I wore a sleep tracker for about a year and a half. For the first few months I checked it every morning before I'd even sat up in bed, and I can tell you with confidence that the score changed how I felt about the day far more than the sleep did.
That's not an argument against trackers. It's an argument for understanding what they're actually measuring, so you know which numbers to trust and which to ignore. Because some of them are quite good, some of them are quite bad, and the app shows all of them with the same confident decimal point.
What a Wearable Actually Measures
Sleep, as sleep scientists define it, is a thing that happens in the brain. The stages — light, deep, REM — are classified by patterns of brain activity, along with eye movement and muscle tone. Measuring that properly means electrodes on the scalp in a sleep lab, a method called polysomnography. It's the reference standard because it's looking at the thing itself.
A wristband is not looking at the thing itself. It's looking at proxies: how much your arm moves, how fast your heart is beating, how that heart rate varies from beat to beat, and sometimes skin temperature or blood oxygen. From those signals, an algorithm infers what your brain is probably doing.
That inference isn't nonsense. Deep sleep does tend to come with a slow, steady heart rate and very little movement. REM tends to come with a more variable heart rate and near-total stillness because your muscles are switched off. But "tends to" is doing a lot of work in those sentences, and the algorithm is guessing at brain states from the wrist, through several layers of assumption.
What Trackers Get Right
The well-established finding, across many comparisons of consumer devices against lab measurement, is that wearables are decent at total sleep time and poor at staging. That's the sentence to remember.
Total sleep time is the easy problem. Being awake and being asleep look very different on a movement sensor and a heart-rate sensor. The device can tell roughly when you stopped moving and your heart rate settled, and roughly when both picked up again in the morning. It will typically be off by some minutes, and it tends to count still-but-awake time as sleep, but as an estimate of "how long did I sleep last night," it's usable.
The same goes for the two timestamps that matter most: when you fell asleep and when you woke up. These are exactly the numbers you'd want if you were trying to fix a bedtime, and they're the numbers the tracker is most reliable at. That's a lucky alignment, and it's the basis for the sensible way to use one, which I'll get to.
What Trackers Get Wrong
Staging is the hard problem, and the wrist is a bad place to solve it.
When researchers compare a wearable's stage-by-stage readout against the lab, the agreement is a lot weaker than for total time. The device will confidently tell you that you got 1 hour 12 minutes of deep sleep, and the honest error bar on that number is wide enough that it could reasonably have been 40 minutes or an hour and forty. Different devices worn on the same night by the same person will disagree with each other, and the same device will produce inconsistent results depending on how tightly it's strapped.
This matters because the stages are where the app puts all its narrative weight. The score, the "restorative sleep" percentage, the little bar chart of your night — most of it rests on the part of the measurement that's least trustworthy. If you want to know what deep sleep and REM actually do and how much you plausibly need, I've covered deep sleep and REM separately. The short version: they matter, and your tracker can't reliably tell you how much of them you got.
So when the app says your deep sleep was low last night, the correct reaction is not concern. It's "maybe, maybe not."
Orthosomnia: When the Score Becomes the Problem
Sleep clinicians have a name for what happened to me in those first few months. They call it orthosomnia — a preoccupation with sleep-tracker data serious enough that it interferes with sleep.
It goes like this. You get a bad score. You worry about the score. The worry makes tonight's sleep worse. Tomorrow's score is also bad, which confirms that something is wrong with your sleep, which is now partially true, because you've spent two nights anxious about it. Some people in this loop report feeling fine in the morning and then feeling tired once they see the number. The app is doing the opposite of its job.
The reason this happens is that a nightly grade invites a nightly judgment. And a judgment based on a measurement that's mostly guessing is a bad foundation for how you feel about your own body. It's a little like weighing yourself on a scale that's accurate to within four kilos and then agonizing over a 300-gram change.
If you already wake up with a knot of anxiety, the last thing you need is a verdict waiting on your wrist before you've had a glass of water. For some people the honest fix is to keep the tracker and stop looking at the score. For others it's to take the thing off for a month and notice whether they sleep better without it. I did the second one, and I slept better.
The Smart-Alarm Promise
There's one tracker feature that deserves a specific mention because it sits right on top of the staging problem: the smart alarm that wakes you "at the optimal moment in light sleep."
The pitch is appealing. Set a thirty-minute window, and the device watches your stages and picks the moment you're sleeping lightest, so you wake up feeling refreshed instead of groggy. And there's a real idea underneath it, because being pulled out of deep sleep genuinely does feel worse than waking from light sleep.
The problem is the execution. To wake you in light sleep, the device has to know your stage right now, in real time, from a wrist sensor. That's the exact thing it's worst at. So what you actually get is an alarm that goes off somewhere in a half-hour window based on a guess, and on the mornings the guess is right, it feels like magic, and on the mornings it isn't, it's just a random alarm that went off early. I've written a longer breakdown of sleep cycle alarm clocks and a separate look at whether the 90-minute rule holds up, and the conclusion in both is the same: the theory is fine, the sensor can't deliver it, and the variable wake time it introduces is itself a cost.
Because a wake-up window means your wake time drifts. And a drifting wake time is the thing most reliably associated with feeling worse in the morning, regardless of what stage you happened to wake from.
How I'd Actually Use One
After all that, I still think a tracker can be useful. It's a matter of asking it the questions it can answer.
Look at trends, not nights. A single night's data is noisy and partly guessed. A four-week average of bedtime, wake time, and total sleep is real information, because the errors wash out and the trend survives. If your average bedtime has crept forty minutes later over a month, that's something you can act on.
Use it as a consistency check. The most useful thing my tracker ever showed me wasn't a score. It was a scatter of wake times across a month — a tight cluster on weekdays, and two dots floating three hours to the right every weekend. That picture explained my Mondays better than any deep-sleep percentage could have.
Ignore the staging and the score. Or at least, don't let them decide how you feel. If you feel rested and the app disagrees, you're right and it's wrong. It's a sensor on your arm, not an authority on your brain.
Stop if it makes you worse. This is the orthosomnia test. If checking the data makes you calmer and more consistent, it's earning its place. If it makes you anxious, it isn't, and there's no prize for persevering.
What Actually Matters More Than the Score
Here's the thing I eventually landed on. The two behaviors that most reliably improved how I felt in the morning had nothing to do with measuring my sleep and everything to do with deciding it.
The first is a fixed wake time that I honor every day, weekends included. Not a window, not a smart range — a time. It's the anchor for everything else, and it's the single most effective variable in fixing a sleep schedule. It doesn't need a sensor to work. It needs an alarm that actually gets me out of bed, and I use Captain Wake for that because it won't stop until I've done something physical — a photo of the bathroom mirror, a barcode on the cereal box — which is a very effective cure for the "I'll get up at the next one" negotiation.
The second is looking at the night before it happens rather than grading it afterward. The same app's sleep forecast shows what I'm on track for if I go to bed now, given tomorrow's alarm, and that number has changed my behavior far more than any morning score ever did, for a simple reason: I can still do something about it. I've written about why a forecast beats a report at length, but the core is that a report tells you what you lost, and a forecast tells you what you can still keep.
A tracker measures what happened. That's fine, as far as it goes. But sleep isn't mostly a measurement problem. It's a decision problem, and the decisions get made at 11 PM and 6:30 AM, not in the app the next morning.
