Last updated: November 2026
An AI visibility audit checks whether AI search tools such as ChatGPT, Google’s AI Overviews and AI Mode, Perplexity and Claude can reach your website, describe your business correctly, and name or cite you when buyers ask about what you sell. It has six parts: access, entity facts, prompt testing, citation sources, content and measurement.
Key takeaways
- An AI visibility audit, a GEO audit, an AI SEO audit and an LLM visibility audit are names for the same job.
- Start with access. A crawler that your robots.txt allows can still be turned away by a firewall or content delivery network.
- Test the questions buyers ask, run each one more than once, and record mentions and citations separately.
- Google’s own guide says you do not need llms.txt, special markup or content rewritten in chunks to appear in its AI features.
- Search Console now reports impressions from AI Overviews and AI Mode. Start measurement there, because it is your own data.
- Score the six parts from 0 to 2 and fix the lowest score first.
What is an AI visibility audit?
An AI visibility audit is a structured review of how your brand shows up in answers written by AI systems. A traditional SEO audit asks whether a page can rank in a list of links. An AI visibility audit asks three different questions:
- Can the systems reach you? Their crawlers have to be able to fetch your pages.
- Do they describe you correctly? The facts they repeat about your business have to be right and consistent.
- Do they name you? When a buyer asks for a provider, a product or an answer in your category, you want to be mentioned, cited or both.
The same work is sold under several names. GEO audit stands for generative engine optimization. AEO audit stands for answer engine optimization. AI SEO audit, AI search audit, AI search visibility audit and LLM visibility audit are used for the same thing. If a proposal uses one of these names, ask which of the six parts below it covers.

How is a GEO audit different from an SEO audit?
A GEO audit builds on an SEO audit. It does not replace it. Google’s guide to its AI search features says the foundations that help a page rank in regular Search are the same ones that make it eligible for AI features.
| SEO audit | GEO or AI visibility audit | |
|---|---|---|
| Main question | Can this page rank for this search? | Is this brand reached, described correctly and named in AI answers? |
| What you inspect | Pages, links, speed, indexing, keywords | Crawler access, entity facts, AI answers, cited sources |
| Unit of measurement | Rankings, clicks, impressions | Mention rate, citation rate, accuracy, AI impressions |
| Typical failure | Page not indexed or outranked | Crawler blocked, wrong facts repeated, competitors named instead |
| Where the data comes from | Search Console, crawl tools, rank trackers | Search Console’s generative AI report, server logs, repeated prompt tests |
If you have not had the first one done, start there. Our guide to a digital marketing audit covers the wider review, and our SEO and local SEO page explains the foundations.
Why run an AI visibility audit now?
Two things changed in 2026.
Buyers moved. In BrightLocal’s Local Consumer Review Survey 2026, the share of consumers using AI tools such as ChatGPT for local business recommendations rose from 6% in 2025 to 45% in 2026. That made AI tools the third most used source, behind Google and Facebook.

Measurement arrived. Google announced generative AI performance reports for Search Console in June 2026, and its help documentation says the report had reached all sites by August 31, 2026. A site owner can now see first-party numbers for AI Overviews and AI Mode instead of relying only on outside estimates.
What does an AI visibility audit check?
It checks six things, in this order. The order matters because each part depends on the one before it.
Part 1: Access. Can AI crawlers reach your site?
Nothing else counts if the crawlers are turned away. Check four layers.
Robots.txt. Look for rules that name the crawlers below, and for a blanket rule that blocks everything.
| Crawler | Operator | Purpose |
|---|---|---|
| OAI-SearchBot | OpenAI | Search results in ChatGPT |
| GPTBot | OpenAI | Model training |
| ChatGPT-User | OpenAI | Fetches a page when a user asks |
| Claude-SearchBot | Anthropic | Search results in Claude |
| ClaudeBot | Anthropic | Model training |
| Claude-User | Anthropic | Fetches a page when a user asks |
| PerplexityBot | Perplexity | Search results in Perplexity, not training |
| Perplexity-User | Perplexity | Fetches a page when a user asks |
| Googlebot | Google Search, including AI Overviews and AI Mode | |
| Google-Extended | Controls use of content in Gemini apps and Vertex AI |
The sources are each company’s own documentation: OpenAI, Anthropic, Perplexity and Google.
Two details are easy to miss. OpenAI says a site that opts out of OAI-SearchBot will not be shown in ChatGPT search answers, so blocking it to avoid training has the wrong effect. Training is controlled by GPTBot. And Google says Google-Extended does not affect inclusion or ranking in Search.
The firewall layer. Robots.txt is a published policy. Your content delivery network, web application firewall and host enforce their own rules, and those can challenge or refuse automated visitors that robots.txt allows. Ask your host or developer which bot protection settings are on, and whether verified search and AI crawlers are exempt.
