AI visibility, AEO and GEO, defined in plain words.
AI visibility is how often AI assistants such as ChatGPT, Perplexity, Claude and Google AI Overviews name your brand when people ask about your category. AEO and GEO are two names for the work of raising it. Every term below opens with a one-sentence definition, then the detail, a source and, for most, a full page of its own.
What is AI visibility, and what are AEO and GEO?
- AI visibilityAlso: AI search visibility
AI visibility is how often AI assistants such as ChatGPT, Perplexity, Claude and Google AI Overviews name your brand when people ask about your category.
It is measured per question and per engine, usually as a mention rate. Unlike a search ranking there is no fixed position: the same question can name you in one answer and leave you out of the next, which is why it has to be sampled, not checked once.
- AEOAnswer engine optimization
AEO is the work of getting your brand named, and your pages cited, in the answers AI assistants give.
In practice it means publishing pages that answer buyer questions directly, earning places on the third-party pages engines cite, and making sure AI crawlers can read your site. An AEO tool measures whether that work is moving the answers.
- GEOGenerative engine optimization
GEO is another name for the same work as AEO: improving how generative AI engines describe and recommend you.
The term comes from a 2023 research paper, GEO: Generative Engine Optimization (Aggarwal et al.). AEO and GEO are used interchangeably; neither is a separate discipline from the other. See what a GEO tool does.
How do AI engines build an answer?
- Large language model (LLM)
A large language model (LLM) is an AI model trained on a very large amount of text to predict and write language, and it is the part of an AI assistant that writes the answer.
Large language model (LLM): definition, measurement and examples
- Answer engine
An answer engine is a search tool that replies to a question with one written answer, usually with sources, instead of a ranked list of links.
- Generative engine
A generative engine is an AI system that finds sources for a question and writes a new answer from them with a large language model.
- Google AI Mode
AI Mode is Google's conversational AI search experience, where a question gets a written answer with links and can be followed with more questions.
- Retrieval-augmented generation (RAG)
Retrieval-augmented generation (RAG) is a method where a language model first retrieves relevant documents and then writes its answer with them in view.
Retrieval-augmented generation (RAG): definition, measurement and examples
- Grounding
Grounding is tying an AI model's answer to outside sources, such as live search results, so its claims rest on retrieved content rather than on memory alone.
- Training data vs live search
Training data is what a model learned before its knowledge cutoff; live search is what it looks up on the web while it answers, and one AI answer can draw on either or both.
Training data vs live search: definition, measurement and examples
How is AI visibility measured?
- Mention rate
Mention rate is the share of AI answers to a question that name your brand.
Named in 6 of 20 answers is a 30% mention rate. On its own that figure hides how uncertain it is, which is what a Wilson interval is for.
- Wilson intervalWilson score interval
A Wilson interval is the range a measured percentage could plausibly sit in, given how many answers it is based on.
Named in 6 of 20 answers is 30%, with a 95% Wilson interval of about 14.5% to 51.9%. Named in 60 of 200 is also 30%, but the interval narrows to about 24.1% to 36.7%. Same number, very different certainty.
It was introduced by E. B. Wilson in 1927 and is the standard interval for a proportion from a small sample (reference). Proofsource puts one on every percentage; see the methodology.
- PromptAlso: tracked question
In AI visibility, a prompt is a question tracked across AI engines, written the way a buyer would ask it.
"What's the best AI visibility tool for a small team?" is a prompt. Tools price by how many prompts you track, because each one is asked on every engine, every time the tool samples.
- Buyer questionAlso: unbranded question
A buyer question is a question about your category that does not name your brand, written the way someone choosing a product would ask it.
"Which electric SUV is best for a family of five?" is a buyer question. These are the real test of AI visibility, because the engine has to choose to name you. In the public Tesla example (Oct 4, 2026), 53 of 100 answers to 25 buyer questions named Tesla.
- Prompt coverageAlso: question coverage
Prompt coverage is how much of your set of tracked buyer questions AI engines answer with your brand named, broken down by question, topic and engine.
An average mention rate can hide a topic where no engine names you. In the Tesla example, Google AI Overviews named Tesla in 4 of 5 solar panel installation answers and Claude in none of the same 5.
- Recommendation
In an AI answer, a recommendation is a mention that endorses your brand, usually by putting it on the answer's shortlist or naming it as the pick for the buyer's situation.
Every recommendation is a mention, but not every mention is a recommendation: a brand named last, or named as the option to avoid, is mentioned without being recommended. Telling them apart takes reading the answer, which is why position and tone are read beside the mention rate.
- Named insteadAlso: competitors in the answer
The brands named instead are the other brands that appear in AI answers to your buyer questions, ranked by how often each is named.
