Mention rate
How often AI names your brand, with the sample size attached
Mention rate is the share of successful runs in which a model named your brand in its answer. One prompt, one model, one country, one moment is a run. Every rate arrives with the number of runs behind it, and anything we never measured comes back as not measured — never as a zero.
By Karl-Gustav Kallasmaa, founder of Attensira · Last updated
In short
Mention rate counts successful runs that named your brand, divided by every successful run in the window. The unit is the run, so an answer that names you eleven times counts once. Failed runs are excluded from the denominator, every rate carries its sample size, and no change is reported until it clears a two-proportion z-test at 95%.
One definition
Runs naming you over all successful runs. No weighting, no blended score, no adjustments.
The sample is visible
Every rate is a value plus n, the count of successful runs it was built from. Read n first.
Not measured is not zero
An unmeasured cell returns null and renders as an em dash. Only a real, measured zero is drawn as zero.
Read from consumer surfaces
Where a consumer surface can be read, that is the primary path. Model APIs are a labelled fallback, never averaged in.