AI SEO and AEO glossary

This glossary defines 53 terms used in search engine optimisation (SEO), answer engine optimisation (AEO) and generative engine optimisation (GEO) — how Google and AI assistants such as ChatGPT, Gemini, Claude and Perplexity find, rank and cite websites. Each definition links to a guide that explains it in depth.

A

AEO — Answer engine optimisation / optimization: The work of making a website easy for AI assistants and AI search features to find, trust and quote in their answers. Learn more →
AI Mode: Google Search's conversational AI experience, which answers follow-up questions and shows links from the Search index. Learn more →
AI Overviews: AI-generated summaries shown at the top of some Google results pages, with links to the sources they draw on. Learn more →
AI SEO: Used in two senses: SEO carried out with AI tools, or optimising a website for AI search and assistants. Learn more →
AI share of voice: The share of citations or brand mentions in AI answers, over a fixed question set, that belong to one brand rather than its competitors. Learn more →
Answer engine: A service that replies to a question with a written answer rather than only a list of links — for example ChatGPT, Gemini, Perplexity, Copilot, Claude or Google AI Overviews. Learn more →
Answer-first passage: A short (about 40–60 words), self-contained answer placed directly under a heading that asks the question, so it can be quoted on its own. Learn more →

B

BM25: A lexical ranking function from the probabilistic relevance framework that scores documents by query-term frequency, rarity (IDF) and document length, with diminishing returns for repeated terms. Learn more →
Brand memory: What an AI model knows about a brand from its training data alone, without searching the web. Learn more →

C

Canonical URL: The single preferred address of a page, declared with rel="canonical", so duplicates are consolidated. Learn more →
ccTLD — Country-code top-level domain: A domain ending tied to a country, such as .in, .de or .co.uk — the strongest signal that a site targets that country. Learn more →
Citation rate: The proportion of sampled AI answers to a question that cite a given page or domain, best reported with a confidence interval. Learn more →
Cloaking: Showing search engines or AI crawlers different content from what people see — a violation of search spam policies. Learn more →
Confidence interval: A range that, under the model's assumptions, contains the true value with a stated probability (for example 95%); it shows how uncertain a measured rate is. Learn more →
Core Web Vitals: Google's user-experience metrics: Largest Contentful Paint (loading), Interaction to Next Paint (responsiveness) and Cumulative Layout Shift (visual stability). Learn more →
Crawler — Bot, spider, user agent: Software that fetches web pages for a search index or AI system, such as Googlebot, Bingbot, OAI-SearchBot, PerplexityBot or ClaudeBot. Learn more →
Cross-encoder re-ranker: A model that reads a query and a candidate passage together and scores their relevance, used to re-order the best candidates from a first-stage retriever. Learn more →

D

Dense retrieval — Vector search, embedding search: Retrieval that represents queries and passages as vectors (embeddings) and finds passages whose meaning is closest to the query. Learn more →

E

E-E-A-T — Experience, expertise, authoritativeness, trustworthiness: The framework Google's quality raters use to judge whether content and its creators can be trusted; trust is its central element. Learn more →
Entity: A uniquely identifiable thing — organisation, person, place, product or concept — that search engines and AI systems can attach facts to. Learn more →

F

False discovery rate — FDR, Benjamini–Hochberg: The expected share of false positives among the findings declared significant; controlling it keeps many-factor audits from reporting chance effects. Learn more →
Featured snippet: An excerpt from a page shown above Google's regular results as a direct answer — an early form of answer engine. Learn more →

G

GEO — Generative engine optimisation / optimization: Improving how visible a source is inside answers written by generative AI search systems; the term comes from a 2023 research paper published at KDD 2024. Learn more →
Google Business Profile: Google's free business listing that appears in Maps and local results, with hours, photos, reviews and services. Learn more →
Google-Extended: A robots.txt token that controls whether Google may use a site's content for Gemini model training and grounding; it does not affect Google Search. Learn more →
Grounding: Giving a language model retrieved, current source content at answer time so its answer is based on, and can cite, those sources. Learn more →

H

Hallucination: A fluent but false or unsupported statement produced by a language model. Learn more →
hreflang: An annotation that tells Google about the language and regional variants of a page, using ISO 639-1 language and ISO 3166-1 alpha-2 region codes. Learn more →
Hidden instructions — Instruction injection: Text planted in web content, often invisibly, that tries to steer an AI system that later reads it — a known attack on AI applications and a breach of search spam policies. Learn more →

I

IndexNow: An open protocol that lets a site notify participating search engines, such as Bing, immediately when URLs are added or changed. Learn more →

J

JSON-LD: The JavaScript Object Notation for Linked Data format recommended by Google for adding schema.org structured data to a page. Learn more →

K

Knowledge graph: A database of entities and the relations between them that a search engine uses to understand and display facts. Learn more →

L

Large language model — LLM: A neural network trained on large text collections to predict and generate language; the core of AI assistants. Learn more →
Lexical retrieval — Sparse retrieval: Retrieval based on matching the words of the query, such as BM25. Learn more →
llms.txt: A 2024 proposal for a Markdown file at a site's root that summarises the site and lists key pages for AI tools; not a ranking standard. Learn more →
Local pack: The map and short list of local businesses Google shows for searches with local intent, such as “dentist near me”. Learn more →
Lost in the middle: The finding that language models use information best when it is at the beginning or end of a long input, and worse when it is in the middle. Learn more →

N

NAP consistency — Name, address, phone: Using exactly the same business name, address and phone number on the website, in structured data and on every profile and directory. Learn more →
nDCG — Normalised discounted cumulative gain: A ranking metric that rewards relevant results more when they appear higher, discounting by the logarithm of position. Learn more →

P

PageRank: The 1998 link-analysis algorithm that scores a page by the number and importance of pages linking to it. Learn more →
Passage indexing: Ranking individual passages of a page, not just the page as a whole, so one well-written section can answer a specific question. Learn more →
Position bias: The tendency of people, and of AI judges, to favour whatever is shown first, independent of quality. Learn more →

Q

Query fan-out: Expanding one question into several related searches and combining their results to compose an answer. Learn more →

R

RAG — Retrieval-augmented generation: An architecture that retrieves relevant documents and gives them to a language model to generate an answer, typically with citations. Learn more →
Rich result: An enhanced Google result (such as product prices or ratings) made possible by structured data that meets Google's requirements. Learn more →
robots.txt: A file at a site's root that tells crawlers which paths they may fetch, standardised as the Robots Exclusion Protocol (RFC 9309). Learn more →

S

Scaled content abuse: Google's term for generating many pages mainly to manipulate rankings rather than help users, however they are produced. Learn more →
Schema.org: A shared vocabulary, launched in 2011 by major search engines, for describing organisations, products, articles and more in structured data. Learn more →
Search intent: What a person wants to achieve with a search — learn, compare, buy, find a place or reach a specific site. Learn more →
SEO — Search engine optimisation / optimization: Improving a website so it ranks higher and earns more clicks in search engine results. Learn more →
Structured data: Machine-readable information added to a page, usually in JSON-LD using schema.org, that describes what the page and its subject are. Learn more →

W

Wilson score interval: A confidence interval for a proportion that stays sensible at small samples and at 0% or 100%, unlike the simple normal approximation. Learn more →

Z

Zero-click search: A search that ends without a click to any website because the answer was shown on the results page or in an AI summary. Learn more →