If you have ever woken up to a 73 from your watch and felt cheated, you are not alone. Sleep scores are everywhere. They are also, in a meaningful sense, made up. This piece is an honest tour of what a sleep wearable can actually see, what it cannot, and how to use one well anyway.
The sleep score lie
Every modern sleep wearable gives you a single 0 to 100 number when you wake up. Apple Watch does it. Fitbit does it. Withings does it. Garmin, Oura, Whoop, all of them.
That number does not exist in clinical sleep medicine. Polysomnography, the overnight study sleep doctors run in a lab, does not produce a sleep score. It produces a hypnogram (a graph of sleep stages across the night), measurements of breathing events, oxygen desaturations, leg movements, and arousals. A trained technician scores it. The output is a report, not a vibe-check on a 100-point scale.
The 0 to 100 score on your wrist was invented by product designers because a single number is easier to engage with than a hypnogram. That does not make it useless. It makes it a marketing convenience that is sometimes correlated with how you actually slept. Treat it accordingly.
What polysomnography measures (and your wearable cannot)
A real sleep study, a polysomnogram or PSG, simultaneously records:
- EEG — brain wave activity through electrodes on your scalp. This is how sleep stages are actually identified.
- EOG — eye movement, which distinguishes REM sleep from other stages.
- EMG — muscle tone at the chin and legs, which drops in REM and spikes during arousals.
- Respiratory effort and airflow — to catch apneas and hypopneas.
- Pulse oximetry — blood oxygen saturation.
- ECG — heart rhythm.
Your wearable, sitting on your wrist or under your mattress, measures a much smaller set: pulse rate, pulse rate variability, body movement, and on better devices, blood oxygen and sometimes single-lead ECG. From those signals it guesses, using a proprietary algorithm, which stage of sleep you were probably in during each 30-second window.
Independent validation studies tell a consistent story. Consumer wearables typically achieve 60 to 75 percent agreement with PSG on stage classification. They are reasonably good at detecting total sleep time and time awake. They are worse at distinguishing light sleep from deep sleep, and worse still at picking out REM in people with unusual physiology. Marketing materials will quote much higher numbers when they cherry-pick the easiest metric. The honest range is 60 to 75 percent.
So why use one at all?
Because polysomnography happens once. Maybe never. Your wearable runs every night for years.
A single PSG is the gold standard for diagnosing a sleep disorder. It is a poor instrument for understanding your sleep across seasons, training cycles, alcohol-free weeks, work stress, jet lag, and the rest of your life. The wearable is bad at any one night and good at thousands.
This is the framing to keep: the absolute numbers are noisy, but the trend is real. If your wearable says you got 18 percent deep sleep last night, treat that as a rough estimate. If it says your 30-day average deep sleep dropped from 18 percent to 11 percent after you started drinking again, take that seriously.
The three sleep devices worth considering in our catalog
We carry a lot of products. For sleep specifically, three earn their place. Different tools for different jobs.
1. Wrist tracker that also handles cardiac health
The ScanWatch 2 is the wearable we point most people toward when they want one device that handles both daily wear and overnight tracking. It is FDA 510(k) cleared for AFib detection, which means it has been reviewed by regulators rather than just marketed at them. Battery lasts about 30 days, which matters more than it sounds, because the wearable you charge every night is the wearable that misses sleep data.
If you want a lighter, less expensive entry point in the same family, the ScanWatch Light drops the ECG and scores around 70 Standard in our system. Good sleep tracking, no cardiac clearance.
2. Non-contact tracker for the best accuracy
The Withings Sleep Tracking Mat is the highest-scoring sleep device in our catalog and the one we would put in our own bed. It is a thin pad that slides under the mattress, plugs into the wall, and records heart rate, respiration rate, movement, snoring, and sleep stages without you ever touching it. There is nothing to wear, nothing to charge, and nothing to forget.
The non-contact form factor matters for accuracy in a way that is easy to miss. A wrist tracker has to filter out arm movement, transient pulse readings, and tattoo or hair interference. An under-mattress sensor sees ballistocardiography, the small mechanical signal of your heartbeat conducted through the mattress, which is a cleaner signal than wrist photoplethysmography. Withings has published k-coefficient validation against PSG for the mat, which is more methodological honesty than most of the category provides.
The trade-off is that the mat does not travel and it does not measure HRV during the day. If you only ever want one sleep device, this is it.
3. Wrist tracker for people already in the Fitbit world
The Fitbit Sense 2 and the original Fitbit Sense both belong in this conversation if you already use Fitbit. The original Sense was the first Fitbit with an FDA-cleared ECG and EDA stress sensor, and it still scores 80 Verified in our system. The Sense 2 refines that platform; we list it refurbished at 79 Verified.
