An AI visibility audit checks whether AI search tools can reach your content, quote it easily and tell who stands behind it. This free tool audits a page for those things and scores it from 0 to 100. It does not track whether ChatGPT or Gemini mention your brand, but a built-in prompt sheet helps you check that by hand.
- What an AI visibility audit covers
- What this tool checks and scores
- How to run the audit and read the score
- Writing style check: what it is and is not
- How to check your brand in ChatGPT, Gemini and Perplexity
- How to improve your AI visibility
- AI visibility audit vs checker vs tracker
- llms.txt and AI crawlers
- AI visibility audit FAQ
- Related tools
What an AI visibility audit covers
AI visibility has three layers. A problem in any one of them can keep you out of AI answers, so it helps to know which layer you are testing.
Can AI crawlers and search bots fetch the page, and can they read it without running JavaScript? Blocked or empty pages cannot be used.
Is the page easy to lift an answer from? Clear headings, a direct answer near the top, lists, tables, dated facts and a visible author and brand help.
When someone asks an assistant about your topic, does it name you? This depends on the wider web as well as your site, and no tool can promise it.
This tool covers layers one and two on any URL you enter. For layer three, use the manual method further down this page. For a broader view of measuring results, see our guide on how to measure AI visibility.
What this tool checks and scores
The first tab, Audit a URL, fetches your page and runs these checks. The result is an overall score from 0 to 100 and a short list of the top fixes.
| Area | What it looks at |
|---|---|
| AI crawler access | Your robots.txt rules for GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, PerplexityBot, Google-Extended, Applebot-Extended, CCBot and Bingbot, plus the meta robots tag and X-Robots-Tag header |
| Readable without JavaScript | Whether the main text is present in the HTML your server sends |
| Answer-ready structure | Headings, headings written as questions, lists and tables, and a direct answer near the top of the page |
| Facts and freshness | Numbers and dates in the text, which give a model something specific to quote |
| Entity and trust | Organization or Person schema, sameAs links, a named author, about and contact pages, and HTTPS |
| Structured data | Which schema types are present on the page |
| llms.txt | Whether the site has one. See the note on llms.txt below for how much it matters |
A low score is a to-do list, not a verdict. If access is the weak area, fix that first, because nothing else matters when a bot is blocked. If structure is weak, restructure the page so each section answers one question in its first sentence. If trust is weak, add a named author, an about page and schema that ties your brand to its profiles.
How to run the audit and read the score
- Open the Audit a URL tab and enter the full address of one page. Start with your homepage, then try a key service or product page.
- Run the audit and wait a few seconds while the page and robots.txt are fetched.
- Read the overall score, then the findings in each area. Note which area looks weakest.
- Work through the top fixes in order. Each one says what to change on the page.
- Change the page, publish, and run the audit again to confirm the score moved.
The audit looks at one page at a time. A strong homepage does not tell you much about your blog or product pages, so audit the pages you most want quoted. To check a whole site for crawl and on-page problems, use the SEO audit.
Writing style check: what it is and is not
The second tab, Writing style check, is a heuristic. You paste text and it looks at wording and sentence rhythm for patterns that often show up in machine-written copy.
Treat it as an editing aid. It cannot prove who or what wrote a text, and it will sometimes flag human writing and miss machine writing. Google says it rewards helpful, original content however it is produced, and explains this in its guidance on generative AI content. So use the check to make copy plainer and more specific, not to chase a score.
How to check your brand in ChatGPT, Gemini and Perplexity
The third tab, Brand prompt check, writes ten prompts for your brand and category. It runs in your browser, so nothing you type is sent to us. You then paste the prompts into each assistant yourself. Use this method:
- Enter your brand and category details on the Brand prompt check tab.
- Copy the ten prompts it creates. They cover the same ground as the table below.
- Open a fresh chat in ChatGPT, Perplexity, Gemini and Google AI Mode, signed out if you can, so earlier chats do not bias the answer.
- Paste one prompt at a time. For each answer, record whether you were mentioned, whether your site was cited as a source, what tone was used and which other sources appeared.
