Blog · AEO glossary and AI search terminology
AEO Glossary: 60 AI Search Terms Every Marketer Should Know
Dharini Shah · September 23, 2026
An AEO glossary is a reference list of the terms used in answer engine optimization, the practice of making your content and brand easy for search engines and AI assistants to find, understand, quote and recommend. The 60 terms below cover the disciplines, the AI surfaces, how AI systems retrieve information, the crawlers you control, the content concepts that matter, and the metrics teams use to report progress.
Each definition is written to stand on its own, so you can share a single entry with a colleague or client without extra context. Where a term depends on a platform policy or a study, the entry links to the primary source. Terms are grouped by theme rather than alphabetically, because most people learn this vocabulary in clusters: first what AEO is, then where answers appear, then how they are produced and measured.
Core disciplines and concepts
These eight terms describe what the field is and why it exists. Most confusion in AI search conversations starts here, because vendors use overlapping labels for similar work.
1. Answer engine optimization (AEO). AEO is the practice of structuring content so a system can lift a direct, self-contained answer from it. The term originally described featured snippets and voice assistants and now includes AI summaries and chat assistants.
2. Generative engine optimization (GEO). GEO is the practice of improving how a brand is understood, represented and surfaced in AI-generated answers. The term comes from the GEO research paper presented at KDD 2024, which found that adding citations, quotations and statistics improved visibility by up to 40% on one metric, while keyword stuffing did not help.
3. Search engine optimization (SEO). SEO is the practice of improving visibility in organic search results. It remains the foundation for AI visibility: Google states that "there are no additional requirements to appear in AI Overviews or AI Mode" beyond its normal search guidance. For a comparison of the three disciplines, see AEO vs GEO vs SEO.
4. LLM optimization (LLMO) and AI SEO. LLMO and AI SEO are informal labels for the same broad goal as GEO: appearing accurately in answers from large language models. No standards body defines these terms, so check what a vendor actually means when they use them.
5. Answer engine. An answer engine is a system that responds to a question with a synthesized answer instead of only a list of links. ChatGPT, Perplexity, Claude, Gemini, Copilot and Google AI Mode all behave as answer engines.
6. AI search. AI search is any search experience where a model generates or summarizes the response, often with source links. It includes AI features inside traditional search engines and standalone assistants with web access.
7. AI visibility. AI visibility is how often and how favorably a brand appears in AI-generated answers for the questions its buyers ask. It is usually measured by sampling prompts rather than tracking a fixed position.
8. Zero-click search. A zero-click search is one where the user gets what they need on the results page and does not visit a website. A Pew Research Center analysis of 68,879 searches found users clicked a result on 8% of visits with a Google AI summary, compared with 15% without one.
AI search surfaces
These nine terms name the places where answers appear. Knowing the surface matters because each one retrieves sources differently and offers different controls.
9. AI Overviews. AI Overviews are AI-generated summaries that appear at the top of some Google results, with links to supporting pages. Ahrefs' 2026 benchmark found them on 21% of keywords studied, and far more often on long queries (46.4% for 7+ words) than one-word queries (9.5%).
10. AI Mode. AI Mode is Google's conversational search experience that answers complex questions and supports follow-ups. Google said at I/O 2026 that AI Mode passed 1 billion monthly users.
11. ChatGPT search. ChatGPT search is the web-connected answer mode inside ChatGPT. According to OpenAI, sites must allow OAI-SearchBot to be eligible, and placement is not guaranteed.
12. Perplexity. Perplexity is an answer engine that cites sources inline for most responses. Its crawler behavior is documented on its crawler page.
13. Microsoft Copilot. Copilot is Microsoft's AI assistant, which draws on Bing's index for web answers. Microsoft's Fabrice Canel has said that schema markup helps Microsoft's LLMs understand content.
14. Gemini. Gemini is Google's AI assistant and model family. Its use of your content for training is controlled separately from Google Search through the Google-Extended token.
15. Featured snippet. A featured snippet is a highlighted excerpt Google shows above or within organic results to answer a query directly. It was the original target of AEO, and Google documents it in its featured snippets guide.
16. People Also Ask (PAA). People Also Ask is a Google results feature that lists related questions with expandable answers. It is a useful source of question phrasing for answer-first content.
17. Voice search. Voice search is spoken querying through assistants and smart devices, which usually return a single answer. It rewards the same concise, direct answers that AEO emphasizes.
How AI systems find and generate answers
These eleven terms explain the mechanics behind an AI answer. Understanding them helps teams separate what they can influence from what they cannot.
18. Large language model (LLM). An LLM is a model trained on large amounts of text to predict and generate language. ChatGPT, Claude and Gemini are products built on LLMs.
