Guide

AI search glossary: GEO, AEO, query fan-out and more

Updated · 15 min read · By the SearchCrow team

This glossary explains, in plain English, the 40 or so terms a small business owner meets when reading about AI search: the optimisation labels (GEO, AEO, LLMO), how assistants find answers (grounding, retrieval, query fan-out), the crawlers that visit your website (GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, Google-Extended, Bingbot) and the ways visibility is measured (visibility score, share of voice). Where a definition describes how a platform works, it comes from that platform's own documentation, linked inline.

Key takeaways

  • GEO, AEO and LLMO are labels for the same goal: being named or cited when an AI assistant writes an answer. Google says that, for Google Search, this is still SEO.
  • Most assistants answer business questions by searching live (grounding), often splitting a question into several searches (query fan-out).
  • Search crawlers and training crawlers are different bots. Blocking a training crawler such as GPTBot does not remove you from ChatGPT's search answers; blocking OAI-SearchBot does.
  • Visibility score, share of voice and prompt tracking are measurement conventions used by tools, not figures the AI platforms publish.

Terms in this glossary: AEO, agentic checkout, AI Mode, AI Overviews, AI search optimisation, Bingbot, ChatGPT search, ChatGPT-User, citation, Claude-SearchBot, Claude-User, ClaudeBot, Copilot, Deep Search, entity, generative engine, GEO, Google Business Profile, Google-Extended, Googlebot, GPTBot, grounding, hallucination, IndexNow, JSON-LD, knowledge panel, LLM, llms.txt, LLMO, LocalBusiness, noindex and snippet controls, OAI-AdsBot, OAI-SearchBot, Perplexity-User, PerplexityBot, product feed, prompt tracking, query fan-out, RAG, robots.txt, schema.org, share of voice, shopping agents, structured data, top cited sources, training data, visibility score, zero-click search.

What do GEO, AEO and LLMO mean?

They are three names for making your business more likely to be used and named in AI-written answers; none is a separate discipline from good SEO.

Generative engine

A generative engine is a search tool that writes an answer from several web sources rather than showing a list of links. The term comes from the research paper that coined GEO, which describes generative engines as systems that "use generative models to gather and summarize information to answer user queries" (arXiv, 2023). ChatGPT search, Perplexity, Google's AI Mode and Copilot all fit this description.

GEO (generative engine optimisation)

GEO means improving how visible your content is inside generative engine answers. It was named in "GEO: Generative Engine Optimization" by Pranjal Aggarwal and colleagues, first posted in November 2023 and accepted to the KDD 2024 conference (arXiv, 2023). Microsoft now uses the term too: Bing's guidelines say GEO "focuses on content eligibility for grounding and reference in AI responses", and add that it "does not guarantee grounding or citations" (Microsoft Bing, 2026). Our GEO, AEO and SEO guide goes deeper.

AEO (answer engine optimisation)

AEO is an industry label for the same goal as GEO: being the source an "answer engine" draws on. Some people also use it to cover featured snippets and voice assistants. Google's guidance on AI features mentions both AEO and GEO and says that, from Google Search's perspective, optimising for generative AI search is still SEO (Google Search Central, 2026).

LLMO (large language model optimisation)

LLMO is another industry label, used mostly by agencies and tool vendors, for optimising your presence in answers from LLMs (large language models, the technology behind ChatGPT, Gemini and Claude). We found no AI platform that uses the term in its own documentation. Treat it as a synonym for GEO.

AI search optimisation

AI search optimisation is the plain-English umbrella for all of the above. For a small business it mostly means letting AI crawlers read your site, publishing clear facts (prices, areas, hours), keeping listings accurate and earning genuine mentions elsewhere. See AI search optimisation for small businesses and the AI visibility checklist.

How do AI assistants find their answers?

Most of them search the web at the moment a customer asks, then write an answer from what they find, with links to the sources.

Training data is the large body of text a model learned from before release; it is fixed at a cut-off date. Live search is when the assistant looks things up while answering. For business recommendations, live search matters more, because it reflects your current website and listings. The crawlers are separate too: OpenAI's GPTBot crawls content "that may be used in training", while OAI-SearchBot is "for search" (OpenAI, 2026).

