What Is a Sleep Score, and Should You Trust It?

You wake up, reach for your phone, and find that last night was a 74. Not terrible. Not good. You felt fine until you saw it.

Sleep scores are now on nearly every wearable, and almost nobody explains what the number is made of or how much confidence it deserves. This is that explanation — including the parts that are less flattering to the companies selling them, ours included.

What a sleep score actually is

A sleep score is a composite index: several separately-measured things about your night, weighted and combined into one number, usually out of 100. It is not a measurement. It is a summary of measurements, and the weighting is a design decision made by the company that built it.

That last point matters more than anything else in this article. There is no standards body for sleep scores. Every vendor picks its own inputs and its own weights. A 74 on one device and a 74 on another are not the same claim about your night — they are two companies’ opinions, expressed in the same units.

What goes into the number

The inputs vary, but nearly every sleep score is built from some combination of these:

Read that list again and notice what it implies: a sleep score mostly rewards sleeping a long time without waking up. The exotic-sounding parts — stage balance, HRV recovery — usually move the number far less than simply being asleep for eight hours.

What your tracker can actually measure

Here is the part most sleep-score explainers skip.

The clinical standard for measuring sleep is polysomnography — an overnight study in a lab. It records electrical activity in your brain (EEG), your eye movements, and your muscle tone, among other signals. Sleep stages are defined by those brain and eye patterns. That is what a sleep stage physically is.

A wrist wearable has none of that. It has two sensors:

• An accelerometer — it detects movement. Lying still for a long stretch looks like sleep.
• An optical heart sensor — it shines light into your skin and reads changes in blood flow, producing heart rate and heart rate variability.

Some add skin temperature or blood oxygen. None of them read your brain. So when your watch tells you that you got 47 minutes of deep sleep, it did not observe deep sleep. It observed that you were still, your heart rate dropped into a particular range, and your heart rate variability behaved a certain way — and a model trained on people wired up in sleep labs concluded that this pattern usually means deep sleep.

This is not a scandal, and it doesn’t make the number useless. It means the number should be read as an estimate with error bars rather than a readout. The problem is that a clean two-digit score on a bright screen doesn’t look like an estimate. It looks like a fact.

Why two devices disagree about the same night

Wear two trackers to bed and you will very likely get two different scores. Three reasons, in order of how much they matter:

1. 1Different formulas. The weights are proprietary. One vendor may treat six hours of sleep as a serious penalty; another may care more about how fragmented those six hours were.
2. 2Different stage models. Each company trains its own classifier on its own data. Given identical raw signals, two models will draw the stage boundaries in different places.
3. 3Different hardware and fit. Sensor placement, strap tightness, skin tone, tattoos and wrist hair all affect optical heart readings. A loose band genuinely produces worse data.

The practical consequence: never compare your score against someone else’s, or against your own on a different device. The only comparison that means anything is you, on the same device, over time.

What counts as a good sleep score?

Most devices use bands roughly like these:

A more useful frame than the bands: most healthy adults sleeping normally score in the 80s. Consistently sitting in the 90s usually means you are sleeping unusually long, not unusually well. And a scattering of low nights in an otherwise steady month is ordinary human sleep, not a problem to solve.

When to ignore your sleep score entirely

There is a documented failure mode here, and it has a name. In 2017 researchers writing in the Journal of Clinical Sleep Medicine described orthosomnia — patients whose determination to achieve perfect sleep-tracker numbers was making their sleep measurably worse. The anxiety about the score was more disruptive than whatever the score was reporting.

It is a genuine risk, and it is the direct consequence of designing a wellness feature like a video game. Some practical guardrails:

• If you feel rested, you slept well. Your own experience outranks the estimate. A poor score after a night you felt good about is a reason to distrust the score, not your body.
• Don’t check it first thing. Reading a low number before you’ve got out of bed reliably colours the whole day.
• Ignore any single night. Illness, alcohol, a hot room, a loose strap, a late meal — a hundred things move one night’s score.
• If it’s making you anxious, turn it off. A wellness feature that costs you sleep has failed at its only job.
• Persistent poor sleep is a doctor’s question, not a watch’s. No consumer wearable can diagnose a sleep disorder, and none should try. If you are consistently exhausted, snoring heavily, or waking gasping, that is a conversation with a clinician.

How to actually improve the number

Since total sleep time and continuity dominate most formulas, the things that move a sleep score are unglamorous and well-established:

1. 1Keep a consistent schedule — same bedtime and wake time, weekends included. Regularity affects the score directly and everything else indirectly.
2. 2Give yourself enough time in bed. Most adults need 7–9 hours. You cannot score well on six.
3. 3Keep the room cool and dark. Both reduce the night wakings that the score penalises most.
4. 4Move the last drink earlier. Alcohol helps you fall asleep and reliably fragments the second half of the night — which is exactly what the score measures.
5. 5Cut caffeine after early afternoon. Its effects last far longer than most people assume.
6. 6Wear the band correctly — snug, slightly above the wrist bone. Poor contact produces bad heart data, and bad heart data produces a bad score regardless of how you slept.

That last one is worth taking seriously. A meaningful share of inexplicably terrible sleep scores are strap problems, not sleep problems.

What we do differently

Two things, and we’d rather state them plainly than imply we’ve solved sleep science.

We interpret your night on your device. The signals from your wrist are read on your watch and phone rather than uploaded to a server for scoring. There is no central record of how you slept — which is a structural difference, not a policy promise.

We describe your night rather than grading it. A number is easy to obsess over and hard to act on. Tribe leans toward plain-language reflection — what changed, what it might relate to, what would help — because that is closer to how sleep actually improves. And there is no subscription attached to any of it.

What we cannot claim: that our stage estimates are meaningfully better than anyone else’s. Nobody wearing a wrist device is measuring brainwaves, and any company implying otherwise is overselling. What we can promise is that the reading stays yours — you can see exactly what we do and don’t collect, and erase all of it in one tap.

So: should you trust your sleep score? Trust it the way you’d trust a bathroom scale — useful over months, noisy day to day, and never the last word on how you actually feel.