Free AI visibility checker
See whether answer engines mention your brand when someone asks a question in your category, how they describe it, and whether they cite your site. One-off scan, no signup to read the page.
The scan runs server-side against your domain and the report arrives by email. One report, no marketing sequence.
What the report looks like
These cards use a made-up roaster, Northwind Coffee, to show the format: the prompt that was asked, how each engine described the brand, whether the tone was positive or neutral, and whether the brand's own site was cited as a source. Your report covers your category's prompts, not these.
Prompt
Best specialty coffee subscriptions for offices
"Northwind Coffee is frequently recommended for office subscriptions, with reviewers pointing to its roast-date transparency and flexible delivery schedule."
Prompt
Northwind Coffee vs other office coffee suppliers
"Northwind Coffee appears alongside several regional roasters. Sources differ on pricing, and no single supplier dominates the comparison results."
Prompt
Which coffee brands publish sourcing information?
"Northwind Coffee publishes origin and roast information per lot, which is cited in several buying guides covering sourcing transparency."
An AI visibility checker tests whether answer engines such as ChatGPT, Perplexity and Claude mention your brand when a user asks a question in your category, how they describe it, and whether they cite your website as a source. This one is free: enter a brand name and domain, preview the mention format, and get the full report by email. It is a one-off scan, not ongoing monitoring.
What this AI visibility checker does, and what it does not
The term "AI visibility" gets used for three different things: whether a model knows your brand, whether it names your brand in an answer, and whether it links to your site. This checker is about the second and third, for a small set of category prompts, at one point in time.
The check asks: when someone puts a question from your category to an answer engine, does the engine name this brand, and if it does, is the site cited as a source? It works from a brand name and a domain. The on-page preview shows the format the report uses, and the report itself is generated for the domain you enter and sent to your email. Three engines are covered: ChatGPT, Perplexity and Claude.
What it is not. It is not a monitor: it does not re-run the prompts every day or alert you when an answer changes. It is not a rank tracker for Google: AI Overviews and AI Mode are discussed below because the mechanics matter, but the report does not query them. And it is not an audit of your pages. It tells you what the engines say; the validators linked on this page tell you why.
Treat the output as a snapshot. A model that mentions you on Tuesday can drop you on Thursday because a retrieval agent fetched a competitor’s updated comparison page. That is the argument for checking, fixing and re-checking rather than reading a single result as the truth about your brand.
How the free AI visibility checker works
Six steps from a brand name to a fix list. The check itself takes a minute; the report follows by email.
- 1
Enter the brand name exactly as customers say it
Type the name people use in conversation, not the legal entity. "Acme" finds more than "Acme Technologies Private Limited". If the brand is known by two names, run the check twice. The domain field is optional but improves the report: the emailed scan runs against the domain, and without one the tool guesses brandname.com.
- 2
Run the check
Click Check Brand Mentions. The page shows which engines the report covers, ChatGPT, Perplexity and Claude, and then a preview of the mention-card format: the engine, the prompt used, the sentiment label, whether the site was cited, and the excerpt where the brand appears.
- 3
Read the first card, then unlock the rest
The first card is fully visible; the remaining cards are blurred behind an email gate. Enter a work email and the tool calls a report generator with your domain and email. Nothing else is collected and there is one email, the report.
- 4
Read the report in this order
Start with presence: is the brand named at all for the category prompts? Then citation: when it is named, is your site the source or is the model paraphrasing a third party? Then sentiment and context: what is the brand described as, and next to which alternatives? The table on this page maps each state to the fix.
- 5
Fix the mechanical causes before the editorial ones
If the brand is absent, check crawler access, rendering and structured data first with the validators linked below. Those are cheap and often the whole problem. Only then move to sources and off-site authority, which take weeks.
- 6
Re-check after each change, or move to monitoring
Answers shift as models retrieve fresh pages, so run the check again after a fix has been live for a couple of weeks. When you need the same prompts tracked weekly across engines and regions, the checker has done its job and the visibility tracker takes over.
