A potential customer asks an AI tool to recommend a business like yours. Does your brand appear in the answer? If it does, is the description accurate? If it does not, which competitors are being suggested instead?
An AI visibility audit helps you answer those questions. It involves testing the searches your customers are likely to make across ChatGPT, Google AI Mode and Perplexity, then recording how each platform responds.
You do not need specialist software to run a useful first audit. A list of customer questions, access to the three platforms and a spreadsheet will give you a starting point.
Why Should You Audit Your Brand’s AI Visibility?
Customers can use AI tools throughout their buying journey. They may ask what is causing a problem, compare ways to solve it or request recommendations for providers. Your business could be relevant at any of those stages, even when the customer has not heard of your brand.
An audit shows whether you are appearing in those conversations. It can reveal that your business is regularly recommended for one service but absent when people ask about another. It can also uncover outdated information, such as an incorrect service area or a product you no longer offer.
Most importantly, it gives you something to measure. Without a baseline, it is difficult to tell whether changes to your website and wider online presence have improved how AI platforms represent your business.
AI answers are not fixed. Responses can vary depending on the wording of a question, the user’s location, the sources available and the platform’s behaviour at the time. Treat each answer as an observation and look for patterns across several searches.
In our client checks at Fly High Media, we have seen mentions increase month on month in smaller AI tools. That observation makes it worthwhile to revisit more than just the largest platforms, although the audit alone cannot tell us what caused the increase.
How Is an AI Visibility Audit Different From a Traditional SEO Audit?
A traditional SEO audit looks at how effectively your website can appear in search results. It may examine crawling, indexing, technical issues, content quality, internal links and organic performance.
An AI visibility audit focuses on what happens when someone asks a question. Does the response mention your brand? Is your website cited as a source? Does the platform understand which customers you serve and why someone might choose you?
You also need to look beyond your website. An AI answer may draw on reviews, directories, publications or other businesses’ content. If those sources describe your company incorrectly, the problem might not be obvious from an audit of your own pages.
The two types of audit support each other. A website that is difficult for search engines to access or understand may struggle to contribute useful information to AI answers. Equally, a technically sound website might still fail to answer the specific questions customers ask.
What Should You Review During an AI Visibility Audit?
For each question you test, record what appears on each platform. Keep three outcomes separate:
- Brand mention: Your business is named in the answer.
- Recommendation: Your business is presented as a relevant option for the customer.
- Citation: The answer links to a page on your website.
These outcomes do not always happen together. Your website might be cited without your brand being recommended, or your brand might appear without a link to your site.
You should also note which competitors appear, what the platform says about your business and which other websites it cites. Save the date and exact question you used. If you return later with different wording, you will not be making a reliable comparison.
A spreadsheet with one row per question and platform is enough. Useful columns include question, customer intent, platform, brand mentioned, brand recommended, your URL cited, competitors, other sources, accuracy and notes.
Step 1: Identify the Questions Your Potential Customers Ask
Start with your customers’ problems and decisions, rather than a list of your service names. Someone considering an SEO agency, for example, may first ask why their website traffic is falling or whether they need an agency at all.
When we choose the first questions to test for a client, we include their core services and, where relevant, the locations they serve. We also put ourselves in their customers’ shoes: what would I ask if I needed this product or service, but did not yet know which business to choose? That helps turn a list of target keywords into questions a potential customer might actually ask.
Your sales team, customer emails and enquiry forms can provide useful examples. Existing keyword research can also show how people describe their needs when they are exploring a topic or comparing options.
Include questions from different stages of the journey:
- Understanding a problem: “Why is my website getting visitors but no enquiries?”
- Exploring solutions: “What are the best ways to improve local search visibility?”
- Comparing options: “Should I hire an SEO agency or an in-house specialist?”
- Choosing a provider: “Which SEO agencies work with businesses in Manchester?”
Add questions about your locations, industries, products or specialisms where they reflect genuine customer needs. Include a few searches containing your brand name to check how accurately you are described, but make most of the list non-branded. That is where you can see whether customers might discover you for the first time.
Start with 15 to 25 questions. A focused list is easier to test consistently and act on than hundreds of prompts with no clear priority.
Step 2: Check Whether ChatGPT Mentions Your Brand
Enter each question into ChatGPT and record the response. Start a new conversation for each main question so earlier messages do not influence what comes next. Avoid mentioning your brand in a non-branded question, as that would change what you are testing.