Grounding

Grounding means supplying a model with current information, usually search results, at the moment it answers, so the answer is based on that material. Google defines it as a technique "used to improve the quality, accuracy, and freshness of AI responses" (Google Search Central, 2026). Google's developer documentation says Grounding with Google Search "connects the Gemini model to real-time web content" (Google AI Studio, 2026). Bing uses "grounding results" to describe pages Copilot draws on (Microsoft Bing, 2026).

Retrieval and RAG

Retrieval-augmented generation (RAG) is the technical name for retrieve-then-write: the system fetches relevant documents first, then the model writes its answer from them. Google says its AI features rely "on our core Search ranking systems to retrieve relevant, up-to-date web pages from our Search index" (Google Search Central, 2026). In practice, RAG and grounding describe the same idea from slightly different angles.

Query fan-out

Query fan-out is Google's name for splitting one question into several related searches. Google says AI Overviews and AI Mode may use it, "issuing multiple related searches across subtopics and data sources" to develop a response (Google Search Central, 2026). ChatGPT documents something similar without the name: it "typically rewrites your query into one or more targeted queries" and may send "additional, more specific queries" (OpenAI Help Center, 2026), and Perplexity says Pro Search "conducts multiple searches across the web" (Perplexity Help Center, 2026). Our guide to query fan-out explains what it means for you.

Citation

A citation is a link or numbered reference to a source an AI answer used. Perplexity says "every answer includes citations linking to the original sources" (Perplexity Help Center, 2026), and Anthropic says the same of Claude's web search (Anthropic, 2026). Being cited (your page is a source) and being named (your business is recommended in the text) are different, and worth tracking separately.

Entity

An entity is a distinct, identifiable thing: a business, person, place or product. Search engines try to understand entities, not just words. Google describes knowledge panels as appearing "when you search for entities" (Google Knowledge Panel Help, 2026), and Bing advises using "clear and consistent naming for people, organizations, products, and locations" because it "improves grounding visibility and citation accuracy" (Microsoft Bing, 2026). For you, that means using the same business name, address and phone number everywhere.

Hallucination

A hallucination is a confident but false statement from an AI model. OpenAI defines hallucinations as "plausible but false statements generated by language models" and says ChatGPT "also hallucinates", though less often in newer models (OpenAI, 2025). For a business, this can mean wrong opening hours, prices or services. The best defence is publishing those facts clearly in several places. See ChatGPT has wrong information about my business.

What are the main AI search products called?

These are the assistants and features your customers are most likely to use.

AI Overviews

AI Overviews are the AI-written summaries at the top of some Google results. Google describes them as "an AI-generated snapshot with key information and links to dig deeper", shown "when our systems determine that generative AI can be especially helpful", and warns that they "can and will make mistakes" (Google Search Help, 2026). To appear as a supporting link, a page "must be indexed and eligible to be shown in Google Search with a snippet" (Google Search Central, 2026). See Google AI Overviews and AI Mode for small businesses.

AI Mode

AI Mode is Google's conversational search experience, which Google calls its "most powerful AI search experience". It works by "dividing your question into subtopics and searching for each one simultaneously" and is available in the United Kingdom in English (Google Search Help, 2026). Google says it is "particularly helpful for queries where further exploration, reasoning, or complex comparisons are needed" (Google Search Central, 2026).

Deep Search is a research feature inside AI Mode, announced in May 2025. Google said it "uses the same query fan-out technique but taken to the next level" and "can issue hundreds of searches" to produce a fully cited report (Google, 2025). It is aimed at longer research questions rather than quick local look-ups.

ChatGPT search is ChatGPT's ability to search the web and answer with "links to relevant sources". It can use location to find local results, and OpenAI says it "sometimes partners with other search providers" (OpenAI Help Center, 2026). To be eligible, OpenAI says to allow OAI-SearchBot and make sure your host or CDN allows OpenAI's published IP addresses; "placement is not guaranteed". See how to get recommended by ChatGPT.