How to read the result
Each mention card carries four signals: whether the brand is named, whether the site is cited, the sentiment label, and the excerpt with its prompt. The combinations below cover nearly every report. Read the row that matches, then jump to the next-steps section.
| Result | What it means | Where to start |
|---|---|---|
| Not mentioned | The engine answered the category prompt without naming the brand. Either its retrieval agent could not read your pages, or the sources it draws on do not associate the brand with the category. This is the most common result for brands under a few years old. | Mechanical checks first: robots.txt, rendering, structured data. Then the sources: are you in the comparison pages and communities the engine cites for this category? |
| Mentioned, not cited | The brand is named but the link, if any, points to a third party: a review site, a listicle, a forum thread. The model knows you from someone else’s page. That is good news for awareness and a risk for accuracy, because you do not control the description. | Make your own pages the better source: a clear answer-first page per category question, Organization and Product schema, and the facts the third party got wrong stated plainly on your site. |
| Mentioned and cited | The engine named the brand and linked your domain as a source. This is the target state for a prompt. Check which page was cited; often it is not the one you would have chosen, which tells you what the model finds quotable. | Protect it. Keep the cited page current, add the same structure to adjacent pages, and move to monitoring so you notice when the citation moves to a competitor. |
| Mentioned with competitors | The brand appears in a list of alternatives. Position in that list matters less than the framing: "best for X" next to your name is a win; "also available" is a shrug. The excerpt shows which. | Own a specific claim the competitors do not. Off-site authority work, comparison content that states the difference, and reviews that repeat the same claim in the same words. |
| Mentioned, negative or inaccurate | The sentiment label reads negative, or the excerpt states something false: an old price, a discontinued product, a founder who left. Models repeat whatever the most-cited source says, so this usually traces to one stale page somewhere. | Find the source page the engine cites, correct it at the source if you can, and publish a canonical fact page on your own domain that a retrieval agent can fetch and quote. |
Two cautions when reading. Sentiment is a label applied to a short excerpt, so "neutral" on a card that lists you among six alternatives is normal and not a problem to fix. And a citation to your domain is only useful if the cited page says what you want said; a cited pricing page from two years ago is a mention and a liability at the same time.
Why a brand is absent from AI answers
When the result is "not mentioned", the temptation is to write more content. Resist it until the mechanical causes are ruled out. They are faster to check, cheaper to fix, and responsible for a surprising share of blank results. Work down the list in order.
✓Retrieval agents are blocked in robots.txt
OAI-SearchBot and ChatGPT-User fetch pages for ChatGPT, PerplexityBot for Perplexity, ClaudeBot for Claude. A robots.txt that disallows them, often copied from a template that meant to block training crawlers, makes the site invisible at answer time. Training crawlers are a separate decision: GPTBot, CCBot and Google-Extended collect pages for model training, and blocking them does not stop retrieval. Paste your file into the robots.txt validator linked below; it names each agent and its status.
✓Content only exists after JavaScript runs
Retrieval agents fetch HTML and, in most documented cases, do not execute scripts the way a browser does. A product page whose name, price and description are rendered client-side reads as an empty shell to them. View the page source, not the inspector: if the facts are not in the raw HTML, server-render or pre-render them.
✓No structured data, or the wrong kind
Organization, Product, FAQPage and Article schema give a model unambiguous facts to quote: the brand name, what it makes, the price, the answer to a question. Pages without JSON-LD, or with JSON-LD that fails to parse, lose that advantage. Run the JSON-LD validator linked below on the pages you most want cited.
✓No llms.txt, or one that points at the wrong pages
llms.txt is a plain-text index of the pages you want a model to read first. No vendor has documented fetching it automatically, so it will not fix absence on its own, but it is the file you hand to any tool or assistant that asks for your site in one page, and it forces you to decide which pages matter. The llms.txt validator checks the format.
✓No cited sources associate the brand with the category
Answer engines lean on a recurring set of sources per category: review sites, comparison pages, forums, trade press. If none of them name your brand next to the category term, the model has nothing to retrieve. This is the editorial cause, and the slowest to fix; it is why it comes last on the list.
✓The brand name is ambiguous
A brand named after a common word or a place competes with everything else that word means. Models resolve ambiguity by context, so the fix is to pair the name with the category everywhere it appears: on the site, in schema, in third-party listings. "Acme inventory software" resolves; "Acme" does not.
