Your Smartwatch Is Lying to You (About Some Things)
What your wearable actually measures well, what it guesses at, and what it completely makes up — based on 6 validation studies across 300+ participants.
Conviction: Moderate (metric-dependent)Tonight, delete your workout calorie count from your food diary. Stop eating back "burned" calories. Your watch overestimates by 20-50%, and that's the average — it could be worse.
Think of your smartwatch like a weather station bolted to your wrist. The thermometer (heart rate sensor) is genuinely accurate — it measures one thing directly and does it well. But the "feels like" temperature (calorie burn) is a guess based on formulas and assumptions about your body — and those assumptions are wrong 20-50% of the time. The sleep stage breakdown is like predicting tomorrow's weather hour by hour: the forecast that it will rain (you're asleep) is reliable, but whether it'll drizzle at 2am vs pour at 3am (light sleep vs deep sleep) is essentially a coin flip.
What your wearable actually measures well, what it guesses at, and what it completely makes up — based on 6 validation studies across 300+ participants.
Conviction: Moderate (metric-dependent)Delete your workout calorie count from your food diary. Stop eating back "burned" calories based on your watch.
Your wearable overestimates calorie burn by 20-50% on average. That "500 calorie" workout might have burned 250-400. Eating those calories back undoes your deficit.
Takes 10 seconds. Open your food app. Delete the exercise entry.
The Verdict
Your watch knows your heart rate. It guesses your calories. It invents your sleep stages.
Think of your smartwatch like a weather station bolted to your wrist. The thermometer (heart rate sensor) is genuinely accurate -- it measures one thing directly and does it well. But the "feels like" temperature (calorie burn) is a guess built on formulas and assumptions about your body -- and those assumptions are wrong 20-50% of the time. The sleep stage breakdown is like predicting tomorrow's weather hour by hour: the overall forecast that it will rain (you're asleep) is reliable, but whether it drizzles at 2am or pours at 3am (light sleep vs deep sleep) is essentially a coin flip.
Want the full evidence? Keep scrolling
What Most People Think
Most people treat their Apple Watch, Garmin, or WHOOP as a miniature medical lab strapped to their wrist. The calorie burn number after a workout feels like a biological fact. The sleep stage breakdown -- 1 hour 23 minutes of deep sleep -- reads like a lab report.
The stress score feels like a direct measurement of their nervous system. Marketing and gamified interfaces reinforce this: the numbers look precise, so they must be accurate.
Here's what's really happening: some of those numbers are genuinely useful. Others are expensive guesses dressed up as measurements. And knowing which is which changes how you should use your device.
What the Evidence Actually Shows
At rest and during steady-state exercise (walking, cycling), wrist-worn devices achieve over 95% agreement with clinical ECG. Errors stay below 5%. Apple Watch consistently outperforms competitors.
This is the one metric your watch genuinely nails. The optical sensor reads blood flow changes through your skin, and when you're sitting still or walking steadily, there's very little noise to confuse it.
Shcherbina et al., 2017, N=60 | Alvarez-Garcia et al., 2024, N=50
No consumer wearable has achieved acceptable error (below 10%) for calorie burn in any validated study. Errors range from 11% to 93% depending on device and activity type. Most land in the 20-30% range.
Think about that: your watch says you burned 500 calories, but the real number could be anywhere from 250 to 450. Devices systematically overestimate during walking and running, and underestimate during resistance training and non-standard activities.
The reason is simple. Heart rate is a direct measurement. Calorie burn is a calculation -- it takes your heart rate, plugs it into a formula with your age, weight, and sex, and spits out an estimate. The formula was built on average bodies. If you're not average (and nobody is), the estimate drifts.
Shcherbina et al., 2017 | Wang et al., 2022, N=20
All major devices achieve 95%+ sensitivity for detecting when you're asleep. Your watch knows you're sleeping. That part works.
The catch: devices frequently misclassify quiet wakefulness as light sleep. If you're lying still in bed but awake (scrolling your phone with minimal movement), your watch thinks you're in light sleep. This means total sleep time is slightly overestimated.
Chinoy et al., 2022, N=35
Apple Watch underestimates deep sleep by up to 43 minutes in a single night compared to a real sleep study. Fitbit Sense detects REM sleep with only 60% sensitivity -- it frequently calls REM "light sleep" instead.
