What does a sleep score actually measure?
Plenty of apps show a number out of 100 every morning. What is really behind it: what a microphone can measure, what a watch adds, and what an honest score should refuse to do.
Every morning, millions of phones deliver a verdict: 72. Or 85. Or 54. A number out of 100, looking very sure of itself, claiming to sum up your night.
The awkward question: measured from what? It deserves a clear answer, because it separates the scores that inform from the scores that decorate.
A score is not a measurement, it is a recipe
Nobody "measures" a sleep score. You measure things — a duration, noises, a heart rate — then blend them according to a house recipe: so many points for duration, so many for calm, so many for regularity. The final number depends as much on the recipe as on your night.
That is why two apps grade the same night differently, and why comparing your score with a friend's is meaningless. The only comparison that holds is you against you: the same recipe, applied to your successive nights.
Provided, that is, the ingredients are real. And that is where everything is decided.
What a microphone really measures
A phone on the nightstand hears plenty of useful things: when calm settles, when the night turns restless, how long the noises last, whether there was snoring and at what time. These are direct measurements — dated, replayable, checkable by ear.
What a microphone cannot do is see your sleep stages. Light, deep and REM sleep are told apart by heart rate, its variability and the body's micro-movements — signals no microphone picks up. An app that has only sound yet still displays "1 h 42 of deep sleep" is not showing you a measurement: it is showing an estimate dressed up as one, typically the average proportions of a textbook night, redistributed over your quiet periods.
Estimating is not a crime — as long as it is announced as such. We covered what a phone alone can and cannot know in sleep stages without a smartwatch.
What a watch adds
A watch or ring worn overnight measures precisely what the microphone lacks: heart and movement. With those signals, stage estimation becomes a genuine scientific approximation — imperfect next to a sleep lab, but grounded in data from your body, that night.
Combining the two is the honest setup: the microphone for what can be heard (noises, snoring, timing), the worn sensor for what must be measured from the body (stages, heart rate). That is the choice Hoot made: without a watch it shows your sounds and your timing — and refuses to compute a score, rather than building one on guessed stages. The score only appears when a watch provides real stages through the Health app.
How to read a score without being fooled
Three reflexes are enough:
- Ask where it comes from. Every displayed number should answer the question "measured or estimated?". A serious app says so; be wary of those that never make the distinction.
- Follow the trend, not the number. 72 means nothing in the absolute. Dropping from 72 to 60 every time you drink in the evening — that is information.
- Keep the ability to check. A score that rises while you feel wrecked must be cross-checkable: replay the night's moments, look at the actual times. If the app shows only the number, it is asking to be taken on faith.
In short
A sleep score is a recipe applied to measurements — and it is worth what its ingredients are worth. A microphone measures sounds and timing, not stages; a watch measures what the microphone cannot; and a score built on guessed stages is a decoration, not data.
The right reflex fits in one question, to ask of any app: "that number — did you measure it, or did you make it up?" The ones that answer plainly deserve your trust. The others, your curiosity — at best.
Frequently asked questions
Can a phone measure my sleep stages?
Not in the strict sense. A microphone hears calm, noises and their timing — enough to estimate when the night was restless. Sleep stages (light, deep, REM) are measured from heart rate and movement, which requires a worn sensor such as a watch or ring.
Why do two apps give different scores for the same night?
Because a score is not a measurement but a recipe: each app picks its ingredients (duration, regularity, noise, stages) and their weighting. Two recipes give two numbers — neither of them is 'the real one'.
What is a score good for, then?
Comparing your nights with each other. The absolute number matters little; how it moves with your habits — alcohol, schedule, screens — is real information, provided the score is built on genuinely measured data.