This GEO glossary explains the terms you'll come across when you want to make your website visible in AI search engines such as ChatGPT, Perplexity, Gemini and Google AI Overviews. Each term starts with a one-sentence definition, followed by a short explanation in plain language. Where a term is covered in depth in its own article, you'll find a link to it.
If you're new to the field, read the complete guide to GEO first. There you'll see how these terms fit together: from the crawler that fetches your page to the citation in the answer a user reads. After that, use the glossary as a reference; every article in the knowledge base links to the terms below.
GEO (Generative Engine Optimization)
GEO (Generative Engine Optimization) is optimising your website and content so that generative AI systems such as ChatGPT, Perplexity and Google AI Overviews mention your brand and cite your pages as a source. Where SEO is about a high position in a list of links, GEO is about being included in the answer itself. That calls for accessible technology, clear structure and content that can be quoted on its own.
Read more: complete guide to GEO
AEO (Answer Engine Optimization)
AEO (Answer Engine Optimization) is optimising content so that search engines and voice assistants can extract a direct answer from it. The term dates from the era of featured snippets and voice search, before generative AI was widely available. In practice AEO and GEO are often used interchangeably; GEO puts more emphasis on citations in answers written by AI.
Read more: difference between GEO and SEO
LLM (Large Language Model)
An LLM (Large Language Model) is an AI model trained on large amounts of text, which enables it to understand and generate text itself. ChatGPT, Gemini, Claude and the models behind Perplexity are examples of applications that run on an LLM. An LLM only knows what was in its training data, unless the system around it fetches current sources through search.
Read more: how AI search engines choose sources
AI Overviews
AI Overviews are AI-generated summaries that Google shows above or among the regular search results, with links to the sources the answer is based on. According to Google's documentation on AI features, they only appear when Google's systems judge that they add something to the classic results. To be eligible as a source, a page must be indexed and able to appear with a snippet in Google Search; Google lists no additional technical requirements.
Read more: how AI search engines choose sources
AI Mode
AI Mode is a Google search mode that gives a fully AI-generated answer, with links to supporting websites, in which you can ask follow-up questions. Google describes it as particularly useful for questions that call for further exploration, reasoning or complex comparisons. The same conditions apply to AI Mode as to AI Overviews: your page must be indexed and able to appear with a snippet (source).
Read more: how AI search engines choose sources
RAG (Retrieval-Augmented Generation)
RAG (Retrieval-Augmented Generation) is a technique in which an AI system first looks up relevant documents and then passes them to the language model as context, so the answer is based on current sources. Without RAG, an LLM draws only on its training data. With RAG, a page published today can already end up in an answer, provided the search bot can fetch it.
Read more: how AI search engines choose sources
Query fan-out
Query fan-out is a technique in which an AI search system splits one question into several related searches across subtopics and sources, and combines the results into a single answer. Google explicitly mentions this technique for AI Overviews and AI Mode. For you, it means a page can also serve as a source for a sub-question, not just for the original query.
Read more: how AI search engines choose sources
Grounding
Grounding is anchoring an AI answer in external, verifiable sources, such as search results, rather than only in what the model remembers from its training. Grounding reduces the chance of made-up information and makes it possible to show sources. Google uses the term for, among other things, supplying content from the Google Search index to Gemini at the moment someone asks a question.
Read more: how AI search engines choose sources
Citation
A citation is a source reference in an AI answer, usually a link or footnote to the page a claim is based on. Citations are to GEO what rankings are to SEO: the visible measure of success. You increase your chance of a citation with sentences that hold up without context, concrete facts with sources and a clear heading structure.
Read more: content that AI cites
Zero-click
A zero-click search is a search in which the user gets their answer directly in the search engine or AI assistant and doesn't click through to any result. AI Overviews and chat assistants reinforce this pattern, because the answer is already summarised on screen. That's why, in GEO, it matters not only whether you get clicks but also whether your brand is mentioned in the answer.
Read more: difference between GEO and SEO
AI crawler
An AI crawler is a bot that fetches web pages for an AI application, for example to train a model or to find current sources for an answer. The key distinction is between training bots, search bots and bots that visit a page on behalf of a user. Each bot has its own user agent, so you can decide per bot in robots.txt what it's allowed to do.
Read more: AI crawlers in robots.txt
GPTBot
GPTBot is OpenAI's crawler that collects content that may be used to train OpenAI's generative AI models. According to OpenAI's documentation on bots, a disallow for GPTBot indicates that your content may not be used for training. Blocking GPTBot doesn't remove you from ChatGPT's search feature; OAI-SearchBot determines that.