Server logs. Logs are the proof. Search them for the crawler names above and look at the status codes. A 200 means the page was served. A run of 403, 429 or 503 responses means something turned the crawler away.
The Search Console setting. Google added a Search generative AI control under Settings in Search Console. Sites are included by default. If someone set it to exclude, your content stops appearing in AI Overviews and AI Mode. Google says the setting is not a ranking signal for the rest of Search and does not affect AI training.
Finally, load a key page with scripts turned off. If the main content is missing from the HTML your server sends, do not assume every crawler will wait for it to appear.
Part 2: Entity facts. Does every source describe you the same way?
AI systems assemble a description of your business from many places. If those places disagree, the answer can come out wrong or vague.
Write down the facts that must match: business name, a one-sentence description of what you do and for whom, locations or service area, the list of services, and contact details. Then compare them across:
- Your home page title and opening paragraph
- Your about page
- The Organization or LocalBusiness structured data on your site
- Your Google Business Profile
- Your LinkedIn company page
- Directory and review profiles in your industry
Google’s guide notes that Business Profiles and Merchant Center feeds can make your services and products visible in AI responses, so treat those as part of the audit and not as an afterthought.
Part 3: Prompt testing. What do AI tools say when buyers ask?
This is the part most people think of as the audit. Do it with a method, because AI answers change from one run to the next.
- Write 20 to 30 prompts in buyer language. Take them from sales calls, from long questions in Search Console and from comparisons buyers make. Include category questions (“who does X for Y”), problem questions (“how do I fix X”) and brand questions (“is [your company] good for X”).
- Run each prompt in the tools your buyers use. For most businesses that means ChatGPT, Google’s AI Mode and AI Overviews, Perplexity, Gemini, Claude and Copilot.
- Run each prompt three times per tool. One run proves little.
- Record five things per run: whether you were named, whether your site was linked, what was said about you, which competitors were named, and which sources were cited.
Then calculate two rates:
- Mention rate = runs in which your brand is named ÷ total runs
- Citation rate = runs in which your site is linked as a source ÷ total runs
Keep a third column for errors: wrong location, wrong services, an old description. Those go straight back to Part 2.
A caution from Google’s guide applies here. It tells site owners to be wary of third-party tools that claim to report metrics from inside its AI systems. Most tracking tools sample answers much as you would. Treat their numbers as estimates and keep your own log.
Part 4: Citation sources. Which pages do the answers lean on?
AI answers in a category tend to draw on a short list of sources. Find that list.
From your prompt log, count every cited domain and sort them into five groups: your own site, competitors, directories and review sites, publications, and forums. Then ask three questions.
- Which of your pages get cited? If they are all blog posts and none are service pages, your expertise is being used but your offer is not being seen.
- Which directories and review sites appear most? Check that your profiles there are complete and current.
- Which publications appear? Those are the places where an earned mention or a contributed article is worth the effort.
Part 5: Content. Can your pages answer the question directly?
Read your ten most important pages as a buyer with one question.
- Is the answer in the first two or three sentences, in plain words?
- Are the facts specific and dated, with a named source where one exists?
- Does the page contain something first-hand: your own data, process, results or examples?
- Is there a clear path from the answer to the service that solves the problem?
Google’s guide puts original, first-hand content at the center of its advice and says there is no need to break content into chunks or write it in a special way for AI. Our post on SEO for AI Overviews sets out a citation-ready content framework you can apply page by page. If you add question and answer sections, our FAQ schema generator builds the markup.
Part 6: Measurement. Can you see change over time?
Set up four measures before you change anything, so you have a baseline.
- Search Console’s generative AI report. Google’s help page says it covers AI Overviews and AI Mode and reports impressions, which you can group by page, country, date or device. It may not appear for sites with very few AI impressions.
- Referral traffic from AI tools. In your analytics, look for visits that arrive from the AI tools’ own domains.
- Your prompt log. Rerun the same prompts on a schedule and track mention rate and citation rate.
- Lead source. Add AI tools as an option wherever you ask new leads how they found you.
How do you score an AI visibility audit?
Score each part 0, 1 or 2 and add them up for a total out of 12. This also works as an AEO audit checklist.
| Part | 0 | 1 | 2 |
|---|---|---|---|
| 1. Access | A search or AI crawler is blocked, or status is unknown | Robots.txt allows crawlers; firewall and logs not checked | Logs confirm crawlers receive pages; Search Console setting is on include |
| 2. Entity facts | Sources disagree on what the business is or does | Core facts match; some profiles are stale | Every source matches, including structured data and profiles |
| 3. Prompt testing | Never tested | Tested once, no repeat runs | Prompt log with repeat runs, mention rate and citation rate |
| 4. Citation sources | Unknown | Cited domains listed | Cited domains listed, gaps identified, plan to close them |
| 5. Content | Key pages do not answer their main question directly | Some pages answer directly; little first-hand material | Key pages answer directly with first-hand data and a path to the service |
| 6. Measurement | Nothing tracked | Search Console report checked occasionally | All four measures tracked on a schedule |
A total of 10 to 12 means the foundations are in place and the work is to keep improving. A total of 6 to 9 means there are specific gaps to close. Below 6, start with Part 1 and work down, because the later parts cannot score well until the earlier ones do.