Perplexity

Perplexity is an answer engine built around search. It says it "searches the internet, gathering information from authoritative sources" and gives numbered citations with every answer (Perplexity Help Center, 2026). Its crawler is PerplexityBot (below). See how Perplexity chooses businesses.

Copilot

Microsoft Copilot sits on top of Bing. Microsoft's guidelines describe how Bing "discovers, crawls, indexes, evaluates, and surfaces content across Bing search experiences, Copilot, and grounding API results" (Microsoft Bing, 2026). Being indexed in Bing is the starting point. See Copilot and Bing for small businesses.

A zero-click search is one where the person gets what they need from the results page or AI answer and doesn't visit any website. Bing acknowledges this directly: content "may surface across Bing search experiences without a traditional click", and "a decline in clicks does not always indicate a loss of visibility" (Microsoft Bing, 2026). It is why being named in the answer matters as much as the click.

Which crawlers and website files should I know about?

A crawler (or bot) is software that visits web pages automatically; each AI company runs its own, and they do different jobs.

robots.txt

robots.txt is a plain text file at the root of your site (yoursite.co.uk/robots.txt). Google defines it as a file that "tells search engine crawlers which URLs the crawler can access on your site", and warns it "is not a mechanism for keeping a web page out of Google" (Google Search Central, 2026). Most AI crawlers below read it. See how to allow ChatGPT to crawl your website.

GPTBot

GPTBot is OpenAI's training crawler. It crawls "content that may be used in training our generative AI foundation models", and disallowing it "indicates a site's content should not be used in training" (OpenAI, 2026). Blocking GPTBot does not stop you appearing in ChatGPT search, because OpenAI says "each setting is independent of the others".

OAI-SearchBot

OAI-SearchBot is the crawler OpenAI uses "to surface websites in search results in ChatGPT's search features". Sites that opt out "will not be shown in ChatGPT search answers", though they can still appear as navigational links, and robots.txt changes take "~24 hours" to apply (OpenAI, 2026). If you want ChatGPT to recommend you, this is the one to allow.

ChatGPT-User

ChatGPT-User is the agent ChatGPT uses when a person's request needs it to visit a page. OpenAI says it "is not used for crawling the web in an automatic fashion", that "robots.txt rules may not apply" because a user started the action, and that it "is not used to determine whether content may appear in Search" (OpenAI, 2026).

OAI-AdsBot

OAI-AdsBot visits landing pages submitted as ads on ChatGPT, to check they meet OpenAI's policies. OpenAI says it "only visits pages submitted as ads" and its data is not used to train models (OpenAI, 2026). You'll only see it if you advertise on ChatGPT.

PerplexityBot and Perplexity-User

PerplexityBot is "designed to surface and link websites in search results on Perplexity" and "is not used to crawl content for AI foundation models". Perplexity-User fetches pages when a user asks something; because the user requested it, it "generally ignores robots.txt rules". Changes can take up to 24 hours to apply (Perplexity, 2026).

ClaudeBot, Claude-SearchBot and Claude-User

Anthropic runs three bots (Anthropic, 2026). ClaudeBot collects content that may be used for training; blocking it excludes your future content from training. Claude-SearchBot "navigates the web to improve search result quality"; blocking it stops your content being indexed for search. Claude-User fetches pages when someone asks Claude a question. Anthropic says its bots honour robots.txt. See how Claude chooses businesses.

Googlebot

Googlebot is Google's main crawler. There is no separate AI Overviews or AI Mode crawler: these features are part of Google Search, so Googlebot's access controls them, and you limit what appears with the usual snippet controls (Google Search Central, 2026).

Google-Extended

Google-Extended is a robots.txt token, not a separate crawler; Google says it "doesn't have a separate HTTP request user agent string". It controls whether content Google crawls can be used to train future Gemini models and for grounding in Gemini Apps and Vertex AI. It "does not impact a site's inclusion in Google Search nor is it used as a ranking signal" (Google Search Central, 2026). Blocking it may reduce Gemini's use of your content, so think before you do. See how Gemini chooses businesses.

