CheckAIVisibility

What is AI visibility?

Updated November 13, 2026

AI visibility is how often, and how prominently, AI assistants such as ChatGPT, Claude and Gemini name your brand, recommend it and link to pages about it when people ask questions in your category. It matters because more buyers now ask an assistant for a shortlist before they visit any website, and a brand that isn’t in the answer isn’t on the shortlist.

Want your own numbers?

The free check asks OpenAI 5 buyer-style questions about your category. No sign-up, about a minute. The full report is a one-time $9, no subscription.

Run the free check →

What is AI visibility, in plain terms?

Think of the question a buyer types into an assistant: “What’s the best scheduling tool for a small physio clinic?” The assistant writes an answer. Your AI visibility is what that answer says about you, across many such questions and many runs: whether you appear at all, how high you appear, and which web pages the assistant used to decide.

The term is young and tools define it slightly differently. Ahrefs describes it as how discoverable your brand is and how often your content gets referenced across AI platforms. Amplitude’s documentation defines its visibility score as the percentage of AI responses that mention your brand. Both come back to the same three things, which are worth keeping apart.

The three levels of AI visibility: mentioned, recommended, cited

1. Mentioned: the answer names you

The most basic level. The answer contains your brand name, whether in a list, a comparison or a passing note. Measured over many runs, this becomes a mention rate: the share of answers to a question set that name you. If you are not mentioned, nothing else matters yet.

2. Recommended: your position when the answer is a list

Many buyer questions get a list back: “here are five options”. Being named fifth with a caveat is not the same as being named first as the best fit. List position captures that. It is useful, but noisier than the mention rate, so treat a single position as a hint and an average over many runs as the signal. Also read the wording: an answer can mention you while recommending someone else, or describe your pricing or features wrongly.

3. Cited: the AI links to a page

When web search is on, assistants usually show the pages they used. A citation can be your own page, or a third party’s page: a review site, a “best X” listicle, a forum thread, a directory. Both count. A citation of a third-party page that recommends you still helps you, and a third-party page that lists your competitors and not you is the most direct fix you can find. Citations tell you which sources shape the answer, which is why they are the most actionable of the three.

Why AI visibility matters for your brand

Why AI visibility varies from run to run

Ask the same question twice and you will often get two different answers. Language models sample their wording, and with web search on, each run may search slightly different queries and read different pages. The effect is large. A SparkToro study with Gumshoe ran 12 brand-recommendation prompts roughly 60 to 100 times each on ChatGPT, Claude and Google’s AI. ChatGPT and Google’s AI returned the same list of brands less than once in 100 runs, and the same list in the same order far less often. How often a brand appeared across runs was much more stable than where it appeared.

The practical rule: never judge AI visibility from one answer. Ask each question more than once, report rates rather than single results, and compare like with like (same questions, same engines) when you re-check.

API answers vs the consumer apps

Most AI visibility tools ask questions through the official APIs rather than typing into the ChatGPT, Claude or Gemini apps. That is the only practical way to run many questions in a controlled way, but it is not the same as what a person sees in the app. Through the API the developer picks the model and the settings. OpenAI’s web search documentation, for example, lets the caller choose the model, how much search context it gets, an approximate user location and allowed or blocked domains. The consumer apps make those choices for the user and can add context from their account, so the same question can come back differently.

Treat API-based results as a consistent, repeatable sample of how the models answer, not a screenshot of every user’s screen. Whether API answers vary in the same way as the apps is still an open question; the SparkToro authors list it as one.

How to measure AI visibility

  1. Pick buyer questions. 10 to 20 questions a real customer would ask: “best X for Y”, “X vs Y”, “alternatives to X”, “is X worth it”. Avoid questions that contain your brand name, which only measure whether the model knows you exist.
  2. Run each question more than once on each engine. Record the full answer text and the cited links.
  3. Score the three levels. Mention rate per engine, average list position when you appear, and the list of cited pages, split into yours and third parties’.
  4. Note who else appears. The competitors named most often, and the pages cited when they are named, show you where to work.
  5. Re-check on the same set after you make changes, typically after several weeks.

If you want a single number, the AI visibility score page explains what usually goes into one and where one number can mislead.

How to improve AI visibility

Start from the cited pages, because they show what the engines actually read. Then work on your own site: let AI crawlers in (OpenAI’s bot documentation separates OAI-SearchBot, which controls appearance in ChatGPT search, from GPTBot, which covers training), keep your main description in server-rendered HTML, and write pages that answer buyer questions directly. In the research paper that defined GEO, adding statistics, quotations and source citations raised a page’s visibility in generative answers, while keyword stuffing did not. For step-by-step tactics, see how to get cited by ChatGPT and AEO vs GEO vs SEO.

Check your own AI visibility

You can run the steps above by hand, or start with our free AI visibility checker to see how your own site appears. The free check asks 5 buyer-style questions on OpenAI with web search, one run each, and checks your site’s AI readiness (robots.txt rules for the main AI crawlers, llms.txt, JSON-LD, title and meta description, text in the server HTML, sitemap.xml), with no sign-up. The one-time $9 full report asks 15 questions on OpenAI and Claude with web search plus Gemini from model knowledge, each twice, for 90 answers, and reports mention rate, list position and a 0–100 score per engine, with every cited page and one suggested action. It shows the full answer text, source links and the score formula. It does not test Perplexity, Google AI Overviews or the consumer apps, and no tool can guarantee a mention.

AI visibility FAQ

What does AI visibility mean?

AI visibility is how often and how prominently AI assistants such as ChatGPT, Claude and Gemini name your brand, recommend it and link to pages about it when people ask questions in your category.

Why does AI visibility matter for my brand?

Buyers increasingly ask an assistant for a shortlist before they visit any website. If the answer names three competitors and not you, you are out of that shortlist, and you won't see it in your analytics because no click happened.

How do I check my AI visibility?

Write the questions a buyer would ask, run each one several times on the engines your customers use, and record whether you are mentioned, where you sit in any list, and which pages are cited. Tools can automate this, but the method is the same.

Why do I get a different AI answer every time?

Language models sample their wording, and with web search on they may read different pages on each run. A SparkToro study found ChatGPT and Google’s AI returned the same brand list less than once in 100 runs, so measure rates across many runs.

Is AI visibility the same as SEO?

No, but they overlap. SEO measures ranked links on a results page. AI visibility measures presence inside written answers, which often draw on pages that don't rank in Google’s top 10. Good SEO helps, because assistants with web search still need to crawl your pages.

Related reading

See where AI answers place you

Start free with 5 questions on OpenAI. If you want the whole picture, the full report covers OpenAI, Claude and Gemini for a one-time $9.