What can you skip?
Several popular tactics are not supported by what Google has published. Skip them, or at least do not count them as progress.
- Treating llms.txt as a lever. Google’s guide says special AI files are not needed and that “Google Search itself doesn’t use them.” Ahrefs studied 137,000 domains and found that 97% of the llms.txt files on them received no requests at all in May 2026. We publish one ourselves because it costs nothing to keep. We do not count it as a lever.
- Rewriting pages into chunks for AI. Google says you do not need to.
- Adding schema “for AI.” Google says structured data is not required for its AI features. Keep your structured data accurate for the regular search features it supports.
- Trusting any tool that claims inside knowledge. Google’s guide warns against tools that claim to report internal metrics from its AI systems.
- Building separate pages for AI crawlers. Showing crawlers something different from what people see is cloaking, which Google’s spam policies prohibit.

How often should you run an AI visibility audit?
Run the full audit once or twice a year, and again after any event that can break access or change your facts: a site migration, a new host or firewall, a rebrand, a new service line or a new location.
Between audits, keep three light routines. Rerun the prompt log monthly. Check the Search Console generative AI report monthly. Review server logs for the main crawlers each quarter.
What is the difference between a free audit and a full AI visibility audit?
A free scan checks what a tool can read from a page. A full audit adds the work a tool cannot do.
Our free website audit scans more than 70 factors in under 60 seconds, including page speed, mobile experience, schema and AEO readiness, and gives you a report to download. It is the right first step and it covers pieces of Parts 1 and 5.
A full AI visibility audit adds firewall and log checks, the entity comparison, the prompt log with repeat runs, the citation source map and the measurement setup. It ends with a scored list of fixes in priority order. That is the work behind our content and AI visibility service, and it starts with a growth assessment at no charge.
Frequently asked questions
What is a GEO audit?
A GEO audit is a review of how a brand appears in answers generated by AI systems. GEO stands for generative engine optimization. It covers the same ground as an AI visibility audit: crawler access, entity facts, prompt testing, citation sources, content and measurement.
What is the difference between a GEO audit and an SEO audit?
An SEO audit checks whether pages can rank in a list of search results. A GEO audit checks whether AI systems can reach the site, describe the business correctly and name or cite it in an answer. The GEO audit depends on the SEO foundations, so the two are done together or in that order.
How do I check the AI visibility of my brand?
Write 20 to 30 questions your buyers ask, run each one several times in the AI tools they use, and record whether you are named, whether your site is cited and what is said. Then check Search Console’s generative AI report for impressions from AI Overviews and AI Mode.
Do I need an llms.txt file?
No. Google says its Search systems do not use it, and Ahrefs found that almost all llms.txt files receive no requests. Publishing one does no harm. It should not be counted as an improvement to AI visibility.
Should I block AI crawlers?
That depends on what you want. Blocking a training crawler such as GPTBot or ClaudeBot tells that company not to collect your content for model training. Blocking a search crawler such as OAI-SearchBot or PerplexityBot can remove you from that tool’s search answers. If your goal is to be named and cited, allow the search crawlers and check that your firewall agrees.
Does schema markup help AI visibility?
Google says structured data is not required for its AI features. It is still worth keeping accurate, because it powers regular search features and because wrong structured data can create conflicting entity facts.
Which AI tools should an audit cover?
Cover the tools your buyers use. For most businesses that means Google’s AI Overviews and AI Mode, ChatGPT, Perplexity, Gemini, Claude and Copilot. If you sell locally, include a few prompts with your town or region in them, because Google’s AI features can draw on Business Profile data. Our guides to SEO and AEO for local businesses, SEO for contractors and SEO for landscapers cover the local side.
The bottom line
An AI visibility audit is not a mystery and it is not a tool subscription. It is six checks in a fixed order: access, entity facts, prompt testing, citation sources, content and measurement. Score them, fix the lowest one, and measure again.
If you want a quick read on the technical side first, run the free website audit. If you want the full audit done with you, book a growth assessment. Strategy means nothing without execution. We do both.
How we know: crawler names and purposes are from OpenAI, Anthropic, Perplexity and Google documentation, checked November 17, 2026. Statements about what Google’s AI features need are from Google’s AI optimization guide, last updated July 10, 2026. Search Console details are from Google’s Search Central blog and help pages. Consumer figures are from BrightLocal’s Local Consumer Review Survey 2026. The llms.txt figure is from Ahrefs.