Here's why this matters: your watch knows you're asleep, but it's essentially guessing what kind of sleep you're in. It's like knowing someone is inside a building but guessing which room they're in by listening through the wall. The stage-by-stage breakdown on your morning summary is entertainment, not measurement.
Chinoy et al., 2022 — Apple Watch 8, Oura Ring Gen3, Fitbit Sense 2 vs polysomnography
Overnight heart rate variability (HRV) from wearables shows strong agreement with medical-grade ECG. For tracking your nervous system's balance, recovery status, and long-term stress trends, the data is reliable enough to guide decisions.
This is the sleeper metric. Most people obsess over their sleep stages (unreliable) and ignore their HRV trends (reliable). If your overnight HRV has been declining for two weeks, that's a genuine signal that your body is under-recovered -- whether from training, stress, poor sleep, or illness.
Kim et al., 2025 | Jimenez-Ocana et al., 2026
The Debate
Side A — 4/10 studies in systematic review
Melanin absorbs green LED light, reducing heart rate accuracy during moderate-to-vigorous exercise. Error rates are significantly higher for darker skin tones on the Fitzpatrick scale.
Side B — 4/10 studies in same review
Studies using Garmin and Fitbit models found no statistically significant interaction between skin tone and heart rate accuracy during walking and jogging protocols.
Verdict: The discrepancy is likely hardware generational -- newer devices dynamically increase LED intensity when detecting poor signal quality. Both findings are probably correct for their respective device generations. The bias is real but shrinking with each hardware release.
Koerber et al., 2022 — ACC systematic review, N=469 across 10 studies
Honest Limitations
Treat published accuracy as a best-case snapshot, not a permanent guarantee.
The population that needs accurate calorie data most gets the least accurate readings.
If you lift weights, your workout calorie count is likely even less accurate than the 20-50% error range suggests.
The Practical Takeaway
Resting heart rate trends, overnight HRV trends, total sleep duration, and the basic "asleep vs awake" detection. These are clinically useful for long-term lifestyle monitoring and recovery tracking.
Workout calorie counts for nutritional decisions. Never eat back "burned" calories based on wearable data. A 20-50% overestimate means your 500 calorie burn might really be 250-400 calories. Build your nutrition plan on intake tracking and weekly weight trends instead.
Sleep staging as directional trends only -- is deep sleep going up or down over weeks? That trend might mean something. But a single night's stage breakdown is noise. Don't make decisions based on one night's data.
Apple Watch and Samsung detect moderate-to-severe sleep apnoea with about 92% accuracy -- useful as a screening flag to get checked, not a replacement for a clinical sleep study. They systematically miss mild cases.
Obsessing over inaccurate sleep data can actually worsen your sleep. If your device says you got 45 minutes of deep sleep but you feel rested and alert, trust your body. The device is probably wrong about the stages. You're fine.
The Nuance
The accuracy hierarchy is device-specific. Apple Watch consistently outperforms competitors in both heart rate and calorie accuracy, though no device passes the clinical bar for energy expenditure. If accuracy matters to you, the device you choose matters too.
Tattoos can completely block the green LED light that optical sensors use to read your pulse. If you have wrist tattoos, heart rate and HRV data from that wrist is essentially useless. Wear the device on the non-tattooed wrist.
Wearables may actually be useful for detecting sleep apnoea -- achieving 92% sensitivity and 93% specificity for moderate-to-severe cases. But they systematically underestimate mild cases. A clean result from your watch doesn't rule out mild sleep apnoea if you're still snoring, waking tired, or getting reports of breathing pauses from a partner.
Sources
Conviction by metric:
Heart rate & resting HRV HIGH
Sleep/wake detection HIGH
Sleep staging LOW
Energy expenditure LOW
What would change this: Open-source validation of algorithms across diverse populations (N>10,000) using doubly labelled water for calorie measurement and longitudinal home EEG for sleep staging. Integration of transdermal metabolite sensors (lactate, glucose) rather than exclusive reliance on optical heart rate and accelerometry.
Want coaching that uses real data instead of wearable guesses? SLH Fit builds your nutrition from intake tracking and weekly weight trends -- the only reliable approach.
Produced by SLH Fit · Truth Engine · Not medical advice.
How strong is the evidence for the claims in this review? Higher = more confidence the claims are supported. This does not measure how large the effect is or how important it is compared with other levers.
Conviction-scored health research in your inbox. What works, what doesn't, and what the studies actually measured.
Subscribe freeConviction-scored verdicts on supplements, nutrition, training, physio, and recovery.