AI Visibility Measurement
Short answer
AI visibility measurement establishes how often AI assistants name a company when asked the questions its buyers actually ask. A fixed prompt set is run repeatedly across ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, every answer is stored verbatim and dated, and the result is a mention rate, a citation rate and a share of voice against named competitors.
Measure first. Everything after a baseline is opinion.
What the engagement covers
- Prompt set design
- Twenty or more questions drawn from how your buyers actually phrase the problem, not from a keyword tool.
- Multi-engine sampling
- The same prompts across five answer surfaces, repeated, because a single run is noise.
- Verbatim archive
- Every answer stored as written, with the date and the engine. A clean before-state cannot be recreated later.
- Share of voice
- Your mentions as a proportion of all mentions, against competitors you name.
How it runs
- Agree the prompt set and the competitor list.
- Run the baseline and archive every response.
- Report mention rate, citation rate and share of voice.
- Re-run monthly. The delta is the report.
Frequently asked
Why not just check ChatGPT myself?
Because a single answer is one sample from a system that varies between runs, sessions and accounts. A rate across repeated runs on five engines is a measurement; one screenshot is an anecdote.
What is a good mention rate?
It depends entirely on the category. In a soft market, being named in eight of twenty prompts is strong. What matters more is the direction of travel from your own baseline.
Do I need this before the optimisation work?
Yes, and it is the one step that cannot be done retrospectively. Once the site changes, the clean before-state is gone permanently.
Contact us
Send us the brief or book a short call. You get concrete next steps in reply.