Checker vs tracker vs report: which one you need
Three tools get bundled under "AI visibility", and people search for all three with the same words. They answer different questions on different timescales.
| Tool | Question it answers | Timescale | Cost |
|---|---|---|---|
| Checker (this page) | Does the engine name my brand for these prompts right now, and does it cite my site? | One-off. Minutes to run; results reflect this moment. | Free. Brand name and domain in, report by email. |
| Tracker | How is my share of AI answers moving, week over week, per engine, per region, against competitors, and which prompts did I gain or lose? | Continuous. The same prompt set re-run on a schedule, with history. | Product. The brand visibility tracker, linked below. |
| Report | What is the full picture for this domain today: presence, citations, sentiment, sources, and the fixes? | Snapshot. A document generated once, usually as a starting point or for a stakeholder. | The emailed output of this checker is a report of this kind. |
Use the checker when the question is "are we in the answer at all?" and the honest reply is that nobody has looked. Use the report to brief someone who was not in the room. Move to the tracker the first time a result changes and you cannot say when or why, because that is the moment a snapshot stops being enough.
The mistake to avoid is treating a checker as a tracker: running it monthly by hand, pasting results into a sheet, and then trying to explain a drop from memory. Two or three prompts checked by hand is fine. Fifty prompts across five engines and three regions is a job for software.
What "mentioned" means on each engine
The engines assemble answers differently, so the same brand can be present on one and absent on another for reasons that have nothing to do with content quality. The checker report covers ChatGPT, Perplexity and Claude; Google and Copilot are included here because the same mechanics decide the outcome, and the tracker covers them.
| Engine | Retrieval agent | What a mention looks like |
|---|---|---|
| ChatGPT | OAI-SearchBot builds the search index; ChatGPT-User fetches a page when a user’s question triggers browsing. | The brand is named in the prose of the answer. When search was used, sources appear as linked citations; a mention can occur without any link when the model answers from what it already knows. Both count as mentioned; only the linked case counts as cited. |
| Perplexity | PerplexityBot. | Nearly every answer carries numbered citations, so a mention is usually paired with a source. Read which source: if it is a review site rather than your domain, you are mentioned but not cited. Perplexity re-retrieves on every query, so results move fastest here. |
| Claude | ClaudeBot. | With web search on, Claude names brands and lists sources under the answer. Without it, the answer comes from training data and reflects how widely the brand was written about. A mention with no source is the second case. |
| Google AI Overviews and AI Mode | Googlebot. Google-Extended is a training control and, per Google, does not affect AI Overviews. | The brand appears in the overview text or in the link cards beside it. Because both are built from Google’s index, a page that ranks for the query has a chance and a page that does not has almost none. The same domain can be cited in an overview and absent from AI Mode’s longer synthesis. |
| Gemini | Googlebot for grounding. | Answers ground in Google Search when the app decides a query needs it, then show sources. A mention here tends to follow Google visibility for the same query. Gemini and AI Overviews can disagree, which is a sign the query is not settled. |
| Microsoft Copilot | Bingbot. | Copilot answers from the Bing index and shows numbered links. A brand invisible in Bing is invisible here regardless of how it performs in Google. Check Bing Webmaster Tools indexing before anything else. |
What to do next, by result
The checker tells you the state; these are the programmes that move each state. Each links to a page that walks through the work in detail.
✓Not mentioned: get cited by AI
Clear the mechanical list above, then build answer-first pages for the category questions and give models a reason to quote them. The "get cited by AI" solution page sets out the page structure, the schema and the prompt set to work from, in order.
✓Mentioned, not cited: make your site the source
The model knows the brand from someone else. Publish the definitive version of what that third party says, on your domain, with schema, and make it more quotable than theirs. Same programme as above, starting from step two.
✓Mentioned with competitors: off-site authority
Lists are won on the sources engines trust, not on your own pages. The off-site authority solution covers the review sites, comparison pages, communities and press that each engine cites for a category, and how to earn a place in them without paying for it.
✓Negative or inaccurate: PR and brand teams
A wrong or unflattering description traces to a source. The PR and brand teams solution covers finding it, correcting it, and publishing the canonical facts so the correction propagates through retrieval rather than waiting for the next model release.
✓Mentioned and cited: monitor
You are in the answer. The risk now is losing the slot without noticing. Move the prompts that matter into the visibility tracker so a change shows up as a delta with a date, not as a surprise in a sales call.
✓Everything at once: answer engine optimization
If more than one state applies across your prompts, which is usual, the answer engine optimization guide is the umbrella: how the engines retrieve, what they cite, and how to sequence the work so mechanical fixes land before editorial ones.
Who uses an AI visibility checker
Four roles run the same check for four different reasons.
