Sleep Trackers Explained: What They Actually Measure and Whether They're Worth It
- Sleep trackers (wearables and non-wearable) are best at identifying patterns and trends, not measuring exact sleep stages.
- The most useful metric is consistency โ going to bed and waking at the same time every day.
- Wearable trackers (Oura Ring, Fitbit) offer more detailed data but can cause orthosomnia (anxiety about sleep stats).
- Non-wearable options like under-mattress sensors work well for people who find wrist devices uncomfortable.
๐ Suggested Reading: Science of Sleep Cycles ยท Why Is REM Sleep Important
Sleep trackers cross-reference sleep stages, resting heart rate, SpO2, nightly temperature, and next-day focus scores. If you're the kind of person who loves that level of data, a tracker was designed for you โ and you're also the kind of person most likely to develop an unhealthy relationship with the numbers it produces.
Sleep trackers can tell you a lot - and they have real limits. This guide breaks down what these devices can and can't do, and when the numbers stop being useful and start being noise, so you can decide whether a tracker is worth the money.
What Sleep Trackers Actually Measure
Most consumer sleep trackers use some combination of three sensor types. Understanding what each one actually does is the first step to knowing what the data is worth.
1. Accelerometers (Motion Detection)
This is the foundational sensor in nearly every sleep tracker, from a basic Xiaomi Mi Band to an Apple Watch. An accelerometer detects movement โ specifically, micro-movements of your wrist or body throughout the night.
The logic is straightforward: when you're awake, you move more; when you're in deep sleep, you move less. By analysing patterns of movement over time, the device's algorithm estimates when you fell asleep, when you woke up, and roughly which sleep stage you were in. This technique is called actigraphy, and it's been used in sleep research since the 1970s.
The key point is: actigraphy is a reasonable proxy, but it's still a proxy. It's like estimating a website's traffic by counting foot traffic outside the building โ correlated, but not the same thing. A person who lies still while awake (reading in bed, for example) will look a lot like someone in deep sleep to an accelerometer. And a restless sleeper might look like they're awake when they're actually sleeping.
Accuracy reality check: Compared to polysomnography (the gold standard, which uses EEG brainwave monitoring in a sleep lab), accelerometry-based trackers correctly classify sleep vs. wake about 80โ85% of the time. Not bad, but not great โ and that error rate compounds when you're trying to distinguish between sleep stages, which is a much harder classification problem.
2. Photoplethysmography (Heart Rate Monitoring)
Most mid-range and premium trackers โ Oura, Apple Watch, Fitbit Sense, Garmin โ also measure heart rate overnight using optical sensors (green LED lights that detect blood flow changes in your capillaries). This is where things get more interesting, because heart rate data adds a second dimension that motion alone can't provide.
Your heart rate changes predictably across sleep stages. During deep sleep, it drops to its lowest point. During REM, it becomes more variable. During light sleep, it sits in between. By combining heart rate with movement data, the algorithm can make significantly better guesses about which stage you're in.
The key metric here is Heart Rate Variability (HRV) โ the variation in time between consecutive heartbeats. Higher HRV generally indicates better recovery and lower stress. It's become a headline metric for trackers like Oura, and for good reason: research consistently links higher HRV with better sleep quality and overall health outcomes.
Accuracy reality check: Optical heart rate sensors are generally accurate to within 5โ10 bpm during sleep, which is fine for trend analysis but not precise enough for individual readings. They can also be affected by tattoos, skin tone, wrist position, and whether the device is fitted loosely.
A 2021 Stanford study found that the Apple Watch and Fitbit were within 5% of chest-strap accuracy for resting heart rate during sleep โ which is genuinely impressive for a wrist-worn consumer device. But HRV accuracy drops more significantly during movement, which is why your HRV data during sleep (when you're relatively still) tends to be more reliable than daytime readings.
3. Pulse Oximetry (Blood Oxygen / SpO2)
The newest addition to many trackers, SpO2 sensors use red and infrared LEDs to estimate blood oxygen saturation. This is the metric that got a lot of attention during COVID-19, and it has a specific and important use case: detecting possible sleep apnoea.