- Use the checklist on the tab to log results with the date and the tool. Repeat monthly, because answers change.
| # | Prompt pattern | What it tests |
|---|---|---|
| 1 | What is [brand]? | Whether the assistant knows you exist |
| 2 | Who are the best [category] providers? | Whether you appear in a ranked list |
| 3 | Best [category] for [type of customer] | Fit with a specific audience |
| 4 | [brand] vs [competitor] | How you are compared |
| 5 | Is [brand] good? What do people say about it? | Sentiment and sources of reviews |
| 6 | How much does [brand] cost? | Accuracy of pricing or plan facts |
| 7 | [Question your customers ask] (answered on your site) | Whether your page is used as the answer |
| 8 | [Category] in [city] | Local presence |
| 9 | Alternatives to [competitor] | Whether you are suggested as an option |
| 10 | Who owns or runs [brand]? | Entity clarity: founder, location, history |
Look for wrong facts as well as missing mentions. If an assistant gets your address, price or services wrong, fix the page that should have the answer, and check that your Organization schema and about page agree with it. This works as a simple AI visibility checker you control, but it is manual and covers only the days you test. Results can differ between runs, so judge trends over several rounds.
How to improve your AI visibility
Work in three phases, and tie each action to a page change.
- Phase 1, open the door. Fix access. Check robots.txt for rules that block the bots you want, with the robots.txt tester. Make sure the main text is in the HTML, not added later by scripts, and that no stray noindex is present. Use the meta tag checker to see robots tags.
- Phase 2, make the page quotable. Put a two or three sentence direct answer under the main heading. Turn vague headings into the questions people ask. Use lists and tables for steps and comparisons. Add real numbers and a date to claims you can support, and show who wrote the page. The heading checker shows your outline.
- Phase 3, be a clear entity. Add Organization schema with sameAs links to your real profiles, keep your name, address and descriptions the same everywhere, and publish about and contact pages. Check the markup with the schema markup validator. Then earn mentions from sites your customers already trust, since assistants lean on the wider web for who is credible.
No tool or tactic guarantees a citation. These steps remove the common reasons a good page gets passed over. For help with the third layer, see our generative engine optimization services and the guide on GEO vs SEO.
AI visibility audit vs checker vs tracker
| Type | What it does | Where this tool fits |
|---|---|---|
| Audit | Examines your own pages and finds what to fix | This tool: layers one and two |
| Checker | Asks AI assistants questions and reports whether you appear, once | Done by hand with the Brand prompt check |
| Tracker | Repeats the questions over time and charts the change | Not offered. A monthly log from the prompt sheet is the manual version |
llms.txt and AI crawlers
The audit reports whether a site has an llms.txt file, but treat it as optional. To allow or block AI bots, edit robots.txt. To build or validate the file, use the llms.txt checker, and read what the llms.txt file is first. Each AI company documents its own crawlers, for example OpenAI's bot page and Google's crawler overview, and the names change, so check them before you write rules.
AI visibility audit FAQ
What is AI visibility?
AI visibility is how often and how accurately AI search tools such as ChatGPT, Gemini, Perplexity and Google AI Mode surface your brand or content in their answers. It depends on whether bots can read your pages, whether the content is easy to quote, and whether trusted sources across the web mention you.
Does llms.txt help AI visibility?
There is no confirmed effect. llms.txt is a proposal, Google Search does not use it, and no major AI company has said its crawler relies on it. A clean file is harmless and cheap, but crawler access, clear content and a strong brand presence matter far more.
How often should I run an AI visibility audit?
Run the page audit after any major content or template change, and then every quarter. Run the brand prompt check monthly, since AI answers change often. Keep a dated log so you can see whether your fixes helped rather than relying on a single result.
Is an AI visibility audit the same as SEO?
No, but they overlap. Both need pages that bots can crawl and content that answers real questions. AI search adds a focus on direct answers, clear entities and mentions on other sites. A page that follows SEO basics is a good start, not a guarantee.
Does this tool track mentions of my brand in AI answers?
No. The page audit checks your own pages, and the writing check works on text you paste. The Brand prompt check only writes the prompts and a checklist. You run them in the assistants yourself, so there is no automatic monitoring or share-of-voice figure.
Can the writing style check tell if AI wrote my text?
No. It looks for patterns that often appear in machine-written text and cannot prove authorship. Use it to spot stock phrases and flat rhythm, then edit for clarity and detail. Original facts and your own examples improve text more than any score.