19. Training data. Training data is the text a model learned from before release. What a model "knows" about your brand without searching comes from this data, which is why off-site mentions have long-term influence.
20. Knowledge cutoff. A knowledge cutoff is the date after which a model has no training data. Anything newer reaches the model only through retrieval.
21. Retrieval-augmented generation (RAG). RAG is a method where a system retrieves relevant documents and passes them to a model to ground its answer. The approach was formalized in a 2020 paper by Lewis et al. and underpins most AI search products.
22. Grounding. Grounding is tying a generated answer to retrieved sources so its claims can be traced. Grounded answers are the ones that show citations.
23. Query fan-out. Query fan-out is the technique of issuing several related searches across subtopics to build one answer. Google describes its use in AI Overviews and AI Mode, which means coverage of related subtopics matters.
24. Hallucination. A hallucination is a confident but false statement generated by a model. Ahrefs' 2026 benchmark reported that most AI models it tested repeated fabricated claims even when official sources contradicted them, which is why accuracy monitoring matters.
25. Embeddings. Embeddings are numerical representations of text that capture meaning, so systems can match a question to passages that use different words. They power semantic retrieval.
26. Semantic search. Semantic search is retrieval based on meaning rather than exact keyword matches. It is one reason clear topic coverage beats repeating a phrase.
27. Context window. A context window is the amount of text a model can consider at once. Retrieved passages compete for this space, which favors concise, information-dense sections.
28. AI agent. An AI agent is a system that takes actions for a user, such as browsing, comparing or booking. Google announced search agents and expanded agentic booking at I/O 2026.
Crawlers, access and technical terms
These twelve terms cover how AI systems reach your site. A single outdated robots.txt rule can remove a site from an AI search product, so this vocabulary is worth learning precisely.
29. robots.txt. robots.txt is a file at the root of a site that tells crawlers which paths they may access. It is the main lever for allowing or blocking individual AI crawlers.
30. AI crawler. An AI crawler is a bot operated by an AI company to fetch web content for search indexing, model training or user requests. Each purpose usually has its own user agent, and you can monitor them through server log analysis.
31. OAI-SearchBot. OAI-SearchBot is OpenAI's crawler for ChatGPT search. OpenAI's bot documentation states that "sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers."
32. GPTBot. GPTBot is OpenAI's crawler for collecting training data. Blocking it does not remove a site from ChatGPT search, which uses OAI-SearchBot.
33. ChatGPT-User. ChatGPT-User is the agent OpenAI uses when a person asks ChatGPT to visit a page. OpenAI notes that robots.txt rules may not apply to these user-initiated requests.
34. ClaudeBot and Claude-SearchBot. ClaudeBot collects training data for Anthropic, Claude-SearchBot indexes content for search, and Claude-User fetches pages on a user's request. According to Search Engine Land, all three respect robots.txt and need separate rules.
35. PerplexityBot and Perplexity-User. PerplexityBot surfaces and links sites in Perplexity and is not used for foundation model training. Perplexity-User handles user-initiated requests and generally ignores robots.txt.
36. Google-Extended. Google-Extended is a robots.txt token that controls whether content is used for Gemini model training. It does not control appearance in Google Search or AI Overviews, which are governed by Googlebot rules and snippet controls such as nosnippet and max-snippet.
37. llms.txt. llms.txt is a proposed Markdown file that summarizes a site for language models, introduced by Jeremy Howard in 2024. Google's John Mueller called it "purely speculative for now" in June 2026.
38. IndexNow. IndexNow is an open protocol that lets sites notify participating search engines, including Bing, when URLs are added, changed or deleted. Microsoft recommended it for keeping fresh content visible. See the IndexNow documentation.
39. Model Context Protocol (MCP). MCP is an open standard for connecting AI applications to external tools and data, introduced by Anthropic in November 2024. It joined the Linux Foundation's Agentic AI Foundation in December 2025.
40. WebMCP. WebMCP is a proposed browser standard that lets websites expose tools directly to AI agents. Chrome released an early preview in February 2026 with declarative and imperative APIs.
Content and authority terms
These ten terms describe what makes content easy for AI systems to trust and quote. They apply equally to classic SEO.
41. Answer-first writing. Answer-first writing places the direct answer in the opening sentences of a page or section, then adds detail. It makes passages extractable without surrounding context. The guide to writing extractable content covers the technique.
42. Extractable passage. An extractable passage is a section that makes sense when lifted out on its own. Avoiding phrases like "as mentioned above" keeps passages self-contained.
