Smartwatches and fitness trackers have exploded in popularity in recent years. With the ability to count your steps, monitor heart rate, and track sleep, these devices promise to quantify your health and wellness like never before. But how accurate are they really?
I put several popular smartwatches – including Apple Watch, Fitbit, and Xiaomi Mi Band – through a series of controlled experiments and data analysis to find out if you can trust the numbers coming off your wrist. Here’s what I discovered:
Step Counting: A Tricky Balance
Step counting is arguably the most basic smartwatch feature, yet it proves surprisingly challenging to perfect. In my testing, all the watches counted steps reasonably well during normal walking. But performance suffered in certain scenarios:
- Carrying objects – Holding a glass of water caused Apple Watch and Mi Band to miss all steps entirely. Fitbit and my phone were closer to the true count.
- Erratic motions – Vigorous arm waving caused Fitbit and Mi Band to overcount substantially. Apple Watch was most accurate.
- Driving – 10 minutes of driving caused Fitbit to tally nearly 1,000 false steps! Apple Watch was far less susceptible to these non-step vibrations.
The key point is that devices must balance sensitivity to true steps with rejection of non-step motions. Apple Watch leans conservative, missing some real steps but also minimizing false counts. Fitbit opts for more sensitivity, capturing more steps but prone to overestimating. Each approach has merits depending on your use.
Heart Rate Tracking: Mostly Solid, With Some Quirks
I evaluated heart rate tracking in three scenarios against a medical-grade chest strap for reference:
During Exercise
All three watches followed my heart rate reasonably well during a varied training run, with Apple Watch performing best at capturing rapid changes:
- Apple Watch missed the peak heart rate by just 2 bpm (186 vs 188 bpm)
- Fitbit missed the peak entirely, recording a lower rate than earlier in the run
- Mi Band was late to respond to heart rate spikes
All-Day Wear
The watches were fairly accurate during daytime non-exercise use, although not without some oddities:
- Apple Watch took very infrequent measurements, missing some short spikes
- Fitbit provided more continuous monitoring with decent accuracy
- Mi Band had the poorest performance with many erroneous high/low readings
Overnight
Heart rate tracking was most reliable overnight while I slept:
- Apple Watch and Fitbit both had average errors under 2 bpm
- Mi Band was slightly higher at 3.4 bpm average error
So optical heart rate sensors are far from perfect, but seem to provide good enough tracking for most purposes if you consider their limitations. Chest straps remain superior for serious training.
Sleep Tracking: Decent for Duration, Unreliable for Stages
Determining sleep duration relies on movement and heart rate patterns. Watch-estimated total sleep times were generally within 30 minutes of my recorded sleep/wake times. However, they can be fooled:
- Lying still while awake would sometimes be mistaken for sleep
- Failing to sync after waking up caused overestimates
Sleep stage detection is more challenging without detailed physiological data. Recent models are improving, but accuracy remains mediocre:
- Apple and Fitbit may miss or misclassify sleep stages
- Light sleep is most often correctly identified
- REM and deep sleep are least accurately detected
For now, focus on the total sleep duration your watch provides as a useful indicator, but take sleep stages with a grain of salt.
The Verdict: Good Enough for Wellness Tracking
While smartwatches have limitations, I believe their readings are accurate enough for general health and wellness tracking, with a few caveats:
- Step counts are reasonable estimates but can vary significantly depending on your motion patterns and watch brand.
- Heart rate is mostly reliable except for intense interval training. Overnight measurements are most accurate.
- Sleep duration estimates are usually within 30 minutes if you sync right after waking. Sleep stages are iffy.
The biggest value is in tracking trends over time within each device’s quirks. If your daily step count, workout frequency, or sleep duration declines week-over-week, it’s usually indicative of a real change even if the absolute numbers aren’t perfect.
Additional standardized accuracy testing across more watch models would be beneficial. But for most users, today’s smartwatches provide measurements that are useful representations of your activity and sleep. Just don’t expect laboratory precision!