✓Founders and marketing leads
The first look. A founder types the brand, sees "not mentioned" for the category question the sales team hears every week, and has a concrete problem to assign. The report is the artefact that turns a hunch into a line item.
✓SEO and AEO leads
A baseline before the work starts and a spot-check after each change ships. Pairs with the validators for a full pre-launch pass: robots.txt, rendering, JSON-LD, llms.txt, then the checker to see whether the engines noticed.
✓PR and brand teams
Watching for what the engines say, not whether they say it. The sentiment label and the excerpt matter more than the citation. An inaccurate description found here is a correction to make before a journalist pastes the same answer into a story.
✓Agencies
A free, repeatable first step in a pitch or an onboarding. Run the prospect’s brand and two competitors, put the three results side by side, and the case for the engagement writes itself. Once signed, the same prompts move into tracking.
Frequently asked questions
What does the AI visibility checker actually check?+
Whether ChatGPT, Perplexity and Claude name your brand when asked category questions, how the excerpt describes it, whether the sentiment reads positive, neutral or negative, and whether your website is cited as a source. You enter a brand name and an optional domain, preview the mention-card format on the page, and receive the full report for the domain by email. It is a single scan at one point in time.
Is the AI visibility checker free?+
Yes. There is no account and no card. The preview on the page needs only a brand name; the full report needs a work email so it can be sent to you, and that is the only thing collected. One report email is sent per request. Ongoing monitoring across engines, regions and competitors is a separate product, the brand visibility tracker, which is where the checker points once you need history.
Which AI engines does it cover?+
The report covers ChatGPT, Perplexity and Claude. Google AI Overviews, AI Mode, Gemini and Microsoft Copilot are described in the per-engine section because the same causes decide whether you appear on them, and because the brand visibility tracker monitors them. If your buyers are mostly on Google, read the AI Overviews row and check your ranking for the same queries first.
What is the difference between being mentioned and being cited?+
Mentioned means the engine wrote your brand name in the answer. Cited means it linked your domain as a source for that answer. A brand can be mentioned from a third-party page, in which case someone else controls the description, or cited from its own pages. Cited is the target state; mentioned without a citation is a prompt to make your own page the better source.
Why does ChatGPT not mention my brand?+
Usually one of four reasons, in order of how often they turn up: robots.txt blocks OAI-SearchBot or ChatGPT-User, the page content only exists after JavaScript runs, the pages carry no structured data, or no source ChatGPT trusts for the category names the brand at all. Check the first three with the validators on this page before spending on content, because they are fast to fix and often the whole cause.
How is an AI visibility checker different from an AI visibility tracker?+
A checker runs once and tells you the state now: mentioned or not, cited or not. A tracker runs the same prompt set on a schedule, keeps the history, and shows the change per engine, per region and against competitors. Use the checker to find out whether there is a problem, and the tracker when you need to know when a result moved and why. This page is the checker.
How accurate is a one-off AI visibility check?+
Accurate for the moment it ran, and for the prompts it used. Engines that retrieve live pages, Perplexity especially, can return a different answer an hour later if a source changed. Treat the result as a snapshot, re-run it after a fix has been live for a couple of weeks, and if you need the number to hold up in a report, use a tracker that averages across repeated runs.
Does blocking GPTBot stop ChatGPT from mentioning my brand?+
No. GPTBot is a training crawler; blocking it keeps your pages out of future model training. Retrieval at answer time uses OAI-SearchBot and ChatGPT-User, and those are what decide whether ChatGPT can read your site today. Many robots.txt files block all three from a template. Allow the two retrieval agents and decide about GPTBot separately; the robots.txt validator lists each one by name.
Can I check a competitor’s AI visibility?+
Yes. Enter their brand name and domain the same way. The report shows whether they are mentioned and cited for the same category prompts, which is the fastest way to see whether the gap is mechanical, such as they allow retrieval agents and you do not, or editorial, such as they appear in comparison pages you are missing from. Side-by-side results for two or three brands are a common starting point for an agency pitch.
What should I do if the AI answer is wrong about my brand?+
Find the source. Models repeat whatever the page they cite says, so an old price or a discontinued product usually traces to one stale third-party page or an outdated page on your own site. Correct it at the source where you can, then publish a canonical facts page on your domain with Organization and Product schema so retrieval agents have a current version to quote. Re-check after two weeks.