Bingbot

Bingbot is Microsoft's "standard crawler" that "handles most of our crawling needs each day" (Microsoft Bing, 2026). Because Copilot relies on the Bing index, Bing lists "blocking Bingbot in your robots.txt file" among things to avoid (Microsoft Bing, 2026).

noindex and snippet controls

These are instructions in a page's code. Google says you limit what appears in AI features with nosnippet, data-nosnippet, max-snippet or noindex (Google Search Central, 2026). Bing adds that NOARCHIVE "prevents content from being used in Copilot responses and grounding results", and NOCACHE limits Copilot to the URL, title and snippet (Microsoft Bing, 2026). Check none is set by accident.

IndexNow

IndexNow is a way to tell Bing immediately when you add, update or remove a page. Bing says timely notifications "reduce outdated or incorrect URL references in Copilot responses and grounding results" (Microsoft Bing, 2026).

llms.txt

llms.txt is a proposal, first published by Jeremy Howard on 3 September 2024, for "a /llms.txt markdown file to websites to provide LLM-friendly content" with a short summary and links to key pages (llmstxt.org, 2026). Google says Google Search ignores these files, and lists them under things you don't need to do (Google Search Central, 2026). Optional, and low priority.

What is structured data, and which kinds matter for a local business?

Structured data is extra code on a page that states facts (name, address, hours) in a format machines read reliably.

Schema.org

Schema.org is the shared vocabulary most structured data uses. Its types include LocalBusiness, defined as "a particular physical business or branch of an organization", with more specific types beneath it such as Dentist and FoodEstablishment (Schema.org, 2026). Google says there is "no special schema.org structured data" needed for AI features, and that markup should match the visible text on the page (Google Search Central, 2026).

LocalBusiness markup

LocalBusiness markup tells Google your hours, departments and other details. Google's required properties are name and address; recommended ones include telephone, url, geo, openingHoursSpecification and priceRange. Google advises using "the most specific LocalBusiness sub-type possible" (Google Search Central, 2026). See our LocalBusiness schema guide.

JSON-LD

JSON-LD is the format Google's LocalBusiness examples use: a block of code placed in the page's HTML that sits apart from the visible content (Google Search Central, 2026). You only need to add it once, usually on your home or contact page.

Knowledge panel

A knowledge panel is the information box Google shows when you search for a known entity. Google says the information "comes from various sources across the web" and is "updated automatically as information changes on the web". For local businesses, Google says Business Profiles "look similar to knowledge panels" (Google Knowledge Panel Help, 2026).

Google Business Profile

A Google Business Profile is your free listing on Google Search and Maps. Google says local results "are mainly based on relevance, distance, and popularity", that "businesses with complete and accurate info are more likely to show up", and that "more reviews and positive ratings can help" (Google Business Profile Help, 2026). Google also says Business Profiles can help your services be visible in AI responses (Google Search Central, 2026). See our Google Business Profile and reviews guide.

How is AI visibility measured?

There is no official AI ranking report, so tools measure visibility by asking assistants real questions and recording who gets named.

Prompt tracking

Prompt tracking means asking AI assistants the same set of questions (prompts) on a schedule and recording the answers: whether you were named, who was named instead and which sources were cited. Because answers vary from run to run, tracking a fixed set over time tells you more than one screenshot. See how to check if ChatGPT recommends your business.

Visibility score

A visibility score summarises prompt tracking as one number, typically the share of tracked answers that named your business. Each tool calculates it in its own way, so scores from different tools are not comparable. See what an AI visibility score means.

Share of voice

Share of voice compares how often you are named with how often competitors are named across the same questions. For example (an illustration, not real data): if a rival appears in 6 of 10 answers and you appear in 3, they have twice your share. It shows who AI favours in your market, which is often more useful than your score alone. See AI competitor analysis.

Top cited sources

Top cited sources are the websites assistants link to most often when answering questions in your area: directories, review sites, local news, competitors' pages. If the same directory keeps appearing and you aren't listed on it, that's a concrete fix. See directories AI assistants cite in the UK.