Read more: AI crawlers in robots.txt
OAI-SearchBot
OAI-SearchBot is OpenAI's crawler that finds and fetches websites so they can be shown in ChatGPT's search features. According to OpenAI, sites that block OAI-SearchBot are not shown in ChatGPT search answers, although they may still appear as a navigational link. The settings for OAI-SearchBot and GPTBot are independent: you can allow search and block training.
Read more: AI crawlers in robots.txt
Google-Extended
Google-Extended is a robots.txt token that lets you decide whether content Google crawls may be used to train future Gemini models and for grounding in Gemini applications. It isn't a separate crawler: crawling is done with Google's existing user agents. According to Google, the token has no effect on your inclusion in Google Search and is not a ranking signal. So it also doesn't control whether you appear in AI Overviews; that runs through Googlebot.
Read more: AI crawlers in robots.txt
Robots.txt
Robots.txt is a text file in the root of your website that tells crawlers which URLs they may and may not fetch. The protocol is set out in RFC 9309. Google stresses that robots.txt is not a way to keep a page out of the search results and not a security measure either, because not every crawler respects it. For GEO, it is mainly the file in which you control access for each AI bot.
Read more: AI crawlers in robots.txt
llms.txt
llms.txt is a proposed Markdown file in the root of your website that gives language models a concise overview of your site and points to the most important pages. The proposal comes from Jeremy Howard and is described on llmstxt.org. The only required element is an H1 with the site's name; ideally it's followed by a summary as a blockquote and H2 sections with lists of Markdown links. It is a proposal, not an official standard, so it's uncertain which AI systems actively use the file.
Read more: llms.txt
Schema markup
Schema markup is structured data in a page's code that uses the schema.org vocabulary to describe what is on the page, such as an organisation, article, product or FAQ. Search engines and AI systems then don't have to infer the meaning from the running text. Schema markup is no guarantee of visibility, but it makes facts about your company unambiguous and machine-readable.
Read more: schema markup for AI
JSON-LD
JSON-LD (JavaScript Object Notation for Linked Data) is a notation that lets you add schema markup to a page as a separate block in a <script type="application/ld+json"> tag. Because the data is separate from the visible HTML, JSON-LD is easier to maintain than markup woven through the text. Google recommends JSON-LD as the preferred format for structured data.
Read more: schema markup for AI
sameAs
sameAs is a schema.org property that points to other web pages that unambiguously describe the same entity, such as your LinkedIn page, Wikipedia article or company register listing. This connects your website to profiles that search engines and AI systems already know. That helps prevent confusion with companies or people of the same name.
Read more: schema markup for AI
Entity
An entity is a uniquely identifiable thing, such as a person, organisation, product or place, that a search engine or AI system can recognise independently of a specific search term. Where classic SEO often revolved around keywords, modern systems look at entities and the relationships between them. The more consistently your name, description and details appear across the web, the better an AI system recognises you as an entity.
Read more: schema markup for AI
E-E-A-T
E-E-A-T stands for Experience, Expertise, Authoritativeness and Trustworthiness: the quality characteristics Google uses in its quality rater guidelines to describe how trustworthy content is. It isn't a technical ranking signal you can switch on, but a framework for what good content is. On your site, you make E-E-A-T visible with signals such as an author, an about-us page, contact details, source citations and a publication date.
Read more: content that AI cites
Topical authority
Topical authority is the extent to which a website is seen as a reliable and complete source on a specific subject. You build it by covering a subject broadly and in depth, with pages that link to each other logically rather than standalone articles. A topic cluster is the usual structure for building topical authority.
Read more: content that AI cites
Server-side rendering
Server-side rendering (SSR) means the server delivers a page as complete HTML, so the text is already in the source code before any JavaScript runs. This matters for GEO because not every AI crawler runs JavaScript. If your content is only built in the browser, such a crawler may see an empty page. You can check this by looking at whether your text is in the raw HTML.
Read more: complete guide to GEO
Knowledge Graph
The Knowledge Graph is Google's knowledge base of facts about entities and the relationships between them, used among other things for the information panels next to the search results. A strong, consistent presence as an entity makes it more likely that Google recognises your organisation correctly. Other search engines and AI providers use similar knowledge structures, but these are less publicly documented.
Read more: schema markup for AI
Pillar page
A pillar page is a comprehensive overview page on a broad subject that briefly covers all subtopics and links through to in-depth articles. The pillar forms the centre of a topic cluster and usually targets a broad search query. In this knowledge base, the complete guide to GEO is the pillar page.
Topic cluster
A topic cluster is a group of closely related pages, made up of one pillar page and several cluster articles that each explore a subtopic in depth and link to each other. The consistent internal links show search engines and AI systems which pages belong together and which one is the main page. That way, every article contributes to the topical authority of the whole.