The short definition
AI visibility is a brand's presence inside AI-generated answers. Where search engine optimisation asks whether you rank on a page of ten links, AI visibility asks whether the model names you at all — and whether what it says about you is accurate.
It is also called generative engine optimisation (GEO) or answer engine optimisation (AEO). The three terms describe the same work: making a brand legible, credible, and quotable to systems that answer instead of list.
Why it matters now
An answer engine returns one paragraph and typically names two or three brands. There is no page two and no consolation traffic for fourth place. Being absent from the answer means being absent from the consideration set entirely.
- Buyers increasingly start research inside an AI platform rather than a search box.
- The model, not the buyer, chooses which sources to read and cite.
- An outdated or wrong description of your brand can persist across millions of answers until the underlying sources change.
How models decide who to name
Three things drive whether a model names a brand: whether it can identify the brand as a distinct entity, whether it can find a clear statement of what the brand does, and whether independent sources it already trusts confirm that statement.
- Entity clarity: one consistent name, description, category, and set of profiles everywhere the model looks.
- Retrievable content: server-rendered pages, structured data, and passages that answer a question directly in the first sentence.
- Third-party corroboration: review platforms, directories, comparisons, and press the model already cites in your category.
How to measure it
Build a prompt set of the questions your buyers actually ask, run it across every platform on a fixed schedule, and record four things per prompt: whether you were mentioned, where in the answer, how you were characterised, and which sources were cited.
That gives a share-of-answer number per platform and a concrete list of the citations you still need to earn. It is the only way to tell improvement from noise, because model outputs vary between runs.
What AI visibility is not
It is not prompt injection, keyword stuffing, or paying a platform for placement. None of those survive a model update. The work that holds is the unglamorous kind: accurate entity data, genuinely useful content, and credible third-party mentions — all of which classic search rewards too.