What do shopping terms like "shopping agents" mean?

These terms matter if you sell products online.

Shopping agents

A shopping agent is an AI assistant that researches products, and sometimes buys them, on a customer's behalf. OpenAI says ChatGPT can show product options with prices and links, and that it considers "structured metadata from first-party and third-party providers" when choosing products (OpenAI Help Center, 2026). See AI shopping agents and your product pages.

Agentic checkout

Agentic checkout means the assistant completes the purchase for the customer. Google announced an "agentic checkout feature" for AI Mode in May 2025, always "with your guidance and oversight" (Google, 2025). OpenAI's Instant Checkout lets ChatGPT show a checkout inside ChatGPT while the merchant's systems process the order (OpenAI, 2026).

Product feed

A product feed is a regularly updated file listing your products with prices, stock and delivery options. OpenAI says merchants can share structured product data in a CSV or JSON feed so ChatGPT displays products accurately (OpenAI, 2026), and that Shopify merchants' product data is already integrated through Shopify Catalog (OpenAI Help Center, 2026). See ChatGPT shopping for small online shops.

What should I do with these terms this week?

You don't need to learn the jargon to act on it. Three things cover most of it:

  1. Open yoursite.co.uk/robots.txt and check it doesn't block OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot or Bingbot.
  2. Check your Google Business Profile for correct hours, services and areas, and make sure the same facts are on your website in plain text.
  3. Ask two or three assistants five questions a customer might ask, and note who gets named and which sites get cited.

How SearchCrow helps

SearchCrow does the third step for you. The free check asks ChatGPT and Perplexity five customer-style questions, checks your site the way AI crawlers see it and shows your visibility score and top three fixes in about 30 seconds. Paid plans track more questions, assistants and competitors over time, with share of voice and top cited sources.

FAQ

What is the difference between GEO and AEO? Very little. Generative engine optimisation (GEO) comes from a 2023 research paper and means improving how visible your content is inside AI-written answers. Answer engine optimisation (AEO) is an industry label for the same goal. Google's own guidance mentions both terms and says that, from Google Search's point of view, optimising for generative AI search is still SEO.

What is the difference between GPTBot and OAI-SearchBot? GPTBot crawls content that may be used to train OpenAI's models. OAI-SearchBot is used to show websites in ChatGPT's search answers. OpenAI says each setting is independent, so you can block GPTBot and still allow OAI-SearchBot. Sites that block OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links.

What is query fan-out? Query fan-out is Google's name for the way AI Mode and AI Overviews may split one question into several related searches across subtopics and data sources, then combine the results into one answer. Other assistants document similar behaviour: ChatGPT says it rewrites questions into one or more targeted queries, and Perplexity says Pro Search runs multiple searches.

What does grounding mean in AI search? Grounding means giving an AI model current information, such as search results, at the moment it answers, so the answer is based on that material rather than only on what the model learned in training. Google describes it as a technique to improve the quality, accuracy and freshness of AI responses. It is why your live website and listings matter.

What does share of voice mean in AI search? Share of voice is the proportion of AI answers, across a set of questions, that name your business compared with your competitors. For example, if you are named in three of ten answers and a rival in six, the rival has twice your share. It is a measurement convention used by tracking tools, not a figure any AI platform publishes.

Do I need an llms.txt file? No. llms.txt is a 2024 proposal for a Markdown file that gives AI tools a summary of your site and links to key pages. Google says Google Search ignores llms.txt files, and the OpenAI, Anthropic and Perplexity crawler pages we checked do not mention it. It is harmless to add, but fix crawler access, listings and clear service pages first.

Sources

Free check

Does ChatGPT recommend your business?

Free, in under a minute. You’ll see whether AI assistants mention you, who they recommend instead, and your top 3 fixes.

We’ll read your website for your business details, so there’s less to type.

Free, no card needed. We ask ChatGPT and Perplexity 5 questions your customers might ask, and check your website. We use your email to send you your results and a few short follow-ups about them, which you can stop with one click; see our privacy notice.