Glossary

An answer engine responds to a query with a composed answer rather than a list of links, drawing on retrieved sources and usually citing some of them. ChatGPT, Perplexity, and Google's AI Overviews and AI Mode are the prominent examples.

AEO — answer engine optimization — is the practice of making content visible, retrievable, and citable inside answer engines: the systems that answer questions directly instead of returning links. It spans machine-readable structure, entity clarity, question-shaped content, and crawler access policy.

GEO — generative engine optimization — is the practice of optimizing content for visibility in generative engines: AI systems that compose answers from retrieved sources. The term entered use through a 2023 research paper of the same name and remains the discipline's preferred name in academic work.

LLMEO — large language model engine optimization — is the same discipline as AEO and GEO under yet another name: making content that LLM-based systems retrieve, trust, and cite. It is the least standardized of the acronyms and is often folded into the other two.

SEO — search engine optimization — is the three-decade-old practice of earning visibility in ranked search results. It is the parent discipline of AEO and GEO, and its center of gravity is moving into them as search engines embed AI answer layers above the rankings SEO was built for.

AI crawlers are the bots AI companies run to gather web content — for model training, for building search indexes, or for live retrieval when answering a query. Major user-agents include GPTBot and OAI-SearchBot (OpenAI), ClaudeBot and Claude-SearchBot (Anthropic), Google-Extended (Google), and PerplexityBot (Perplexity).

Citation share is the fraction of an answer engine's citations that go to a given domain or brand across a set of queries — the emerging visibility metric of AI search, where being cited replaces being ranked.

RAG — retrieval-augmented generation — is the architecture behind most answer engines: retrieve documents relevant to the query, then have a language model compose its answer from them, grounding the response in sources it can cite.

A zero-click search ends without a click to any website: the query was answered on the results surface itself. AI answer layers intensify the pattern, which makes citation presence and brand visibility the remaining prize on those queries.

AI visibility is how findable, retrievable, and citable a brand or site is across AI surfaces — answer engines, chatbots, and the AI features inside search engines. It is the discipline's umbrella measure: citation share, brand mentions in answers, and AI referral traffic are all ways of counting it.

Four names for substantially the same practice, each emphasizing a different corner: AEO the answer surfaces, GEO the generative architecture (the research literature's term), LLMEO the model doing the answering, and SEO the three-decade-old parent discipline of ranked search. The work — retrievable, trustworthy, citable content — is shared.

llms.txt is a proposed standard: a plain-text file at a site's root that indexes what the site offers to AI systems, with an optional llms-full.txt carrying the content itself. Google's documentation says no such file is needed to appear in Search or its AI features.

AI Overviews are Google's AI-generated answers, composed from web sources and shown above the traditional results. They are the largest AI answer surface on the web: a Similarweb analysis put them on 43% of Google searches as of July 2026, up from 15% a year earlier.

AI Mode is Google's conversational search surface: a full AI-generated answer with citations, ads and app integrations, running alongside classic results. Datos and SparkToro measured it at 0.13% of U.S. search-related visits in Q2 2026 — heavily instrumented, lightly used.

Agentic search is the step past answering: an AI system that browses, acts and transacts on a user's behalf rather than returning an answer for the user to act on. It reads a site as an agent, not a reader, which makes machine readability, actionability and access policy the optimization surface.

Grounding is the step that ties a generated answer to retrieved source documents, so the response reflects — and can cite — specific material rather than the model's parameters alone. The UK's CMA treats it as a distinct use of publisher content, separate from training and fine-tuning.

Query fan-out is the step where an answer engine turns one user question into its own set of background search queries and retrieves against those, rather than against the words the user typed. What the engine puts in those queries shapes the answer more than the pages it later fetches.

Citation drift is the rotation of which sources an AI engine cites for the same prompt over time, with the prompt and the page both unchanged. It is not malfunction — it is how retrieval-based answers behave, and it is why AI visibility is a distribution to move rather than a position to hold.

Coati is a name Google's VP of Search used once, at a conference in November 2022, describing it as an update that replaced the Panda algorithm. Google has published nothing about it since, and it appears in no Google documentation — including the ranking-systems guide Google maintains for exactly this purpose.

Google Discover is a personalised feed Google surfaces without a query, in the Google app and on Android home screens. For news publishers it now supplies more Google traffic than Search does: 67.51% against 27% in Q4 2025, across the 400+ publishers NewzDash tracks.