The AEO and GEO Glossary: 62 Terms, Clearly Defined

This glossary defines the 62 terms that come up most often in answer engine optimisation (AEO), generative engine optimisation (GEO) and technical SEO. Each definition is short enough to be read on its own, and most link to a longer guide on this site. Terms are grouped alphabetically.

A to C

AEO (answer engine optimisation)

Work to make a site’s content the answer shown directly on search results pages, such as featured snippets and “People also ask” boxes. See AEO services.

AI Mode

Google’s conversational search experience, which breaks a question into subtopics, runs many searches and writes an answer with links. See what AI Mode means for SEO.

AI Overviews

AI-generated summaries shown at the top of some Google search results, with links to the pages used.

AI referral traffic

Visits that arrive from an AI assistant rather than a search results page, usually identified in analytics by a referrer such as chatgpt.com or perplexity.ai. Volumes are small next to search, so this understates AI visibility rather than measuring it.

Answer block

A short, self-contained passage written so it can be lifted whole into an answer, normally a direct answer of two or three sentences placed immediately under a question heading.

Answer-first writing

Putting the direct answer in the first sentence or two of a section, before any context or background. See quotability.

Backlink

A link from another website to yours. It remains a ranking signal in search and is also one route to the third-party mentions AI systems use to identify a brand. See digital PR for AI visibility.

BM25

A widely used keyword-matching method that scores documents by how often query words appear, adjusted for document length and word rarity.

Canonical source

The page a search engine or AI system treats as the original home of a fact or a piece of content. Being the canonical source for your own data matters more than being quick to republish someone else’s figure.

Canonical URL

The version of a page you want search engines to treat as the main one when several URLs show the same content, declared with rel=”canonical”.

Chunking

Splitting a page into smaller passages before storing them for retrieval by an AI system. See content chunking.

Citation (in AI answers)

A link or reference an AI assistant shows to indicate which page a statement came from.

Claude-SearchBot

Anthropic’s crawler that helps surface web pages in Claude’s search results.

Context window

The amount of text a language model can consider at once, measured in tokens. See context windows.

Core Web Vitals

Google’s three user experience metrics: Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift.

Corpus of trust

The informal set of sources a search or AI system reuses most often for a subject. Entering it is a matter of being accurate, consistent and corroborated over time, not a single technical change.

Cosine similarity

A measure of how similar two embeddings are, based on the angle between them. See embeddings.

Crawl budget

The number of URLs a search engine is willing and able to crawl on a site in a given period. It matters mainly for large sites.

D to G

Dense retrieval

Finding passages by comparing the meaning of a question and passages as embeddings, instead of matching exact words. See dense retrieval and re-ranking.

E-E-A-T

Experience, expertise, authoritativeness and trustworthiness: the qualities Google’s quality rater guidelines use to assess content. See E-E-A-T.

Embedding

A list of numbers that represents the meaning of a piece of text, so similar texts have similar numbers.

Entity

A distinct, identifiable thing, such as a company, person or place, that search engines and knowledge graphs store facts about.

Entity home

The single page you want search engines and AI systems to treat as the authoritative description of your organisation, usually the About page, carrying Organization schema and sameAs links to your real profiles. See entity resolution.

Featured snippet

A highlighted answer at the top of Google results, taken from a ranking page, in paragraph, list or table form.

GEO (generative engine optimisation)

Work to get a brand named and cited accurately in answers written by AI assistants. See GEO vs SEO.

Google-Extended

A robots.txt token that controls whether Google may use a site’s content for Gemini training and grounding. It does not affect Google Search.

GPTBot

OpenAI’s crawler for collecting data that may be used to train its models. Blocking it does not affect ChatGPT search.

Grounding

Basing an AI answer on retrieved sources instead of the model’s memory alone. See hallucination and grounding.

H to M

Hallucination

A statement by an AI system that sounds plausible but is false or unsupported by its sources.

hreflang

An annotation that tells search engines which language or regional version of a page to show to which users. See implementing hreflang.

IndexNow

A protocol that lets a site notify participating search engines, including Bing, when pages change. Google does not use it. See IndexNow.

Information gain

How much new, useful information a page adds compared with other pages on the same topic. See information gain.

Knowledge Graph

Google’s database of entities and the facts connecting them, used for knowledge panels and understanding. See knowledge graphs.

