// Glossary

GEO glossary

Plain-English definitions of the terms you'll keep running into when working on generative engine optimisation. Linkable, structured, and AI-readable.

Core concepts

GEO: Generative Engine Optimisation. The practice of helping AI search and answer engines crawl, understand, trust and cite a website.
AEO: Answer Engine Optimisation. Often used interchangeably with GEO; emphasises optimising for systems that return synthesised answers rather than link lists.
Generative engine: Any system that produces a synthesised, natural-language answer from web content – ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, Copilot.
Answer engine: A search interface that returns answers rather than ranked links. Often built on top of a generative engine.
Citation readiness: How easily a generative engine can attribute a specific piece of information to your website. Driven by schema, clarity, authorship and accessible content.
Entity: A distinct thing the web describes – a company, person, product, place. Generative engines reason about entities, not just keywords.

AI crawlers

GPTBot: OpenAI's primary web crawler, used to discover and ingest content. Controlled via robots.txt User-agent: GPTBot.
ChatGPT-User: OpenAI's live-fetch agent. Activates when a ChatGPT user requests a specific URL during a conversation.
OAI-SearchBot: OpenAI's crawler powering ChatGPT's search experience.
ClaudeBot: Anthropic's training crawler for Claude.
Claude-Web: Anthropic's live retrieval agent that fetches pages on behalf of a Claude user.
PerplexityBot: Perplexity's index crawler.
Perplexity-User: Perplexity's live fetch agent for user-initiated queries.
Google-Extended: A robots.txt token Google honours specifically to opt out of generative training, distinct from regular Googlebot.
Applebot-Extended: Apple's equivalent opt-out token for Apple Intelligence training.
CCBot: Common Crawl's bot. Common Crawl is widely used as a training source by many AI providers.

Files and standards

robots.txt: A plain-text file at the site root that tells crawlers which paths they may or may not access. Honoured by all major AI bots.
sitemap.xml: An XML index of canonical URLs that helps crawlers discover content.
llms.txt: A proposed Markdown file at the site root summarising key pages for LLM crawlers. Adoption is emerging.
llms-full.txt: An optional companion to llms.txt that includes the full text of the listed pages.
ai.txt: A separate proposed convention for declaring AI training permissions. Less widely adopted than robots.txt.

Structured data

Structured data: Machine-readable markup that explicitly describes entities, relationships and content types.
JSON-LD: JavaScript Object Notation for Linked Data. The recommended format for embedding schema.org markup in web pages.
Schema markup: Vocabulary from schema.org used to describe Organisations, Articles, Products, FAQs and more.
Organization schema: JSON-LD describing a business as an entity – name, url, logo, sameAs, contactPoint.
FAQPage schema: JSON-LD describing question-and-answer pairs. Highly extractable by AI engines.
Product schema: JSON-LD describing a product – price, availability, brand, ratings – essential for ecommerce GEO.
Article schema: JSON-LD describing editorial content – headline, datePublished, author, mainEntityOfPage.
BreadcrumbList: JSON-LD describing the navigational path to the current page. Cheap and useful.

Trust and content

E-E-A-T: Experience, Expertise, Authoritativeness, Trustworthiness. Google's framework for evaluating content quality; useful as a GEO heuristic too.
Authorship signals: Bylines, author bios, Person schema and consistent attribution that help engines tie content to a real human.
sameAs: A schema.org property used to link an entity to its verified profiles elsewhere – LinkedIn, Wikipedia, Crunchbase, Companies House.
Server-side rendering: Producing meaningful HTML on the server before it reaches the browser. AI crawlers parse HTML, not arbitrary JavaScript output.
Hallucination: When a generative engine fabricates a fact. Strong structured data and clear authorship reduce the chance of being misquoted.
Ground truth: Verifiable, source-of-truth content an engine can cite. Your job in GEO is to be that source.

Products and surfaces

AI Overviews: Google's generative answer feature that summarises content above traditional search results.
ChatGPT search: OpenAI's web-aware search experience inside ChatGPT, powered partly by OAI-SearchBot.
Perplexity: An answer engine that synthesises responses with inline citations. Heavy emphasis on source attribution.
Copilot: Microsoft's generative assistant, integrated with Bing search results.