Wearables, measured honestly
What your wearable actually measures, and what it quietly guesses
Your watch is genuinely good at one thing and genuinely bad at another, and almost nobody tells you which is which. We coach clients who sync this data to us every week, so we see what happens when people trust the wrong number. Here is what the validation research supports, metric by metric.
The accuracy gap, in one chart
Devices are not uniformly good or bad. They are excellent at the thing they measure directly and poor at the things they model from it.
Longer bar means a less trustworthy number
In Stanford testing, six of seven devices measured heart rate within 5 percent error. This is the one number most wearables genuinely earn your trust on.
Reasonable for tracking your own trend week to week. Not precise enough to compare against someone else's device, and it miscounts pushing a trolley or holding a handrail.
Averaged roughly 31 percent error across devices. Systematic reviews found no brand inside acceptable accuracy limits for energy expenditure.
Error jumped to about 52 percent. A wrist device cannot see the resistance you are pushing, so it is inferring the hardest variable from the least useful signal.
Figures from published validation research, including a systematic review of commercially available wearables in JMIR mHealth and uHealth and Stanford University device testing. Error rates vary by device, activity and individual, so treat these as the typical case rather than a guarantee for your specific watch.
Every metric, and whether to act on it
The practical question is never "is this accurate" in the abstract. It is "should I change what I do tomorrow because of this number".
| Metric | How it is measured | Verdict | Act on it? |
|---|---|---|---|
| Heart rate | Optical PPG, green light through the skin | Accurate at rest and during steady effort. Degrades during rapid intervals and heavy weights, when a chest strap is meaningfully better. | Trust it |
| Resting heart rate trend | Overnight PPG averaging | The single most useful long-term number on the device. Absolute value matters less than a sustained rise, which often precedes illness or accumulated fatigue. | Trust it |
| HRV | Beat-to-beat interval variation from PPG | Useful as your own baseline and trend. Almost meaningless compared against another person, and different brands compute it differently, so numbers do not transfer between devices. | Directional only |
| Sleep duration | Accelerometer plus heart rate | Broadly reasonable for total time asleep. Devices tend to mistake quiet wakefulness for sleep, so expect it to flatter you slightly. | Directional only |
| Sleep stages (deep, REM, light) | Inferred from movement and heart rate | Against clinical polysomnography, every consumer device tested showed bias toward scoring light sleep. Treat stage percentages as entertainment, not a diagnosis. | Do not act on it |
| Calories burned | Modelled from heart rate, movement and your profile | The least reliable headline number on any wearable. Never eat back calories a device claims you burned, which is the most common way this data actively harms fat loss. | Do not act on it |
| SpO2 / blood oxygen | Red and infrared PPG | Consumer implementations are wellness features, not medical devices. Useful for spotting a pattern worth raising with a doctor, not for self-diagnosis. | Do not act on it |
Watch, ring or strap: why they disagree
The disagreement is not brand quality. It is physics: where the sensor sits and what it is physically measuring.
Wrist, screen
Smartwatch
Optical PPG on the wrist plus an accelerometer. The screen gives you live feedback mid-set, which no other form factor does.
Strong at: Live workout data, steps, GPS pace, everyday convenience
Weak at: Heart rate during heavy lifting and fast intervals, when wrist flexion and movement disrupt the optical signal
Finger, screenless
Smart ring
The same PPG sensor class, but the finger has richer blood flow and moves less in sleep. Oura Ring 4 and the Bengaluru-built Ultrahuman rings lead this category in India.
Strong at: Overnight metrics, resting heart rate, HRV trend, all-day comfort
Weak at: No live feedback during training, and gym bars or heavy grips can damage or mark the ring
Chest, screenless
Chest strap
Measures the heart's electrical signal directly, the same principle as an ECG, rather than inferring beats from light.
Strong at: The most accurate heart rate available to consumers, especially in intervals
Weak at: Nobody wears one to sleep, so it gives you nothing on recovery or overnight trend
The screenless shift
The fastest-growing category has no screen at all. Smart rings from Oura and the Bengaluru-built Ultrahuman, and upper-arm bands from Whoop, deliberately remove notifications so the device measures you without interrupting you. That trade is the whole proposition: you give up live feedback during a set and gain a device you will genuinely wear to bed, which is where recovery data actually comes from. Pricing models differ sharply too, and in India the subscription question changes the real cost substantially over a few years.
What we actually look at when a client syncs their data
- 1
Resting heart rate trend
A sustained rise over several days usually means poor sleep, illness or too much training. It is the earliest warning the device gives you.
- 2
HRV direction, not value
We never compare your HRV to anyone else's. We watch whether your own baseline is drifting down while training load climbs.
- 3
Total sleep time
Not the stage breakdown. Hours asleep is the number that predicts whether the next session goes well.
- 4
Steps on non-training days
Movement outside the gym is where most of the daily calorie burn actually lives, and it collapses quietly on busy weeks.
Notice what is missing: calories burned, and sleep stage percentages. We ignore both, deliberately.
Go deeper on the metrics
Common questions
Are fitness trackers accurate?
It depends entirely on the metric. Heart rate is genuinely accurate, with six of seven devices in Stanford testing measuring within 5 percent error. Calorie burn is not, averaging roughly 31 percent error for walking and running and about 52 percent for cycling, with systematic reviews finding no brand inside acceptable limits. Treating one number as reliable and the other as a rough guide is the correct approach.
Should I trust my watch's calorie count?
No, and specifically you should never eat back the calories it claims you burned. Energy expenditure is the least accurate headline figure on any wearable. Use it to compare today against your own yesterday, never as a budget to spend food against.
Is a smart ring better than a smartwatch?
They answer different questions. A ring sits on a finger with better blood flow and less movement overnight, so it is stronger for sleep, resting heart rate and HRV trend. A watch has a screen, so it is the only option that shows heart rate live during a set. Many people who train seriously use a chest strap for sessions and a ring or watch for the other 23 hours.
Why does my watch say I get almost no deep sleep?
Because sleep staging from a wrist is genuinely hard. In a multicentre study validating 11 consumer trackers against clinical polysomnography, every device showed a bias toward scoring light sleep. Total sleep time is worth watching; the deep and REM percentages are not worth losing sleep over.
Do wearables work the same on Indian skin tones?
This is under-discussed and worth knowing. Optical sensors use green light, which melanin absorbs, and several studies report reduced signal quality on darker skin, sometimes combined with higher body fat. The evidence is not unanimous, with at least one large study finding activity type matters more than skin tone. If your device gives erratic readings, a chest strap removes the variable entirely because it measures electrical signal rather than light.
Want someone reading this data with you?
Our coaching syncs your wearable data and tells you which numbers to act on and which to ignore, every week.
Start Your JourneySources
- Reliability and Validity of Commercially Available Wearable Devices for Measuring Steps, Energy Expenditure, and Heart Rate: Systematic Review · JMIR mHealth and uHealth
- Accuracy of 11 Wearable, Nearable, and Airable Consumer Sleep Trackers: Prospective Multicenter Validation Study · JMIR mHealth and uHealth
- Monte Carlo analysis of optical heart rate sensors: the effect of skin tone and obesity on the PPG signal · PMC
- Accuracy and Acceptability of Wrist-Wearable Activity-Tracking Devices: Systematic Review · Journal of Medical Internet Research
Device prices and subscription terms change frequently and vary by retailer. Coaching guidance on this site is educational and is not a substitute for medical advice. See how we use sources.