In brief
An AI visibility audit checks whether AI engines name your business, describe it correctly and cite your pages when buyers ask. You can do one by hand in a day. Run 20 to 30 buyer questions across the main engines, record what they say, then check crawler access, structured data and outside sources to find the causes.
Key points
- A useful audit starts with the questions buyers really ask, not with your brand name.
- Record four things per answer, whether you are named, where, what is said, and which sources are cited.
- Checking the facts in each answer often finds more urgent problems than checking visibility.
- Crawler access, CDN settings and structured data are quick technical checks that explain many gaps.
- The sources AI engines cite show where outside work will count.
- A one-day audit is a snapshot, so repeat it or track it to see real trends.
The steps at a glance
- 1
Write 20 to 30 buyer prompts
List the questions buyers ask before they choose a business like yours, across services, locations and stages, mostly without your brand name.
- 2
Run them across the main engines
Ask each prompt in ChatGPT, Perplexity, Gemini, Claude and Google, in a private window, and paste each answer into a spreadsheet.
- 3
Score every answer
Record whether you are named, your position, the competitors named, any facts about you and every cited source.
- 4
Check the facts
Mark each claim about your business as correct, wrong, out of date or missing, and note which source it seems to come from.
- 5
Check crawler access and schema
Review robots.txt and CDN settings for AI search bots, then test key pages for Organization, LocalBusiness and Person markup.
- 6
Review outside sources
Look at the most cited domains and check your presence and details on each, including Business Profile, directories and review sites.
- 7
Pick five actions
Turn the findings into a short list of fixes, ordered by impact, with an owner and a date for each.
What is an AI visibility audit?
An AI visibility audit checks how AI engines talk about your business when buyers ask them questions. It covers whether you are named, what is said, which pages are cited, and why.
AI visibility is different from a Google ranking. There is no single position to check. An answer might name you first, bury you in a list, describe you wrongly or leave you out. An audit gathers enough answers to see the pattern and then looks for the causes.
You can do a good first audit by hand in a day with a spreadsheet. The steps below are the ones we use, minus the automation.
Step 1. Which questions should you test?
Start with 20 to 30 questions buyers ask before they choose a business like yours. Most should not include your brand name. Someone who does not know you yet will not type it.
Spread the questions across three groups.
- Discovery questions. "Best kitchen fitter in Leeds" or "which CRM suits a small agency".
- Comparison questions. "Is quartz or laminate better for a worktop" or "brand A versus brand B".
- Detail questions. "How much do dental implants cost in Dundee" or "do they offer finance".
Add two or three branded questions, such as "what is Example Windows like", to see how you are described. Pull wording from sales enquiries, People Also Ask and Reddit threads. The prompt research guide goes into more depth on choosing questions.
Step 2. How do you run the prompts across engines?
Ask each question in each engine and save the full answer. Cover the engines your buyers use. For most businesses that means ChatGPT, Perplexity, Gemini, Claude and Google.
A few habits make the results fairer.
- Use a private or logged-out window where you can, so your own history has less effect.
- Ask each question in a fresh chat, not as a follow-up.
- For Google, search the question and note whether an AI Overview appears and what it says.
- Save the date and engine next to each answer.
- If time allows, run your ten most important prompts twice, on different days.
Consumer apps personalise answers and change between runs. That is normal. It is why a one-day audit is a snapshot, not a verdict.
Step 3. How do you score each answer?
Record the same fields for every answer, so you can compare them. A simple spreadsheet works well.
| Field | What to record |
|---|---|
| Named | Yes or no |
| Position | First, second, third and so on, or not named |
| Recommended | Whether the answer suggests choosing you |
| Competitors | Every other business named, in order |
| What is said | The words used to describe you |
| Sources | Every page or site cited |
Once it is filled in, patterns jump out. You may be named in Perplexity and missing in ChatGPT. You may lead on one service and never appear for another. One competitor may appear in almost every answer. Each pattern points to a different cause.
