Why Sleep Scores Feel Broken (and What to Look at Instead)
Dovy Paukstys
Founder, Komori Care
What a Consumer Sleep Score Is Optimizing For
You woke up exhausted. You know you slept badly. You open your sleep app and it says 87 — "Good Sleep."
Or: you slept great. You feel rested. The app says 62 — "Needs Improvement."
Neither experience is rare. The mismatch happens because sleep scores are optimized for engagement, not your calendar. They're designed to be a number you check every morning, not a tool that explains what changed.
A sleep score is a weighted average of multiple signals — duration, heart rate, movement, maybe sleep stages — compressed into one 0-100 grade. The weights are proprietary. The algorithm may adjust over time. The score you get today for the same data might not match the score you'd get three months from now, because the company shipped an update.
That's not a bug. It's a product decision. A score is sticky, trackable, rankable. You can compare it to yesterday, graph it over weeks, share it with friends. It creates a daily ritual: check the app, see the number, feel validated or anxious, come back tomorrow.
What it doesn't do: tell you why last night was different from the night before, or what part of your routine, environment, or behavior actually changed.
Quick Answer
A single sleep score collapses different nights (travel, shift, sick, partner noise) into one grade that often tracks the algorithm more than your rest. If your job is understanding what changed overnight, look at movement dose, position timeline, and bedroom noise events — trends you can act on — not a letter grade.
Why "Good Score / Bad Night" Happens
The classic mismatch: high score, terrible sleep. Or low score, you feel fine.
This happens when the algorithm weighs signals differently than your body does. A few common failure modes:
The score rewards stillness, but you had restless sleep because of pain or discomfort. If you barely moved and your heart rate stayed low, some apps score that as "deep restorative sleep" even if you were awake half the night staring at the ceiling.
The score penalizes short sleep, but you're a natural short sleeper. Six hours feels great for you. The app deducts points because the American Academy of Sleep Medicine says adults need 7-9. You're docked for not being average.
The score ignores context. You slept 8 hours after a red-eye flight. The app says "Great!" Your body says "jet lag." The score doesn't know you crossed four time zones. For more on that gap, see jet lag and sleep data.
The algorithm changed. Your routine hasn't. Your data hasn't. But three weeks ago the company updated the weights, and now your baseline 80 is a 72. You're not sleeping worse. The formula shifted.
This is why orthosomnia — obsessive score-chasing — is a documented problem in the sleep-tracking literature. People start structuring their nights around getting a good score instead of getting good sleep. That's the opposite of wellness.
Better Lenses: Movement, Position, Noise
If a single score collapses too much, what should you look at instead?
Three concrete dimensions that reveal what changed without assigning a grade:
1. Movement frequency
How many times did you shift positions? How restless was the night compared to your baseline?
Healthy adults shift positions 20-40 times per night. A spike above your normal — say, 55 shifts when you usually log 28 — correlates with stress, alcohol, late caffeine, or early stages of illness. A drop below normal can mean deeper sleep, exhaustion catching up, or medication effects.
Movement frequency is a proxy for sleep continuity. It doesn't tell you whether you got enough REM or deep sleep (that requires EEG), but it shows you how settled your night was. Plot it over weeks and you'll see the travel nights, the argument nights, the nights the neighbor's dog barked until 2 a.m.
2. Position timeline
How much time did you spend on your back vs your side vs your stomach?
Position matters for snoring, acid reflux, and apnea severity. If you're a side sleeper trying to reduce reflux, you want to see "6.5 hours left side, 1 hour right side, 30 minutes back" — not a score that says "85 / Good Sleep" with no breakdown.
Position also reveals habit changes. If you're trying to stay on your side or stop sleeping on your stomach, the timeline shows whether the effort is working. A score can't tell you that. For more on position tracking without a camera, see how to track sleep position without a camera.
3. Bedroom noise events
How many times did a sound spike wake or disturb you?
Most sleep apps ignore audio entirely. A few (like Komori, coming 2027) classify noise events on-device and log their timing. When you see "3 noise events clustered between 2:15 and 3:00 a.m." and you also see a movement spike in that window, you have a hypothesis: something loud happened, and your sleep fragmented in response.
That's actionable. A score of 68 with no explanation is not. For a deeper look at bedroom sound, see noise and sleep: what decibels actually matter.
How to Review a Night Timeline Without Overfitting Meaning
Once you shift from score-watching to timeline-watching, a new risk emerges: overinterpreting single-night data.