They are your competitors as the engines see them, which is not always the list you would write. In the Tesla example, the competitors named most often across the same 25 buyer questions were Rivian and SunPower, in 13 answers each, then Ford in 10.
- AI answer sampling
AI answer sampling is asking the same questions on each AI engine more than once and reading every visibility figure as an estimate from that sample, not a fixed fact.
- AI answer volatility
AI answer volatility is how much the brands an AI engine names change when the same question is asked again, or asked in different words.
- AI visibility score
An AI visibility score is one number that sums up how often, and how prominently, AI engines name your brand across a fixed set of questions.
- Position in AI answers
Position in an AI answer is where your brand sits among the brands the answer names: first, third or last on its shortlist.
Position in AI answers: definition, measurement and examples
- Prompt volume
Prompt volume is an estimate of how often people ask AI assistants a given question or about a given topic.
- AI sentiment
AI sentiment is the tone an AI answer takes when it names your brand: positive, neutral or negative.
What do AI engines read and cite?
- Citation
A citation is a web page an AI engine links to or lists as a source for its answer.
Being named and being cited are different things. An engine can name your brand while citing a competitor's comparison page, or cite your page without naming you. In the public Tesla example (Oct 4, 2026), 100 answers to buyer questions cited 455 different websites, and fewer than 2 in every 100 citations went to tesla.com.
- Query fan-out
Query fan-out is when an AI engine splits one question into several related searches and builds its answer from all the results.
Google says both AI Overviews and AI Mode may issue multiple related searches across subtopics and data sources to build a response (Google Search Central). It is why a page can be cited for a question it never mentions: it answered one of the sub-searches.
- Google AI Overviews
Google AI Overviews are the AI-written summaries Google shows above some search results, with links to the pages they draw on.
Google says a page must be indexed and eligible to show with a snippet to appear in them (Google Search Central). They are one of the four engines Proofsource tracks.
- Source reach
Source reach is the share of AI answers that cite a given website at least once.
It is the companion to citation share. A site cited ten times in one long answer and a site cited once in each of ten answers have the same citation count, but very different reach.
- Third-party sources
Third-party sources are pages about your category written by someone other than you, such as reviews, roundups, forum threads, videos and competitors' sites, that AI engines read and cite.
- Listicles and roundups
A listicle, or roundup, is an article that lists and ranks several products for one need, such as the best project management tools for agencies.
Listicles and roundups: definition, measurement and examples
- Reddit citations
A Reddit citation is a link to a Reddit thread that an AI engine lists as a source for its answer.
- YouTube citations
A YouTube citation is a link to a YouTube video, often to a moment inside it, that an AI engine lists as a source for its answer.
What do AI answers do to your traffic?
- Zero-click search
A zero-click search is a search that ends without the person clicking through to any website, often because an answer was shown on the results page itself.
- AI referral traffic
AI referral traffic is the visits that reach your website from links in AI assistants' answers, such as ChatGPT, Perplexity, Claude or Copilot.
Does AI describe your brand correctly?
- Brand questionAlso: branded question
A brand question is a question that names your brand, such as "What is Tesla?" or "Is Tesla worth it?", used to check how AI describes you rather than whether it names you.
A brand question will mention you almost every time: in the Tesla example, 19 of 20 answers to brand questions named Tesla. That is why brand questions are tracked separately from buyer questions and read for accuracy, not counted as visibility.
- Entity confusionAlso: brand identity confusion
Entity confusion is when an AI answer about your brand describes, or opens with, a different company or thing that shares or resembles your name.
Even a famous name is not immune: in the Tesla example, Google AI Overviews answered "What is Tesla?" by opening with the magnetic unit of the same name. A newer brand whose name is close to a better-known one can be described as someone else entirely. One plain sentence on your site saying what you are, and what you are not, gives engines something to quote.
- Brand accuracy
Brand accuracy is whether AI answers describe your brand correctly: what you sell, who it is for, what it costs and how you compare.
- AI hallucination
A hallucination is a confident statement in an AI answer that is false or unsupported, such as a wrong price, a feature that does not exist or a source that was never written.
How do engines know who you are?
- Entity
In search and AI, an entity is one distinct, identifiable thing, such as a company, product, person or place, that a system can recognize and keep apart from things with similar names.
- Knowledge graph
A knowledge graph is a database of entities and the facts that connect them, and Google's Knowledge Graph is the best-known one, with millions of entries about people, places and things.
- Wikidata
Wikidata is a free, collaborative, multilingual knowledge base that supports Wikipedia and stores facts about entities as structured statements anyone can read and reuse.
- Schema markup (structured data)
Schema markup, or structured data, is code on a page, usually JSON-LD using the schema.org vocabulary, that states facts about the page and its subject in a form machines can read.