Step 4. Are the facts about you correct?
Go through every answer that mentions you and check each claim. Mark it correct, wrong, out of date or missing.
This step often finds the most urgent problems. Old prices, closed branches, services you stopped offering and mixed-up competitors all turn up. A buyer who reads a wrong fact may never contact you to find out it is wrong.
Where the engine cites a source, open it and look for the wrong detail. Often it is an old page on your own site or a stale directory listing. The guide on when AI gets your business wrong covers how to trace and fix these.
Step 5. Can AI search crawlers reach your site?
Check this early, because a blocked crawler explains a lot. Open yourdomain.com/robots.txt and look for rules that block AI search bots.
OpenAI's crawler overview says sites that block OAI-SearchBot will not be shown in ChatGPT search answers. Perplexity and Anthropic have similar search bots. The AI crawlers and robots.txt guide lists them all.
Then check your CDN or firewall. Some hosting services offer settings that block AI bots before they reach robots.txt. If your key pages rely on heavy scripts to show their text, view them with scripts turned off to see what a simple crawler sees.
Step 6. Does your site describe who you are clearly?
Test your home page, about page, a service page and an article. Run each through the Schema Markup Validator and Google's Rich Results Test.
Look for four things.
- One Organization or LocalBusiness block with the same name, address and phone as your Business Profile.
- A fixed @id for the business, used the same way on every page.
- sameAs links to your real profiles and records.
- Named authors marked up as Person entities.
Google's guide to generative AI features says structured data is not required. It still helps every system that reads your pages identify the business correctly. The entity SEO guide explains why a clear identity matters.
Step 7. Which outside sources do AI engines rely on?
Sort the sources column from step 3 by how often each domain appears. The top ten domains tell you where the engines get their information in your topic.
For each one, check three things. Are you listed or mentioned there? Are your details correct? Are your competitors there when you are not? Directories, review sites, trade bodies, local press and Reddit threads are common. These are the places where outside work will count most.
Step 8. How do you turn findings into actions?
Pick five actions. More than that rarely gets done. Order them by impact and effort.
A typical first list looks like this.
- Correct the wrong facts at their source.
- Unblock any AI search crawlers.
- Fix or add Organization and author markup.
- Improve the page behind your weakest important prompt.
- Get listed, or fix your listing, on the most cited directory where you are missing.
Give each action an owner and a date. Then plan to repeat the audit in a month or two. Since answers vary, the second audit is what tells you whether things moved.
What can a free AI visibility checker tell you?
A free checker is useful for a first look. It cannot replace repeated tracking, because one run of any prompt is only a sample. Be wary of any free tool that gives you a single score without showing the answers behind it.
The most useful free checks are the ones that look at causes rather than results. Crawler access, structured data and entity clarity are all things you can check once and fix. Those fixes help every engine at once.
When you do use a free checker, ask what it actually ran. How many prompts, on which engines, and on what date? A score based on three questions asked once tells you very little. A short report that shows the real answers, with sources, tells you far more than a number on a dial.
How does the free entity audit help?
Our free entity audit covers step 6 for you. It crawls your key pages and rebuilds the entity graph your site publishes. It checks for stable @ids, connected schema nodes, name variants, verifiable identifiers and authors marked up as Person entities.
It returns scored findings. Each one says what was found, why it matters and how to fix it. That turns the most technical part of the audit into a short list you can hand to a developer.
How Axiom GEO helps
The entity audit is free to try at the public tool and is also built into the platform. AI visibility tracking runs your prompts on a schedule across ChatGPT, Perplexity, Gemini and Claude, and records mention, position, sentiment, competitors, recommendation and cited pages. That turns a one-day snapshot into a trend. On the Professional plan and above, answer accuracy checks every claim against a fact sheet you approve. AI crawler analytics reads your server logs to show which AI bots actually reach your pages.
Sources
These are the external pages this guide relies on.