43. Entity. An entity is a distinct, identifiable thing such as a company, product, person or place. Search engines and models organize knowledge around entities rather than keywords.
44. Entity clarity. Entity clarity is how consistently and unambiguously the web describes what your company is, what it sells and who it serves. Inconsistent descriptions across your site, review profiles and directories tend to produce inconsistent AI answers.
45. Knowledge graph. A knowledge graph is a database of entities and the relationships between them. Google uses one to power knowledge panels and to understand what a query refers to.
46. Structured data (schema markup). Structured data is code that labels page content using a shared vocabulary, usually Schema.org. Its effect on AI citations is unclear: an Ahrefs study found no significant lift in AI Mode or ChatGPT after pages added schema.
47. JSON-LD. JSON-LD is the script format Google recommends for structured data. A February 2026 test found ChatGPT and Perplexity read an address placed only in invalid JSON-LD, which suggests they treat it as page text.
48. E-E-A-T. E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness. Google's helpful content guidance says it "isn't a specific ranking factor" and that "trust is most important."
49. Brand mention. A brand mention is any reference to your brand on another site, linked or not. Ahrefs' 75,000-brand study found branded web mentions correlated with AI Overview visibility at 0.664, versus 0.218 for backlinks, though correlation is not causation.
50. Content freshness. Content freshness is how recently a page was published or meaningfully updated. An Ahrefs study of about 17 million citations found AI-cited content was 25.7% fresher than organic results, with ChatGPT showing the strongest preference.
Measurement terms
These seven terms describe how teams report AI visibility. Because AI answers vary between runs, measurement relies on repeated sampling rather than fixed positions.
51. Prompt. A prompt is the question or instruction a user types into an AI assistant. Prompts tend to be longer and more specific than search keywords.
52. Prompt research. Prompt research is the process of identifying the questions buyers ask AI assistants, along with their intent and the brands those assistants recommend. It is the AI search counterpart of keyword research.
53. Visibility rate. Visibility rate is the percentage of sampled prompt runs in which your brand appears. SparkToro and Gumshoe research found less than a 1 in 100 chance of getting the same brand list twice, so percentages across many runs are more reliable than single "rankings."
54. Citation and citation rate. A citation is a link or named source an AI answer credits. Citation rate is the share of sampled answers that cite your pages. A brand can be mentioned without being cited, and the reverse.
55. Recommendation share. Recommendation share, sometimes called AI share of voice, is how often a brand is recommended relative to competitors across a prompt set. It shows competitive position more clearly than raw visibility alone.
56. AI sentiment. AI sentiment is the tone and accuracy of how an AI answer describes your brand. Positive visibility with outdated pricing or wrong positioning is still a problem worth fixing.
57. AI referral traffic. AI referral traffic is visits that arrive from links in AI answers. ChatGPT adds utm_source=chatgpt.com to referral URLs, which makes that traffic identifiable in analytics.
Commerce and emerging terms
These three terms describe AI systems that complete transactions, a newer area that affects ecommerce and service brands.
58. Agentic commerce. Agentic commerce is buying that happens inside an AI assistant or through an agent acting for the user. OpenAI's Instant Checkout launched in September 2025, and OpenAI states that product results are "organic and unsponsored."
59. Agentic Commerce Protocol (ACP). ACP is the open protocol OpenAI developed with Stripe to let merchants sell through ChatGPT. It underpins Instant Checkout.
60. Universal Commerce Protocol (UCP). UCP is Google's protocol for checkout inside AI Mode and the Gemini app, announced in January 2026 and co-developed with retailers including Shopify, Etsy, Target and Walmart.
How to use this glossary with your team
A glossary is most useful when a whole team agrees on it. The recommendations below are practical suggestions, not rules.
Pick one term for each concept. Decide whether your team says AEO, GEO or AI visibility, and use it consistently in reports and briefs. Mixed labels make stakeholders think you are running separate projects.
Separate training from retrieval. When a stakeholder asks why an AI assistant says something wrong about the brand, start by asking whether the answer came from training data or from retrieved sources. The fix differs: retrieval problems often respond to page and crawler changes within weeks, while training data reflects the broader web over a longer period.
Define metrics before you report them. Write down how you calculate visibility rate, citation rate and recommendation share, including which prompts, models and run counts you use. Without a written definition, month-over-month comparisons become unreliable.
Build your own brand glossary too. An industry glossary like this one explains the field. A brand glossary defines your own products, categories and differentiators so every page describes them the same way. The guide to building industry glossaries for AI search explains how to publish one.
Common mistakes with AEO terminology
Treating "AI ranking" as a real metric. AI recommendation lists change from run to run, so a single screenshot showing position one proves little. Report percentages across repeated samples.