Schema markup (structured data): definition, measurement and examples
- E-E-A-T
E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness, the qualities Google says its systems look for in helpful, reliable content.
Can AI crawlers reach and read your site?
- Crawler access
Crawler access is whether the bots AI companies send can fetch and read your pages.
OpenAI, for example, runs separate bots for search (OAI-SearchBot), for training (GPTBot) and for fetching a page a user asked about (ChatGPT-User) (OpenAI docs). A bot can be blocked in robots.txt or, less visibly, at your CDN's edge by a bot-blocking setting. And a bot that gets through can still see an empty page if your text only appears after JavaScript runs.
- GPTBotOpenAI's training crawler
GPTBot is OpenAI's web crawler for collecting content that may be used to train OpenAI's generative AI foundation models.
Blocking it is a training decision, not a search one: ChatGPT search is decided by OAI-SearchBot, and OpenAI says each robots.txt setting is independent.
- OAI-SearchBotOpenAI's search crawler
OAI-SearchBot is OpenAI's crawler for surfacing websites in ChatGPT's search features, and sites that opt out of it are not shown in ChatGPT search answers.
- ClaudeBotAnthropic's training crawler
ClaudeBot is Anthropic's web crawler that collects web content that could contribute to training Anthropic's Claude models.
Anthropic runs two more bots with their own robots.txt names: Claude-SearchBot for search indexing and Claude-User for pages a user's question needs.
- PerplexityBotPerplexity's search crawler
PerplexityBot is Perplexity's crawler that indexes web pages so they can be surfaced and linked in Perplexity's search results.
- llms.txt
llms.txt is a proposed plain-text file at the root of a website that lists its most useful pages for language models.
It was proposed at llmstxt.org. It is cheap to add and harmless, but it is not a ranking lever: none of the four engines we track has said it uses llms.txt to decide what to name or cite, and Google states that you do not need new machine-readable files, AI text files or markup to appear in AI Overviews or AI Mode (Google Search Central).
Treat it as documentation for agents, not as optimization.
- robots.txt
robots.txt is a plain-text file at the root of a website that tells crawlers which paths they may fetch, under the Robots Exclusion Protocol (RFC 9309).
- llms-full.txt
llms-full.txt is an optional companion to llms.txt: one markdown file holding the full text of a site's key pages, so a model or agent can read them in a single fetch.
- Google-Extended
Google-Extended is a robots.txt token that controls whether content Google crawls may train future Gemini models and ground Gemini answers, without any effect on Google Search.
- Bingbot
Bingbot is the web crawler that builds Microsoft's Bing index, the index Microsoft Copilot searches when it grounds an answer in web results.
- Applebot-Extended
Applebot-Extended is a robots.txt token that lets a site opt out of Apple using its content to train Apple's foundation models, and it does not crawl anything itself.
- ChatGPT-User
ChatGPT-User is the agent OpenAI uses when ChatGPT or a custom GPT visits a web page because of something a user asked.
- Claude-User
Claude-User is the agent Anthropic uses to fetch a web page when a person asks Claude a question that needs it.
- Claude-SearchBot
Claude-SearchBot is Anthropic's crawler that reads the web to improve the quality and relevance of Claude's search results.
- Perplexity-User
Perplexity-User is the agent Perplexity uses to visit a web page when a user's question needs it, and Perplexity says it generally ignores robots.txt.
- CCBot
CCBot is the web crawler of Common Crawl, a nonprofit that publishes a free, open copy of the web that researchers and AI developers use as training data.
What people ask about AI visibility terms
Is AEO the same as GEO?
Yes, in practice. AEO (answer engine optimization) and GEO (generative engine optimization) both mean getting your brand named and your pages cited in AI answers. GEO comes from a 2023 research paper; AEO is more common in marketing.
What is a good AI mention rate?
There is no universal benchmark, because it depends on how crowded your category is. Compare your rate with the brands named most often for the same questions on the same engines, and read the interval: a rate from 20 answers can sit a long way from the true figure.
What is the difference between a mention and a citation?
A mention is your brand's name in the answer text. A citation is a web page the engine links to as a source. You can be mentioned without being cited, or cited without being mentioned, so track both.
Does llms.txt improve AI visibility?
There is no evidence that it does. Google says no special AI text files are needed to appear in AI Overviews or AI Mode, and none of the four engines Proofsource tracks has said it uses llms.txt for ranking. Readable pages and open crawler access matter far more.
See your own mention rate, with its interval.
See which buyer questions name your brand, who gets named instead, and what to fix. 100 answers. 25 Prompts/Question. Top AI engines. Free. No card.