One restless night doesn't mean you have a sleep disorder. One night of 9 hours doesn't mean you're suddenly a long sleeper. One night with 60 position shifts doesn't mean your mattress is broken. Variance is normal.
What matters is the pattern:
- Compare to your own baseline, not to population averages. If you usually move 25 times a night and last night was 50, that's worth noting. If you usually move 40 and last night was 50, that's noise.
- Look for clusters. Three bad nights in a row after switching pillows is a signal. One bad night after staying up late is not.
- Pair timeline data with subjective recall. Did you feel rested? Did you remember waking? How's your mood? The device logs what happened; you decide if it mattered.
- Track trends, not grades. A graph that shows your average movement spiking every Sunday (because you stay up late watching TV on Saturday) is more useful than a Sunday score of 58.
For more on what your nighttime movement patterns say about your sleep, including when to worry and when not to, see that dedicated post.
What Komori Intentionally Does Not Show
Komori Care, the contactless sleep monitor we're building (coming 2027), does not calculate a sleep score.
That's by design. Komori shows:
- Position timeline: back / side / stomach time, minute by minute.
- Movement frequency: how many shifts, when they clustered, how they compare to your baseline.
- Bed presence: when you got in, when you got out, how long you stayed.
- Noise events: classified on-device (fan, voice, door, unknown), logged by time. Raw audio is not retained unless explicitly requested.
- Bedroom environment: temperature, humidity, CO2 if integrated with external sensors.
No algorithm weights these into a single number. You get the timeline. You decide what matters.
Why? Because the job of a sleep monitor is to show you what happened, not to judge whether it was good or bad. A score implies there's one right answer for everyone. There isn't. Some people thrive on 6 hours. Some people need 9. Some people move 15 times a night and wake refreshed. Some move 40 and feel the same.
What helps is consistency: your patterns, over time, on your mattress, in your room. When something breaks the pattern, you see it. No grade required.
For a deeper explanation of this design philosophy, see why Komori has no sleep score.
When a Sleep Study Actually Makes Sense
Timeline data is not a replacement for a sleep study. If you have symptoms that suggest a medical sleep disorder — chronic daytime sleepiness, witnessed apneas, gasping awake, severe snoring, unrefreshing sleep despite adequate duration — you need polysomnography, not a bedside monitor.
Waitlist for a Night Review Without the Scoreboard
If you're tired of chasing a number that doesn't reflect how you actually slept, Komori is being built for you.
A Night Review instead of a score. Position, movement, environment, and events — the raw timeline of what happened, so you can see the pattern instead of the grade.
Komori is a general wellness product in development, coming 2027. It is not a medical device, not FDA-cleared, and not intended to diagnose, treat, prevent, or cure any condition. It is not for seizure detection or SUDEP.
Join the waitlist to get notified when we launch. Kickstarter is planned for Q1 2027. The Insider Pass ($49) is optional for early access — details at komoricare.com.
The Honest Bottom Line
Sleep scores sell subscriptions. They create a feedback loop: check app, see number, feel something, come back tomorrow. That's good for retention. It's bad for understanding.
If your goal is knowing what changed — whether the new pillow helped, whether side-sleeping reduced your reflux, whether the neighbor's renovation kept you up — a score won't tell you. A timeline will.
Movement frequency. Position distribution. Noise events. Room temperature. Those are the signals you can actually act on. A single composite number is just noise.
Stop grade-watching. Start pattern-watching.
Komori Care is a contactless sleep monitor in development — radar + ambient + on-device audio, no wearable, no camera. Kickstarter launching Q1 2027. Join the waitlist to get notified when we ship.
Keep reading
- Why Komori has no sleep score — the design decision explained in full.
- How often the average person moves at night — what movement frequency reveals about sleep quality.
- Orthosomnia: when sleep tracking makes you sleep worse — the clinical literature on score anxiety.
- How to read a sleep report without overinterpreting — what consumer data can and can't tell you.
- Sleep tracker accuracy claims: what to trust — the gap between marketing and clinical validation.
You can choose a position at lights-out. Knowing what you held until morning is the hard part.
Komori is a contactless monitor that logs which position you slept in, through blankets, with no camera and nothing to wear. Pre-launch — join the list and we'll tell you when it ships.
Keep reading
Want to see your sleep position data?
Join the free waitlist for launch updates. Or lock in founder pricing with the Insider Pass ($49, credited toward your purchase).