Method
Noise threshold
The noise threshold is the smallest change between two periods that Zebora will describe as a rise or a fall. Below it, a move cannot be distinguished from what you get by measuring the same thing twice, so it is reported as stable. The threshold is worked out for each comparison rather than fixed.
Why a fixed rule would not work
How much a figure can wobble depends on the size of the slice and how close it sits to the extremes, so the same movement can be meaningful for the overall score and meaningless for a single tag. One blanket threshold would be too strict in one place and far too loose in another.
It covers sampling only
The threshold accounts for the variation that comes from sampling a finite set of answers. Genuine model drift sits on top of that. Which is why two consecutive moves in the same direction are much stronger evidence than one large move in a single period.
A fall that was not a fall
A brand's overall figure drops by five points between two months. On a slice that size, repeating the same measurement twice produces moves of roughly that magnitude on its own. The drop sits below the threshold, so it is reported as stable. A dashboard showing it as a red arrow would be narrating the measurement, not the market.
Related terms
- Within measurement noise — What a below-threshold move is called in the reporting.
- Confidence range — The same uncertainty, expressed on a single figure rather than a change.
- Effective sample size — Smaller slices get a higher bar, because of this.
Last reviewed . Definitions are reviewed quarterly, and whenever the underlying measurement changes.
Don't let AI decide your brand's future without you.
See exactly where you stand vs. competitors—and what to do about it.