Measurement library
How to measure AI answer visibility
Short, sourced explanations of the measurement questions that come up when a team starts checking whether AI answers cite them. Each page defines the metric, names its denominator, cites the provider documentation it depends on, and states what it does not prove.
Run a free auditMeasurement questions
-
Do Perplexity Sonar and Sonar Pro return different sources?
Yes.
-
How to measure whether ChatGPT cites your company
Ask a fixed set of unbranded buyer questions, repeat each one, and record for every answer whether your hostname appeared in the returned source list.
-
How to calculate an AI citation rate
An AI citation rate is the number of checks that cited your domain divided by the number of checks that completed.
-
What unmeasured should mean in an AI visibility report
Not cited means a check completed, returned a usable source list, and your domain was not in it.
-
OAI-SearchBot vs GPTBot
OAI-SearchBot and GPTBot are separate crawlers with separate jobs, and one robots.txt rule does not govern both.
-
Googlebot vs Google-Extended for AI Overviews
Google-Extended is not a Google Search control and does not determine whether a page can appear in an AI Overview.
-
Gemini citations versus Google AI Overviews
A citation from the Gemini model and a link inside a Google AI Overview come from two different systems.
How these pages are written
- Each page answers its title question in the opening paragraph, without the product.
- Every claim about a provider carries a link to that provider's own documentation.
- Anything that needs a real measurement is shown as a labelled evidence gap rather than asserted.
- Each page states what it does not establish.
The counting rules these pages describe are documented in the GetSounding measurement method, and the reliability gates for published benchmarks are in the SoundingScore Index.