Glossary
AI search glossary
50 terms you will meet when buying AEO, GEO or AI search work, in plain English, with a note on how settled each one is.
What do AEO and GEO mean?
Answer engine optimization (AEO) makes your pages the clearest answer to buyers’ questions. Generative engine optimization (GEO) improves how AI assistants describe and cite your business across the web. The work overlaps heavily, and Google says optimizing for its AI features is still SEO. Reviewed 2026-11-16; AI products change quickly, so each term notes whether a platform documents it, the industry uses it, or it is mainly marketing language.
Disciplines
- Answer engine optimization(AEO)
- Making your pages the clearest, most quotable answer to the questions buyers ask, so search features and AI assistants can use them.
- Why it matters: It turns vague marketing pages into direct answers. Sources disagree on whether AEO is part of GEO or the same thing; in practice the work overlaps heavily.
- Industry: widely used; no single ownerWikipedia: Generative engine optimization
- Generative engine optimization(GEO)
- Improving how generative AI systems describe, recommend and cite a business: the facts they repeat and the sources they use.
- Why it matters: AI answers are assembled from many sources, so consistency and evidence across the web matter as much as your own pages.
- Industry: widely used; no single ownerWikipedia: Generative engine optimization
- AI SEO(LLMO, GSO, AI search optimization)
- Umbrella labels for optimizing for AI-driven search. They describe the same territory as AEO and GEO rather than separate disciplines.
- Why it matters: Buy the work, not the acronym. Google’s own guide says optimizing for its generative AI features is still SEO.
- Marketing: mostly vendor languageGoogle: optimizing for generative AI features
- Search engine optimization(SEO)
- Helping search engines find, understand and rank your pages. It remains the foundation: Google says AI Overviews and AI Mode use its core ranking and quality systems.
- Why it matters: A page that cannot be crawled, indexed and shown with a snippet cannot appear in Google’s AI features either.
- Industry: widely used; no single ownerGoogle: AI features and your website
- AI Overviews
- Google’s AI-generated summaries shown above or among search results, with links to the sources used.
- Why it matters: Being linked there is a new form of visibility; being summarised without a click is a new form of zero-click.
- Platform: documented by a search or ai companyGoogle: AI features and your website
- AI Mode
- Google Search’s conversational mode, where follow-up questions are answered with AI and linked sources.
- Why it matters: It answers multi-step questions, so pages that cover a topic completely and clearly have more chances to be used.
- Platform: documented by a search or ai companyGoogle: AI features and your website
- Search generative AI setting
- A Search Console setting that includes (by default) or excludes a site from AI Overviews, AI Mode and generative AI in Discover.
- Why it matters: Excluding a site removes it from those features without changing normal rankings; reports suggest it can also affect preferred-source eligibility. It does not cover the Gemini app.
- Platform: documented by a search or ai companyGoogle: preferred sources
- Search Generative AI performance reports
- Search Console reports, announced in June 2026, showing how a site performs in Google’s generative AI features.
- Why it matters: The first first-party measurement of AI visibility in Google. Use it alongside your own dated observations.
- Platform: documented by a search or ai companyGoogle: Search Generative AI performance reports
- Preferred sources
- A Google feature that lets people choose publishers they want to see more of, including in AI Mode and AI Overviews for eligible sites.
- Why it matters: Loyal readers can now raise a site’s presence in their own results.
- Platform: documented by a search or ai companyGoogle: preferred sources
- Snippet controls(nosnippet, max-snippet, data-nosnippet)
- Page-level directives that limit what Google may show from a page, including in AI features.
- Why it matters: Use them to keep specific text out of answers, knowing they also limit normal snippets.
- Platform: documented by a search or ai companyGoogle: AI features and your website
- Scaled content abuse
- Google’s spam policy against producing many low-value pages, by AI or otherwise, mainly to manipulate rankings.
- Why it matters: Thousands of near-identical city or keyword pages are a risk, not an AI search strategy.
- Platform: documented by a search or ai companyGoogle Search spam policies
How AI answers work
- Query fan-out
- An AI system splitting one question into several related searches, then combining what it finds.
- Why it matters: It explains why results vary between runs and why covering the related questions around a topic helps. Google warns against mass-producing near-duplicate pages to chase it.
- Industry: widely used; no single ownerGoogle Search spam policies
- Grounding
- Giving an AI model fresh sources, such as search results, to base its answer on, instead of memory alone.
- Why it matters: Grounded answers can cite your pages, which is why being crawlable and clear still matters.
