Method
Confidence range
A confidence range is the band a reported score would be expected to fall within if the same measurement were repeated. Zebora publishes one on every figure, at every level of the roll-up, because a visibility score is an estimate from a finite sample of variable answers rather than a count of something fixed.
Why a bare percentage misleads
A figure quoted without a range invites a comparison it cannot support. Two numbers a few points apart may be statistically indistinguishable, and the only way a reader can know that is if the range is on the page next to them. Publishing it is a deliberate choice to be harder to over-read.
Ranges widen as slices narrow
The overall score rests on the whole taxonomy and carries a reasonably tight range. A single tag may rest on a couple of dozen prompts and carries a much wider one — wide enough that most month-to-month movement inside a tag is not a finding at all.
Two numbers that are the same number
One slice reports 41% and another 44%. Quoted bare, that reads as a three-point gap. Published with their ranges, the two bands overlap almost entirely, and the honest reading is that the slices are indistinguishable. The range is what stops a reader drawing a conclusion the measurement cannot support.
Related terms
- Effective sample size — What the width is calculated from.
- Noise threshold — The related question: not how wide, but how big a move counts.
- Within measurement noise — What we say when the ranges overlap too much to call it.
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.