AI visibility is whether the AI engines recommend you, merely mention you, or leave you out entirely when customers ask the questions you should win. An AI visibility agency should show you those numbers, measured repeatedly across engines, before it takes any of your money. Ours does: the first thing a prospect sees is who the AI names today in their trade, free.
The outcomes that actually get scored
Every measured answer lands in one of five buckets: recommended (the engine puts you forward), mentioned (you appear without endorsement), cited only (your site is a source but you are not named in the answer), negative (named as a warning), or absent. Collapsing these into one "visibility score" hides exactly the information you are paying for: absent and negative need entirely different fixes, and a mention is not a recommendation. Ask any provider which buckets they score. If the answer is a single blended number, ask what is inside it.
Why a screenshot is not visibility
AI answers are not stable. In our published dataset, 19% of same-question, same-engine repeats change their verdict, and engines disagree with each other on 29% of questions. A screenshot is one draw from that distribution. Honest measurement asks each question repeatedly, quotes proportions with a confidence range, and holds back control queries so that when numbers move, you know whether the work moved them. The full method is public: how we measure, shown on real data. So are our readings, published measurements of whole UK trades, hundreds of answers at a time, so you can see what the instrument produces before pointing it at your own business.
Searched for a different name?
This discipline is young enough that the market has not settled on one name for it. The same work is sold as: AEO agency · GEO agency · AI SEO agency · LLM SEO · AI search optimisation agency. Whatever the label on the proposal, ask to see the five buckets, measured repeatedly.
Common questions
What is a good AI visibility score?
There is no universal number; visibility is relative to your market. The readings we publish show whole UK trades where most answers name nobody local at all, and others where one national brand takes a large share of every answer. What matters is your share of the questions you should win, against named competitors, measured the same way every time.
Can an agency guarantee AI visibility?
No. Nobody controls the engines, and answers move on their own: 19% of identical repeat runs change their verdict in our published data. What an honest agency can do is prove movement, a measured baseline, fixes, then a re-measure against untouched control queries so an engine-wide swing cannot be dressed up as their work.
How often should AI visibility be measured?
Within a measurement, repeatedly: the same question asked once is noise. Between measurements, re-measure after each batch of fixes has had time to be re-crawled, and quarterly as a floor once things are stable. Daily dashboards mostly chart the engines' own churn, which we quantified before selling anything.
The numbers this page describes are what the free ten-question audit hands you: who each engine recommends today on the ten questions you should win, before any money changes hands. If your visibility is already good, that is the report. Pricing is public: pilot from £500, monthly from £700.
Updated July 2026