- Industry: widely used; no single owner
- Retrieval-augmented generation(RAG)
- A system that retrieves relevant documents first, then has a model write an answer from them.
- Why it matters: The same method powers company knowledge assistants and many AI search products.
- Industry: widely used; no single owner
- Embeddings and vector search
- Turning text into numbers that capture meaning, so systems can find passages that mean the same thing even with different words.
- Why it matters: Matching by meaning rewards pages that explain a topic clearly, not pages stuffed with keywords.
- Industry: widely used; no single owner
- Reranking
- A second step that reorders retrieved passages by how well they answer the question.
- Why it matters: Direct, specific answers tend to survive this step.
- Industry: widely used; no single owner
- Hallucination
- An AI stating something false with confidence.
- Why it matters: Consistent, correct facts across your site and profiles reduce what models can get wrong; nobody can promise zero errors.
- Industry: widely used; no single owner
- Context window
- How much text a model can consider at once.
- Why it matters: Long, unfocused pages compete with themselves; a clear answer near the top is easier to use.
- Industry: widely used; no single owner
- Large language model(LLM)
- The kind of AI model behind ChatGPT, Claude, Gemini and similar assistants.
- Why it matters: Each product combines a model with its own search and source choices, so results differ by product.
- Industry: widely used; no single owner
Measurement
- Mention rate
- The share of recorded AI answers that name your business, out of the valid answers for an agreed question set.
- Why it matters: Always ask for the denominator, the dates and the platforms. A single answer is not a rate.
- Industry: widely used; no single owner
- Citation rate
- The share of recorded AI answers that link to your domain as a source.
- Why it matters: A mention without a link sends no traffic; a link without a mention may still help. Report them separately.
- Industry: widely used; no single owner
- Answer accuracy
- Whether what an AI says about your business is correct: name, services, prices, location, contact.
- Why it matters: Being mentioned wrongly can cost more than not being mentioned. Track errors and their likely source.
- Industry: widely used; no single owner
- Visibility baseline
- A dated record of how AI answers treat you before any work starts: questions, platforms, locale, answers saved.
- Why it matters: Without it, nobody can show whether anything changed.
- Industry: widely used; no single owner
- Zero-click search
- A search answered on the results page, so the person never visits a website.
- Why it matters: Measure enquiries and brand demand, not only clicks.
- Industry: widely used; no single owner
- AI visibility score
- A single number some tools give for AI presence.
- Why it matters: Ask how it is calculated. A score without the questions, platforms, dates and denominator behind it cannot be checked.
- Marketing: mostly vendor language
- Buyer question set
- The fixed list of questions, agreed with the business, that AI answers are tested against over time.
- Why it matters: Changing the questions each month makes any trend meaningless.
- Industry: widely used; no single owner
Crawlers and access
- Google-Extended
- A robots.txt token, not a separate crawler, that controls whether Google may use your content for Gemini training and grounding in some other Google systems.
- Why it matters: Blocking it does not remove you from AI Overviews or AI Mode; those follow Googlebot and the Search generative AI setting.
- Platform: documented by a search or ai companyGoogle: common crawlers
- AI search crawler(OAI-SearchBot, Claude-SearchBot, PerplexityBot)
- A bot that indexes pages so an AI product can find and cite them in answers.
- Why it matters: Blocking these can remove you from that product’s answers.
- Platform: documented by a search or ai companyOpenAI: overview of crawlers
- User-triggered fetcher(ChatGPT-User, Claude-User, Perplexity-User)
- A fetch made because a person asked an assistant to read or check a page.
- Why it matters: Vendors document different robots.txt behaviour for these; check each vendor’s current documentation.
- Platform: documented by a search or ai companyOpenAI: overview of crawlers
- Training crawler(GPTBot, ClaudeBot, Meta-ExternalAgent, Applebot-Extended, CCBot)
- A bot collecting content to train AI models, separate from search indexing.
- Why it matters: Allowing or blocking training is a separate business decision from appearing in AI search.
- Platform: documented by a search or ai companyOpenAI: overview of crawlers
- llms.txt
- A proposed Markdown file at /llms.txt that summarises a site for language models.
- Why it matters: Google says Search ignores it, and no major AI company has confirmed using it for answers. It can still help developer tools and agents. It is not a ranking requirement.
- Industry: widely used; no single ownerllmstxt.org proposal
- IndexNow
- An open protocol, started by Bing and Yandex in 2021, for telling participating search engines that a URL was added, changed or deleted.
