How smartwatches and fitness trackers count your steps
Wrist wearables count steps by reading the rhythm of arm swing from a motion sensor. That is why a stroller, a treadmill rail or a bumpy car ride can throw the count off.
Pulse, covering wearables and health sensors
, 3 min read
3 axesof motion an accelerometer measures to find each step
What to know
- Step counts come from an accelerometer, a chip that measures movement along three axes.
- Software looks for a repeating rhythm at walking pace and ignores isolated jolts.
- A wrist device infers steps from arm swing, so walking with a still arm is undercounted.
Almost every smartwatch, fitness band and smart ring shows a daily step count. None of them sees your feet. The number is an estimate built from a small motion sensor and software that guesses which movements were steps.
The sensor
The core part is an accelerometer, a tiny chip that measures acceleration along three axes: up and down, side to side, and forward and back. It samples many times a second, producing a stream of numbers that rise and fall as the body moves. Many devices also have a gyroscope, which measures rotation and helps the software tell an arm swinging from a wrist turning.
The same sensors drive other features, such as fall detection and automatic workout detection.
Finding the rhythm
Walking has a signature. Each step sends a small jolt up through the body, and on the wrist the arm swings forward and back at the same pace. In the sensor data that shows up as a series of peaks that repeat at a steady rate, typically a little under two steps a second for walking and faster for running.
Step-counting software looks for that pattern. A common approach is to:
- Filter the signal to remove slow drift and fast noise.
- Find peaks that are strong enough to be a step.
- Check the rhythm, accepting peaks only when several arrive at a plausible walking or running pace.
The rhythm check is why a single bump or a wave of the hand usually does not add a step, and why some devices hold back a few steps before counting a walk, then add them in once the pattern is clear. Newer devices increasingly use machine-learning models trained on recorded walking, in the lab and in daily life, instead of hand-tuned rules. The exact methods are proprietary and change with software updates.
Why the count goes wrong
A wrist device measures the arm, not the legs, and infers steps from it. That leads to predictable errors.
Undercounting happens when you walk but your arm stays still: pushing a stroller or shopping cart, holding a treadmill rail, carrying bags, walking a dog on a lead, or keeping your hands in your pockets. Very slow walking, such as shuffling around a kitchen, can also fall below the rhythm threshold.
Overcounting happens when the arm moves rhythmically but you do not walk: brushing teeth, folding laundry, clapping, drumming, or riding on a bumpy road.
Placement matters too. A device on the hip or ankle sits closer to the legs and generally counts walking steps more reliably than one on the wrist, which is why research studies often use hip- or ankle-worn reference pedometers. A ring sits on the hand, so it faces the same arm-swing problem as a watch.
Comparing devices
Two trackers worn at once often disagree by hundreds or thousands of steps a day, because each maker uses its own thresholds and models, and some count the day from midnight while others use their own daily cycle. Accuracy studies generally find wrist devices are closer for steady walking than for daily life, and that errors vary by brand, activity and how the device is worn.
The practical upshot: a step count is most useful compared with itself over time on the same device, much like resting heart rate trends or heart rate variability. A sudden jump or drop after a software update can come from the counting method, not from how much you moved.
Why it matters
Step counts are no longer just a number on a watch face. Some apps feed them into fitness scores and "body age" estimates, and doctors and researchers increasingly look at them. Knowing that the count is an estimate from arm movement explains why it misses a walk behind a stroller, why two devices disagree, and why trends on one device mean more than any single day's total.
This article is general information about how technology works. It is not medical advice. Talk to a clinician about your own health.
WearablesDigital health and AI in careSmart rings
Written by an AI correspondent under standards set by editor Jacob Knight. Spotted an error? Tell us and we will fix it.