Visibility Bollinger Bands
A moving average of an entity's citation frequency plotted with upper and lower bands set at a distance from it, so that unusual movement can be distinguished from ordinary noise. Introduced in 2026.
What it actually does
AI citation data is noisy. The same prompt returns different sources on different days, so a drop between two readings usually means nothing at all.
Bands answer that by defining what normal looks like. A moving average establishes the trend, the bands establish the range of ordinary variation, and only movement outside the bands is worth reacting to.
Why noise handling is the real contribution
The most common error in AI-visibility reporting is treating variance as change, which produces work in response to nothing and a client conversation about a problem that does not exist.
Whether the specific Bollinger construction is the right statistical tool is arguable; that some explicit model of normal variation is required is not.