- Why it matters: Copilot’s answers rest on the Bing index, so getting changes into Bing quickly matters.
- Platform: documented by a search or ai companyIndexNow
- Perplexity Publishers Program
- A revenue-sharing programme for publishers whose content Perplexity uses, announced in July 2024.
- Why it matters: A commercial deal for publishers; ordinary sites are cited without it.
- Platform: documented by a search or ai companyPerplexity Publishers Program
- AGENTS.md
- A Markdown file of instructions for AI coding and task agents working with a project or site.
- Why it matters: Useful for developer-facing products; irrelevant to most buyers’ search visibility.
- Industry: widely used; no single owner
Agents and commerce
- AI agent
- An AI system that takes steps on someone’s behalf, such as searching, comparing, filling forms or buying.
- Why it matters: Agents read sites to act, not just to answer, so clear prices, specs and policies matter.
- Industry: widely used; no single owner
- Model Context Protocol(MCP)
- An open protocol that lets AI assistants connect to tools and data sources in a standard way.
- Why it matters: A business can expose search, booking or catalogue functions to assistants through an MCP server.
- Industry: widely used; no single ownerModel Context Protocol
- WebMCP
- An emerging proposal for websites to expose MCP-style tools to browser-based agents.
- Why it matters: Early and still changing in 2026; worth watching, not yet a requirement.
- Industry: widely used; no single owner
- NLWeb
- A Microsoft open project that gives a website a natural-language question endpoint built on its structured data.
- Why it matters: Adoption is low so far; good structured data is the part that helps everywhere.
- Industry: widely used; no single owner
- Agentic commerce
- Shopping where an AI agent discovers, compares and sometimes pays for products for a person.
- Why it matters: Accurate product data, stock and policies become what an agent judges you on.
- Industry: widely used; no single ownerWikipedia: Agentic commerce
- Agentic Commerce Protocol(ACP)
- An open standard from OpenAI and Stripe for checkout flows between buyers, AI agents and merchants.
- Why it matters: Reports disagree on how in-chat checkout is offered today; check current documentation before planning around it.
- Platform: documented by a search or ai companyWikipedia: Agentic commerce
- Universal Commerce Protocol(UCP)
- A commerce protocol announced by Google with retail partners in January 2026, covering discovery, cart, checkout and after-sale steps.
- Why it matters: Availability to merchants was limited during 2026; watch the official programme.
- Platform: documented by a search or ai companyWikipedia: Agentic commerce
- Agent Payments Protocol(AP2)
- A protocol for proving that a person authorised an AI agent to make a payment.
- Why it matters: It addresses trust in agent purchases; its governance was moving during 2026.
- Platform: documented by a search or ai companyWikipedia: Agentic commerce
- Agent readiness
- How easily AI agents can read and use a site: crawl access, clear content, structured data and, optionally, agent endpoints.
- Why it matters: The foundations are the same as good SEO; extra protocols are optional and changing.
- Marketing: mostly vendor language
Content and entities
- Content chunking
- Rewriting pages into small self-contained blocks in the belief that AI systems prefer them.
- Why it matters: Google’s May 2026 guide lists it among tactics you can skip. Write clearly for people; clear structure helps everyone.
- Marketing: mostly vendor languageGoogle: optimizing for generative AI features
- Entity
- A distinct thing a search system recognises, such as your company, a product or a person, with facts attached.
- Why it matters: Consistent names, addresses and descriptions help systems treat you as one business, not several.
- Industry: widely used; no single owner
- Knowledge graph
- A search engine’s database of entities and the facts linking them.
- Why it matters: Panels and AI answers draw on it; no one can promise you a panel.
- Platform: documented by a search or ai company
- Structured data(schema.org markup)
- Code that states facts about a page, such as organisation, service, product or breadcrumb, in a standard vocabulary.
- Why it matters: Google says no special markup is needed for AI features; accurate standard markup still helps search understand pages.
- Platform: documented by a search or ai companyGoogle: AI features and your website
- E-E-A-T
- Experience, expertise, authoritativeness and trust: the qualities Google’s quality raters look for in content.
- Why it matters: Real authors, real evidence and real contact details are the honest way to show them.
- Platform: documented by a search or ai company
- Answer-ready page
- A page that answers its main question directly near the top, then gives the detail, evidence and next step.
- Why it matters: Useful for people first; the same clarity helps AI systems quote it accurately.
- Marketing: mostly vendor language
Want to know how AI assistants describe you?
Start with a dated AI search audit: the questions, the answers, the gaps and a written plan.

