You and a friend walk the same lane, side by side.
At the gate you compare wrists.
The numbers are hundreds apart.
One of you feels cheated and the other feels smug.
Neither device is broken.
They simply disagree about what a step looks like.
How does a watch decide that was a step?
Your band has no idea what your feet are doing.
It only feels your wrist moving, many times a second.
Software then hunts through that movement for a pattern.
When a rise and fall repeats the way walking does, it counts a step.
Every brand writes that rule differently, and keeps it secret.
So a step is not a fact.
It is a judgement made by code.
Why do two devices disagree on the same walk?
Because each has drawn the line in a different place.
One counts cautiously and misses your shuffle around the kitchen.
Another counts generously and rewards you for standing and chatting.
Where you wear it matters too.
A wrist swings far more than a pocket does.
Your phone in a handbag records almost nothing.
Arm position changes it again.
Pushing a trolley or holding an umbrella lowers the count.
Where do the phantom steps come from?
From anything rhythmic your wrist does while you sit still.
Clapping at a wedding will add steps.
So will chopping vegetables or brushing your teeth.
A bumpy autorickshaw ride shakes your arm into a convincing walking rhythm.
Devices miss real steps in the other direction too.
A treadmill with your hands on the rails is the classic case.
The errors run both ways.
What should you do with the number?
Use it as a rough diary of your own activity.
Compare this week with last week, on the same device.
Ignore the round target the app suggests.
It was never based on you.
If your usual day is short, aim for a little more.
The useful question is whether you move more than you did a month ago.
Things to Remember
- A step count is an estimate, never an exact tally.
- Compare counts only from the same device on the same wrist.
- Expect phantom steps from clapping, cooking and rough roads.
- Ignore the default target and beat your own average.
- Judge the trend across weeks, not the score at bedtime.