The same check, indoors and out
MOTION
Accuracy
95.2%
Step Count Accuracy, Indoor Walk
on a treadmill, against an observer’s hand tally count; outdoors, on pavement and kerbs, 91.8%
Reference standard
Hand tally counter
Sample size
22 participants
On a treadmill the Ultrahuman Ring lands within 4.8% of a hand count of every step taken. Step outside, where the surface tilts and you stop at kerbs, and it lands within 8.2%.
95.2%indoors, on a treadmill
91.8%outdoors, at an everyday pace
3.3 ptswhat stepping outside costs
950 / 946steps recorded per 1,000 taken across all walks, indoors and out
A walk is not one thing, and a step counter feels every difference. Indoors on a belt your stride repeats almost identically for as long as you stay on it: same surface, same pace, same arm swing, minute after minute. Step outside and none of it holds still. The pavement tilts, you slow at kerbs and speed up again between them, you turn, and you break stride for a door. Those are genuinely different signals arriving at the same sensor.
The gap between them is 3.3 percentage points, and the one published comparison that ran both conditions against manually counted steps found a gap in the same direction. Its overground leg was an indoor track rather than a street, so if anything it understates what stepping outside costs.
That is why a single figure measured in one setting tells you very little, and why both are on this card. The dashed line marks the tighter end of the range the expert consensus on step-count validation treats as acceptable for general use — an error of 10% to 15% — and the line is drawn at the 10% end. It is on the chart because a number without a threshold beside it is hard to judge.