and comparing your score with a friend's is comparing two different formulas. Scores are relative to your own baseline, so they mean little in the first few weeks while the device is still learning your normal range. And a score can only reflect what it can see, which is why a stressful day, an argument or a long flight often shows up while a genuinely hard session sometimes does not.

They are also easy to mislead. A late meal, a warm room, alcohol or a night with the watch worn loose can all move the number without saying anything meaningful about your fitness or your health.

What they are genuinely good for

Used as a trend rather than a verdict, these numbers earn their place. A resting heart rate that has drifted up several beats and stayed there across a week is a real signal worth respecting, usually meaning accumulated fatigue, poor sleep or the start of an illness. Many people notice a clear overnight shift the night before they feel a cold coming on, which is genuinely useful for deciding whether to do the hard session. That is noticing a change in your own baseline, and it is not the device detecting illness or any medical condition.

Alcohol is the other one people find persuasive. A drink or three usually shows up as an elevated overnight heart rate and a suppressed recovery score, and seeing that pattern repeat changes behaviour more effectively than any amount of good intentions.

And the consistency chart catches slippage early. Bedtimes creeping later across a busy month is the kind of thing you only see in a graph.

Orthosomnia, and knowing when to stop looking

There is a well described trap here, sometimes called orthosomnia: the pursuit of a perfect sleep score becomes a source of anxiety, and the anxiety makes sleep worse. It is a real pattern, not a joke about worried people. Checking the score first thing shapes how you feel about a day that has not started yet. A bad number can talk you into feeling tired when you woke up fine, and lying in bed wondering whether you are getting enough deep sleep is a good way to get less of it.

The advice is plain. If the score is making you anxious, stop looking at it. Turn off the morning notification, hide the score in the app, or take the watch off at night for a fortnight and see how you feel. Nothing is lost: the device will still be there, and your sleep is not made worse by going unmeasured. Judge your rest by how you actually feel and function, which remains a better instrument than any wrist device, and if the worry persists, that is worth raising with a doctor too.

Common mistakes

Reading a single night as a verdict is the most common. Nightly numbers bounce around; only the weekly shape means anything. Chasing deep sleep specifically is next, since it is the least reliable part of the estimate and not something you can directly control. Third is comparing scores with someone on a different brand, which compares two formulas rather than two people. Fourth is skipping training because the number said so when you feel fine, or, worse, pushing through a session you feel terrible about because the number said you were ready. The score is one input, and your own sense of how you feel outranks it.

The last mistake is treating any of it as a health assessment. Blood oxygen and ECG style features on consumer watches are wellness features, not clinical measurements, and no watch can tell you whether you have a sleep disorder. Snoring, gasping, long pauses in breathing or exhaustion that does not respond to more sleep are all reasons to see a doctor.

Questions people ask

Why do two devices give me different sleep stages?

Because each one is guessing from movement and heart rate using its own model. Neither is reading your brain. Disagreement between devices is the expected outcome, not a fault in either.

Is my sleep score wrong if I feel fine?

How you feel is the better measure. A score is a model's summary of a few signals, and it does not know about your morning, your mood or your week. If it disagrees with you, believe yourself.

Should I wear my watch to bed at all?

Only if it is comfortable and you find the duration and consistency data useful. Big multisport watches are unpleasant to sleep in and need charging anyway; a band is far easier to live with overnight. Charging habits matter more than you expect, so it is worth reading what really drains watch battery before committing to nightly wear.

Which numbers should I actually pay attention to?

Sleep duration, sleep consistency and resting heart rate as a trend. The rest is context. For the wider version of that argument across every metric your watch produces, see which health metrics are worth tracking, and for why overnight heart rate is more trustworthy than the number during intervals, see wrist against chest strap.