Step counters aren't magic — they're accelerometer pattern-matchers, and they're right most of the time but wrong in predictable ways. Knowing where your tracker over- and under-counts helps you trust the numbers (or not) for the things that actually matter.
How accurate are different devices?
iPhone (HealthKit): ±5% in normal walking, worse on slow walking and stair climbing. Apple Watch: ±3%, the most accurate consumer tracker for general adults. Android phones via Health Connect: ±5–10% depending on phone manufacturer; Samsung devices tend to undercount. Garmin / Fitbit: ±5%, calibrated well over years. Cheap fitness bands ($20–40): ±15–25%, often optimistic.
When trackers undercount
Slow walking under 3 km/h. Treadmill walking with handrails. Pushing a stroller / shopping cart (arm doesn't swing). Hands in pockets. Phone in a bag rather than on your body. Walking with a leashed dog (frequent stops/turns confuse the algorithm).
When trackers overcount
Driving on bumpy roads (rare but happens). Vigorous fidgeting at desk. Riding a bicycle with the watch on the handlebar-side hand. Power tools, hammering, repetitive arm work that mimics gait pattern.
How to test your tracker
Walk a known distance (track lap, mapped sidewalk) at normal pace. Count manually as a check. Repeat once at brisk pace and once slow. Most consumer trackers will hit within 5% of manual count at normal pace; if yours is off by more than 15%, the algorithm is miscalibrated for your gait. Apple Watch and Garmin support stride-calibration walks to fix this.
FAQ
- Why does my phone count fewer steps than my friend's?
- Where you carry it matters most. Pocket > bag > belt. If it's in a backpack or handbag, it can undercount by 30%+.
- Should I trust GPS-derived distance more than step-derived?
- For outdoor walks, yes — GPS is more accurate for distance. Step counts are still useful as the universal portable metric.