LLM SEO / AI SEO

Informal names for the same work as GEO: getting content retrieved, used and cited by search products built on language models. Worth knowing because clients and tools use the labels interchangeably.

llms.txt

A proposed Markdown file at a site’s root that summarises the site for language models. No major AI provider has said it uses the file to choose sources. See llms.txt.

Lost in the middle

A research finding that language models use information at the start and end of long inputs better than information in the middle. See lost in the middle.

Misinformation inoculation

Publishing a clear, well-sourced correction of a wrong claim about your company or your market, so retrieval has an accurate passage to find. Useful when an assistant keeps repeating an outdated fact about you.

N to Q

NAP

Name, address and phone number. Keeping them identical everywhere helps search engines and AI systems identify a business. See NAP consistency.

noindex

An instruction, in a meta tag or HTTP header, telling search engines not to include a page in their index. See robots.txt, noindex or canonical.

OAI-SearchBot

OpenAI’s crawler for ChatGPT search. Blocking it keeps pages out of ChatGPT search results. See getting cited in ChatGPT search.

Organic traffic

Visits from unpaid search results. It no longer captures the whole picture, because an AI answer can inform a buyer without producing any visit at all.

Passage ranking

Google’s ability, introduced in 2021, to rank a page based on one well-matched passage within it.

People Also Ask

A Google results feature listing related questions with short answers taken from web pages.

PerplexityBot

Perplexity’s crawler for building its search index. See getting cited by Perplexity.

Query fan-out

The technique of breaking one question into several sub-queries and searching them in parallel, used by Google’s AI Mode. See query fan-out.

R to S

RAG (retrieval-augmented generation)

A method in which an AI system retrieves relevant passages and writes an answer based on them. See retrieval-augmented generation.

Recommendation (in AI answers)

When an assistant suggests a company as a solution rather than only citing it as a source. It generally requires stronger identity signals and more third-party corroboration than a citation does.

Re-ranking

A second, more precise scoring of retrieved passages to choose the best few for an answer.

Rendering

Running a page’s JavaScript to build the final content. Googlebot renders; most AI crawlers do not. See our render test.

robots.txt

A file at a site’s root that tells crawlers which paths they may fetch. It controls crawling, not indexing.

sameAs

A structured data property that links an entity on your site to its profiles elsewhere, such as LinkedIn or Wikidata. See entity resolution.

Schema markup (JSON-LD)

Structured data in the schema.org vocabulary, usually written as JSON-LD, that describes a page’s content in machine-readable form. See JSON-LD schema.

Search intent

What the person actually wants from a query: a definition, a comparison, a product or a supplier. The live results for a query are the best available evidence of the intent a search engine has assigned to it.

Semantic specificity

Writing with explicit nouns instead of pronouns and vague back-references, so a passage still makes sense when it is retrieved on its own.

SEO (search engine optimisation)

Work to make a site rank in unpaid search results. It remains the base layer under AEO and GEO, because assistants retrieve from search indexes before they answer. See SEO services.

Share of answer

The share of a fixed question set on which an assistant mentions your brand, measured per platform and tracked over months. Usually the headline number in an AI visibility report.

Share of voice (AI)

The percentage of test runs in which an AI assistant mentions a brand for a set of questions. See tracking brand mentions.

T to Z

Topical authority

The depth and consistency of a site’s coverage of a subject, built by answering its questions thoroughly and linking the pages together. See topical authority.

Unlinked brand mention

A reference to a company by name with no hyperlink. It passes no link equity but still contributes to how search engines and AI systems identify and describe a brand. See brand mentions and co-occurrence.

Wikidata

An open, structured knowledge base whose items are widely reused by search engines and other systems. See entity resolution, sameAs and Wikidata.

YMYL (Your Money or Your Life)

Google’s term for topics such as health, finance and safety, where inaccurate content could cause harm, and which are held to the highest trust standard.

Zero-click search

A search that ends without a click to any website, because the answer was shown on the results page. See zero-click searches.

Sources

1. Google Search Central documentation on AI features, robots.txt, canonicalisation, hreflang and structured data
2. OpenAI, Anthropic and Perplexity crawler documentation
3. Liu et al., “Lost in the Middle: How Language Models Use Long Contexts”, 2023
4. Lewis et al., “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”, 2020