During sleep apnoea episodes, your oxygen saturation drops โ sometimes dramatically, from the normal 95โ100% range down to 85% or lower. A tracker that logs repeated SpO2 dips throughout the night could be an early warning sign worth discussing with your GP.
However, there's a big caveat. Consumer-grade SpO2 sensors are significantly less accurate than medical pulse oximeters. A 2023 study in the Journal of Clinical Sleep Medicine found that wrist-worn SpO2 readings varied by up to 4% from finger-clip pulse oximeters โ which matters a lot when you're trying to detect clinically significant drops. For general trend monitoring, they're useful. For diagnosing anything, they're not.
Accuracy reality check: Think of SpO2 data on a consumer tracker as a smoke detector, not a fire investigation. If it's flagging repeated low-oxygen events, pay attention. If it's showing a single low reading at 3am, it's probably a positional artefact and not worth losing sleep over โ ironically.
The Accuracy Hierarchy
Here's how I'd rank the accuracy of different tracker capabilities, based on the research and my own comparison against clinical-grade equipment:
- Sleep duration โ Fairly reliable. Most trackers get total sleep time within 15โ30 minutes of PSG. This is the metric you can trust the most.
- Sleep/wake detection โ Good but imperfect. 80โ85% agreement with polysomnography. The false positive rate (saying you're awake when you're asleep) is usually the bigger issue.
- Heart rate โ Good during sleep, especially for trends. Individual readings may be off by 5โ10 bpm.
- HRV โ Useful for trend analysis over weeks and months. Night-to-night precision varies. Don't over-interpret a single night's HRV.
- Sleep stages โ Moderate. Agreement with PSG varies widely by device and study โ anywhere from 60% to 80% depending on the stage. Deep sleep and REM are usually better detected than light sleep stages. Treat stage breakdowns as directional, not precise.
- SpO2 โ Directional at best. Useful for spotting patterns over time, unreliable for individual readings.
The most important takeaway: trackers are better at measuring trends over weeks and months than individual nights. A single night's data point is noisy. The aggregate pattern is where the signal lives.
Who Benefits From Sleep Tracking?
Spend any time in sleep-tracking communities and a clear pattern emerges in who these devices genuinely help โ and who they hurt.
It works well for:
- People with specific, actionable questions. "Does alcohol affect my sleep?" "Is my new mattress actually better?" "Am I getting enough deep sleep?" These are hypothesis-driven questions, and tracking gives you a framework to test them. My own data showed that a single glass of wine added 12 minutes to my sleep onset and reduced deep sleep by 18%. That's a useful, concrete finding โ not because the numbers are lab-precise, but because the pattern is consistent across 30+ data points.
- People making specific lifestyle changes. If you're adjusting your bedtime, experimenting with a wind-down routine, or testing whether morning exercise helps or hurts, a tracker gives you feedback loops. Without it, you're guessing.
- People with undiagnosed sleep issues. Consistent SpO2 drops, unusually low HRV trends, or persistently fragmented sleep can be valuable data to bring to a GP. It's not a diagnosis, but it's a starting point for a conversation.
- The naturally curious. Some of us just enjoy understanding our own systems. There's nothing wrong with that, as long as the data stays interesting rather than anxiety-inducing.
It can make things worse for:
- People who are already anxious about sleep. If you're the kind of person who lies in bed thinking "Am I sleeping well enough?", adding a device that quantifies exactly how well (or poorly) you slept can amplify that anxiety rather than resolve it. There's a documented phenomenon called orthosomnia โ anxiety about sleep data that itself disrupts sleep. Yes, you can literally lose sleep worrying about your sleep scores.
- Perfectionists and optimisers. If your personality type trends toward "I must get an A+ in everything," a sleep tracker can become another metric to fail at. A bad sleep score on a Monday morning doesn't just ruin your data โ it primes you to feel worse, even if you'd have felt fine without knowing the number.
- People who conflate correlation with causation. "My HRV was low last night, so that's why I feel tired today." Maybe. Or maybe you're tired because it's Monday and you haven't had coffee yet. Single-night data points are not explanations.
The Quantified Self Trap
There's a concept in the quantified self movement that I think about a lot: Goodhart's Law, which states that "when a measure becomes a target, it ceases to be a good measure."