Confusing training crawlers with search crawlers. Blocking GPTBot or ClaudeBot limits training use, while blocking OAI-SearchBot or Claude-SearchBot can remove you from those assistants' search answers. Many sites block the wrong one.
Assuming special files are required. Google says no new machine-readable files or AI text files are needed for its AI features. llms.txt and schema may have uses, but neither is a shortcut to visibility.
Using mention and citation interchangeably. A mention names your brand. A citation links to or credits your page as a source. Both matter, and they often move independently.
How Bob Builds AI helps
Bob Builds AI is an AEO and GEO platform and agency that turns much of this vocabulary into working reports. Visibility Monitoring tracks visibility rate, citation rate, recommendation share, citation sources and sentiment across ChatGPT, Gemini, Claude, Perplexity, Copilot and Google AI Overviews and AI Mode, measuring the real chat and search interfaces instead of raw model APIs. Prompt Research uncovers the questions customers ask AI, Brand Memory keeps your product facts and messaging consistent, and Agent Analytics shows how AI crawlers access your site. For context on how these terms fit into a broader search program, see why AEO is replacing parts of traditional SEO.
FAQ
What is the difference between AEO and GEO?
AEO focuses on structuring content so a system can extract a direct answer, which applies to featured snippets, voice assistants and AI summaries. GEO focuses on how a brand is represented across AI-generated answers, including off-site mentions, entity accuracy and recommendation share. The practical work overlaps heavily, and many teams use the terms interchangeably. What matters most is choosing one label and defining it clearly for your stakeholders.
What is the difference between a mention and a citation in AI search?
A mention is when an AI answer names your brand. A citation is when the answer credits or links to a specific page as a source. A brand can be mentioned based on training data without any citation, and a page can be cited for a fact without the brand being recommended. Tracking both separately shows whether a model knows about you and whether it trusts your pages as evidence.
What does query fan-out mean?
Query fan-out is a technique where an AI search system breaks one question into several related searches across subtopics, then combines the results into one answer. Google describes using it in AI Overviews and AI Mode. For marketers, it means a page can contribute to an answer by ranking for a sub-question, so covering related subtopics clearly matters as much as targeting one head term.
Which AI crawlers should I allow in robots.txt?
It depends on your goals. To appear in AI search answers, sites generally need to allow the search crawlers, such as OAI-SearchBot for ChatGPT search, Claude-SearchBot for Claude and PerplexityBot for Perplexity. Training crawlers, such as GPTBot, ClaudeBot and the Google-Extended token, are separate choices. Review each user agent individually rather than blocking all AI bots with one rule.
Is llms.txt an official standard?
No. llms.txt is a proposal published by Jeremy Howard in September 2024, not a formal standard. In June 2026, Google's John Mueller described it as "purely speculative for now" and said none of the AI systems use it. Some teams publish one as a low-cost experiment, but it should not replace crawl access, clear content and consistent brand information.
What is RAG and why does it matter for marketers?
Retrieval-augmented generation, or RAG, is a method where an AI system retrieves relevant documents and uses them to ground its answer. Most AI search products work this way. For marketers, it means current, crawlable, clearly written pages can shape answers even after a model's knowledge cutoff, because the model reads retrieved content at the time the question is asked.
How is visibility rate calculated?
Visibility rate is the percentage of sampled prompt runs in which your brand appears. A team defines a set of realistic buyer prompts, runs them repeatedly across the AI models its audience uses, and divides the runs that mention the brand by total runs. Repeated sampling matters because research from SparkToro and Gumshoe found AI brand recommendations vary substantially between runs.
Is E-E-A-T a ranking factor?
Google says E-E-A-T itself is not a specific ranking factor. It describes experience, expertise, authoritativeness and trustworthiness, qualities Google's systems try to identify using a mix of signals, with trust described as the most important. Google applies extra weight to these qualities for topics that affect health, finances or safety. The same qualities make content more credible for AI systems to cite.
Conclusion
AEO vocabulary falls into a few clear groups: the disciplines themselves, the surfaces where answers appear, the retrieval mechanics behind them, the crawlers you control, the content qualities that earn trust, and the metrics that show progress. Learning the terms in those clusters makes the field much easier to follow than memorizing an alphabetical list.
The practical value of a shared glossary is fewer mistakes: blocking the wrong crawler, reporting an AI "ranking" that will not repeat, or treating a mention as a citation. A good next step is to agree on definitions for your core metrics and check your robots.txt against the crawler names above. If you want to track those metrics across AI models in one place, Bob Builds AI can help you set up the monitoring and prioritize what to fix.