Fitbit's stage classification algorithm is mature and Fitbit publishes more about its validation methodology than most competitors. If you already have a Fitbit account with years of trend data, do not throw that data away to switch ecosystems. The trend is the point.
What the numbers actually mean
If you only look at one number, look at total sleep time. If you look at five, look at these. None of them are precise on any given night, but their averages over weeks are meaningful.
- Deep sleep percentage — Healthy adults typically spend 13 to 23 percent of total sleep in slow-wave (deep) sleep. This is the stage associated with memory consolidation, growth hormone release, and immune function. Alcohol and late-evening exercise consistently suppress it.
- REM sleep percentage — Healthy adults typically spend 20 to 25 percent of total sleep in REM. REM is tied to emotional processing and procedural memory. It is heavily concentrated in the last third of the night, so cutting your night short cuts REM disproportionately.
- Overnight HRV — Heart rate variability during sleep is the cleanest objective stress marker most wearables can give you. Your absolute HRV number is not comparable to anyone else's. Your own trend is. A sustained drop usually means something: training overload, illness coming on, alcohol, or psychological stress.
- Sleep latency — Time from lights out to actually asleep. Under 20 minutes is normal. Under 5 minutes consistently can suggest sleep deprivation, not enviable sleep skill.
- Wake after sleep onset (WASO) — Total time awake during the night after you first fell asleep. Under 30 minutes is considered healthy. Rising WASO across weeks is often the earliest signal that something has changed.
A 30-day protocol that actually tells you something
Most people put a wearable on and then drown in dashboards. Try this instead.
- Days 1 to 7: baseline. Wear the device every night. Do not change anything. Record nothing in your head except a vague note of how you feel. At the end of the week, write down your seven-day averages for total sleep time, deep sleep percentage, REM percentage, overnight HRV, and WASO.
- Days 8 to 30: change exactly one variable. Pick one. Caffeine cutoff at 12pm. Or no screens after 9pm. Or bedroom at 65 degrees Fahrenheit. Or zero alcohol. One variable. For three full weeks.
- Day 30: compare averages. Look at the same five metrics over the 23-day intervention window versus the 7-day baseline. If the averages moved meaningfully and consistently, that variable matters for you. If they did not, it probably does not, and you should try a different one for the next cycle.
This is the only way to extract a real signal from noisy nightly data. One variable, three weeks, compare averages. Do not try to optimize seven things at once; you will not be able to tell what is working.
Sleep environment beats sleep tracker
Here is the part that no one selling sleep wearables wants to say out loud. For most people most of the time, a dark, cool, quiet bedroom does more for sleep quality than any tracker ever will. The wearable tells you what happened. The environment determines what happens.
Blackout curtains. Room temperature in the 60 to 67 degree Fahrenheit range. White or pink noise to mask intermittent sounds. A consistent wake time. None of that requires a Bluetooth connection.
The Hatch Restore is a Standard product in our catalog, not Verified, because it does not have the clinical-grade clearances some other devices do. That is the right tier. It is not pretending to be a medical device. It is a well-built environmental tool for the bedroom, and for a lot of beginners it will move sleep quality more than upgrading your wrist tracker would.
What is not in the catalog yet
To be transparent about gaps: there are good sleep devices we do not currently carry. If you are shopping broadly, we would expect the following to score in the Verified range when we get to them.
- Oura Ring (Gen 3 or Gen 4) — Finger ring form factor with strong published validation work on sleep staging and HRV. Expected Verified-plus when added.
- Whoop 4.0 — Subscription-based wrist band with detailed strain and recovery analytics. Strong on HRV trending; weaker on consumer health credentials.
- Eight Sleep Pod — A heating and cooling smart mattress cover that actively manipulates sleep environment rather than just measuring it. Expensive, well-engineered, and the only device on this list that changes what your bed does.
The honest bottom line
A sleep wearable is a long-running, imperfect instrument that becomes a powerful one when you stop treating each night as a verdict and start treating monthly averages as the unit of analysis. The single 0 to 100 score on your wrist is not real. The HRV trend across 90 nights is.
If you want maximum accuracy and you never want to think about it, the Withings Sleep Tracking Mat is the cleanest answer. If you want one device that handles sleep, HRV, and cardiac screening on your wrist, the ScanWatch 2 is the one we would recommend. If you already live in Fitbit, stay there. And before you spend a dollar on any of them, take a look at your bedroom.