What is AI visibility?
Published 20 September 2026. Last checked 20 September 2026.AI visibility is whether an AI answer system names your business when somebody asks it a question your business could answer. It is measured by putting real buyer questions to each system and recording who gets named, in what order, and for what stated reason. It is not the same thing as a search ranking.
Why it is not a search ranking
A search engine returns a list of links and the user chooses. An answer system returns an answer, and in most cases the user does not go any further.
That changes what visibility means. In search, being on page one gives you a chance. In an answer, there is no page one — there is a handful of named firms and everybody else, and the systems that do the naming publish no ranking, notify nobody, and vary their answer between one run and the next.
Which systems count
- Google AI Overview
- Google AI Mode
- ChatGPT
- Microsoft Copilot
- Claude
- Perplexity
- Gemini
Seven systems give answers rather than lists: Google’s AI Overview, Google’s AI Mode, ChatGPT, Microsoft Copilot, Claude, Perplexity and Gemini.
Three distinctions are routinely collapsed and should not be. Google AI Overview and Google AI Mode are different products with different behaviour. Gemini is not Google’s AI answers — it is a separate assistant with a separate index. Copilot is not ChatGPT despite sharing underlying models; it answers differently and, in testing, it reaches for smaller and less-known firms than any of the others.
How it is measured
Put a fixed set of buyer questions to each system, under controlled conditions, on a stated date. Record every firm named, the order, the stated reason, and what the system cited as its source. Four conditions decide whether the reading means anything:
No personalisation. A signed-in account returns an answer shaped by that account's history, not what a buyer sees.
The full product, not a downgraded one. On some platforms, signing out returns a weaker model than the one people actually use, which is a different measurement rather than a cleaner one.
A fixed location. Answers vary by country.
More than one run. Answers move between consecutive runs on the same day. A single screenshot proves nothing.
Three results, not one
A business can be recommended by name, merely mentioned in passing, or cited as a source for something the system says. These have different causes and different remedies, and a single visibility score hides which one has happened.
What moves it
Nobody outside these companies knows the mechanism, and any supplier claiming otherwise is guessing. What is observable from the readings themselves is consistent enough to be worth stating:
Checkable facts get quoted; adjectives do not. Small, unknown firms that publish a price and a scope get named. Firms describing themselves as comprehensive, trusted or leading do not.
Independent corroboration outweighs self-description. What a register, a directory or a third-party page says about a business carries further than what the business says about itself.
The systems cite pages that answer the question directly. A page whose first paragraph answers the question it is titled with is the passage that gets lifted.
Entity clarity matters. A business a system can describe confidently — legal name, location, what it sells, to whom — gets named more readily than one it cannot pin down.
What nobody can promise
Placement. Results are directional, they vary by location, account, prompt, browsing mode and time, and no supplier controls the systems doing the naming. A guarantee of being recommended by an AI answer system is a guarantee about somebody else’s product.
How to act on it is answer engine optimisation, the same practice under its other name is LLM SEO, and the measurement itself is the AI Visibility Audit. The free website health check covers the machine-readability half of it.
Questions
Find out whether AI names your business.
25 agreed buyer questions, seven answer surfaces, two runs, one report.
Request an AI Visibility Audit