GEO and AI search glossary

The work of understanding and improving how AI tools such as ChatGPT, Gemini, Claude, Perplexity and Google's AI Overviews describe, cite and recommend a brand. The term comes from a 2023 research paper led by Princeton University. See What is GEO?

Another name for much the same work. Some people use AEO for Google's AI summaries and GEO for chat tools. In practice they overlap almost completely.

Any search experience where an AI writes an answer instead of, or as well as, returning a list of links. Includes ChatGPT search, Perplexity, Google's AI Overviews and AI Mode, Copilot and Gemini.

A loose term for how present and how well described a brand is in AI answers. Useful as shorthand; be wary of anyone who turns it into one precise number.

The technology behind AI chat tools. It learns patterns from huge amounts of text and generates answers from them. Many now also search the web live before answering.

The AI-written summaries Google shows above, or among, its normal search results, with links to sources. Powered by Gemini.

Google's conversational search mode, where the whole result is an AI answer you can follow up on. It tends to cite a wide range of sources, most of which don't rank in the top ten for the original query.

When an AI tool links to or names a source as evidence for its answer. A citation can be a link to a page (your site, a news story, a review, a forum thread) or a named reference within the answer.

A mention is when the AI names your brand in its answer. A citation is when it points to a source. You can be mentioned without being cited, and your site can be cited without your brand being recommended. Both are worth knowing.

A source ChatGPT links to in an answer, usually when it has searched the web. Answers built only from training data often have no citations at all, even when they mention brands.

Perplexity shows its sources prominently, usually as numbered references. It tends to cite more sources per answer than other tools, which makes it easier to see what it's relying on.

How often your brand, or your sources, appear across a set of AI answers compared with competitors. Also called AI citation share. It's the AI version of a familiar PR measure, and it only means something across many questions and many runs.

How much AI answers change when the same question is asked again. It's high: one 2026 study found the same list of brands came back less than 1 time in 100. This is why single checks and "rankings" in AI are unreliable.

When an AI tool turns one question into several related searches behind the scenes, then builds one answer from all the results. It's a big reason AI answers cite pages that don't rank for the original question.

Coverage and mentions you didn't pay for and don't own: journalism, reviews, commentary, reference sites, community discussion. Most AI citations point to this kind of source, though studies define it differently.

A frame from Seer Interactive's 2026 research: brands do better in AI answers when they have a clear identity, third-party validation and people talking about them. Useful as a lens, not a checklist.

A distinct thing an AI or search engine recognises: a company, person, product or place. Using one consistent name and description helps AI treat you as one clear entity.

When an AI states something false with confidence, such as the wrong founder, a product you don't make or a closed office. Clear, consistent, up-to-date information across the web reduces the risk, but can't remove it.

A search that ends without the person clicking through to any website. About two thirds of US Google searches now end this way, and AI answers push the share higher.

A bot an AI company uses to read web pages, either to train models or to fetch pages live for answers. Examples include GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot. If they're blocked, your pages can't be cited.

A proposed text file that summarises a website for AI tools. Harmless, but there's little evidence AI crawlers read it.

An open standard for publishing a machine-readable description of a business, its people and what it does, so AI tools can read it reliably. This site publishes one at /entitymap.json.

Structured data added to web pages to describe what's on them. Good housekeeping for search, but a 2026 Ahrefs study found adding it made no meaningful difference to AI citations.

Guidance from AMEC, the global body for communications measurement, published in May 2026. It warns against treating AI visibility as a stand-alone score and calls for measurement linked to outcomes.