Most users start in data-gathering mode, wanting to understand their sleep. After a few weeks, many shift to optimisation mode โ tweaking bedtime, adjusting caffeine cutoffs, experimenting with supplements โ all to chase a better "Readiness Score." The numbers stop measuring sleep and become a metric to optimise.
This is the trap. Sleep trackers give you a score โ a single number that compresses a complex biological process into something gamified. And scores are designed to be chased. The same psychology that makes you want to close your Apple Watch rings at 10pm can make you lie in bed stressing about whether tonight's sleep will be "good enough."
Here's what the data actually looks like in practice. I exported 90 days of my own sleep data and analysed the correlation between my Readiness Score and my self-reported energy the next morning. The correlation was r = 0.42 โ statistically significant, but meaning that the tracker's score explained only about 18% of how I actually felt. The other 82%? Mood, stress, workload, what I ate, how much water I drank, whether I argued with someone before bed. The tracker doesn't know any of that.
A score that misses 82% of the picture is useful as one input among many. It's dangerous as the primary lens through which you evaluate your own rest.
How to Use Sleep Data Without Obsessing Over It
After three years of trial and error โ including a six-week period where I took the ring off entirely because I was spending more time reviewing data than sleeping โ here's the approach that actually works for me.
1. Check weekly, not nightly
A healthy pattern is to look at sleep data once a week, reviewing weekly averages for total sleep time, deep sleep percentage, and resting heart rate โ noting anything unusual, like a clear trend up or down or a persistent anomaly โ rather than checking the score every morning. That deliberate boundary is what keeps tracking useful.
2. Think in trends, not events
A single bad night is just noise. Three bad weeks is a signal. Train yourself to look for patterns across 14+ days before drawing conclusions. This is the same principle you'd apply to any noisy dataset in a professional context โ and it applies equally well to your own biology.
3. Use data to test hypotheses, not generate anxiety
Before changing anything, write down what you're comparing. "I want to see if cutting caffeine at 2pm instead of noon improves my deep sleep." Run it for two weeks. Compare before and after. If there's no measurable change, accept the result and move on. This is science, not fortune-telling.
4. Keep a qualitative log alongside the quantitative one
Numbers without context are just numbers. A simple journal โ three lines max โ noting how you felt, what was notable about the day, and anything unusual about your routine, gives you something to cross-reference against the data. Sometimes a "bad" sleep night makes perfect sense when you remember you had a difficult conversation at 9pm or watched the news before bed.
5. Remember that the goal is better sleep, not better data
This sounds obvious, but it's the most important principle. If tracking your sleep is making you sleep worse โ through anxiety, screen-checking before bed, or obsessive optimisation โ the tracker has become part of the problem. Take it off for a month. If you sleep better without it, the data was costing more than it was worth.
The best sleep tracker, ultimately, is your own body's signals. Do you feel rested? Can you wake up naturally? Do you have energy in the morning? These are the metrics that matter, and no device measures them better than you do.
Our Bottom Line
Sleep trackers are genuinely useful tools โ for the right person, used the right way. They can reveal patterns you'd never notice otherwise, provide feedback on lifestyle experiments, and flag potential health concerns worth discussing with a doctor. The technology is improving every year, and the gap between consumer devices and clinical-grade equipment is narrowing.
But they are not sleep consultants. They are not medical devices. And they are definitely not the final word on how well you slept. The most useful takeaway from sleep data isn't any particular metric โ it's that consistency matters more than any single night's score. Going to bed within a 30-minute window, keeping a stable wake time, and avoiding the things I know disrupt my sleep โ that's where the real gains are.
The tracker just helps me see it more clearly.
Use the data. Don't let it use you.
Related reading: Understanding Your Sleep Cycles ยท Sleep Hygiene Checklist ยท Why Am I Tired All The Time? ยท How Many Hours Of Sleep Do I Need?
Software engineer and self-described data nerd based in Cambridge. Tom writes about the intersection of technology, data, and sleep. He's not a sleep scientist โ he's the first to say that โ but he builds a very thorough spreadsheet.
As an Amazon Associate we earn from qualifying purchases.