For questions about current businesses or services, check whether ChatGPT has searched the web. If the answer includes citations, open them. Record whether your brand appears in the response, whether your website is linked and what reasons are given for recommending each business.
Look closely at the wording. “Here are several agencies you could consider” is different from a response that uses your article to explain a topic but recommends another provider. Both are useful findings, but they tell you different things about your visibility.
If you do appear, check whether the description matches your actual offer. A recommendation based on a service you do not provide could mislead a potential customer.
Repeat a few commercially important questions on a different day. One answer should not be treated as a permanent ranking. Patterns across repeated checks are more useful when deciding where to focus your ChatGPT optimisation efforts.
Step 3: Review Your Visibility in Google AI Mode
Ask the same questions in Google AI Mode. Note whether your business is named in the answer and whether any of your pages appear among the linked sources. Then try a natural follow-up question, such as asking for options in a particular location or for a specific type of customer.
It helps to understand how Google AI Mode works when interpreting these results. Keep what you see in AI Mode separate from conventional Google Search listings. Appearing in a standard search result does not tell you whether your business was included in the AI answer to that question.
Check whether the answer responds to the customer’s intent. If you offer a specialist service but AI Mode repeatedly suggests broader providers, investigate whether your website makes that specialism clear. If another site is linked for an explanation you could provide, review how thoroughly your own content answers the question.
Google Search Console can provide additional performance data for Google’s AI search features where reporting is available for your site. Use that information alongside your manual checks. Search Console can show broader trends, while your prompt testing shows what appeared for the specific customer questions you selected.
Step 4: Search for Your Brand and Services in Perplexity
Enter your question list into Perplexity and inspect both its answer and its citations. Open the linked pages to see which information supports the response.
Test non-branded questions first. Then ask a few branded questions, such as “What services does [brand] offer?” or “Is [brand] suitable for [specific need]?” These searches reveal whether Perplexity understands your business once a customer knows its name.
Record the exact pages it cites. A link to a detailed service page tells you something different from a link to a directory profile. If a competitor’s guide is repeatedly cited for questions your customers ask, read it and assess whether your own website addresses the same need.
Look for patterns across the results. If your brand is mentioned but another source is consistently used to explain your service, that may point to an opportunity to make your own pages more useful. Those findings can help set priorities for Perplexity optimisation.
Step 5: Record Which Competitors Are Being Recommended
For every question, note which other businesses appear and why the platform suggests them. After testing all three tools, you may find that the same competitors appear repeatedly for particular services or locations.
Check whether they are genuinely comparable to your business. An AI tool might suggest a national provider for a local search or include a company that serves a different audience. Mark those mismatches rather than assuming every recommendation is useful to the customer. Similar business names need an extra check: the platform may have confused your brand with a different company altogether.
For relevant competitors, review how they present the service in question. Do they have a dedicated page? Do they explain who it is suitable for? Are they featured in independent articles or directories that the AI platforms cite?
You are looking for gaps in how clearly your own business answers a customer’s question. You may find that your website explains what you do but gives little help to someone deciding whether it is right for them.
Step 6: Check Which Sources AI Platforms Are Citing
Group cited pages into three categories: your own website, competitors’ websites and independent sources. Independent sources might include industry publications, review sites or business directories.
For each citation, ask why that page is useful for the question. Does it provide a direct answer, explain a process, compare options or include details that are hard to find elsewhere? The aim is to understand what information the response uses, not simply to collect a list of URLs.
Check third-party pages that mention your business, too. An old directory listing could show the wrong address or describe services that have changed. If a platform cites that page, you have found a specific issue to address.
For local businesses, we pay particular attention to missing citations when a question names both the service and the location. We check whether the business has a useful page for that service and area and whether its relevant listings accurately describe what it offers. These are practical gaps to investigate, rather than a guarantee that adding a page or listing will secure a citation.
Remember that a citation and a recommendation are different. A platform may link to your useful guide while recommending another business. Record both outcomes so you can see whether your content is contributing to answers and whether your brand is being considered as a provider.
Step 7: Assess How Accurately AI Platforms Describe Your Business
Read every brand mention from a customer’s point of view. Could someone understand what you offer, where you operate and whether you are suitable for their needs?
Check factual details such as your services, products, locations, target customers and pricing approach. Watch for claims about credentials, results or guarantees that your business has not made. AI tools can also confuse similarly named businesses, so check that the answer is referring to the right company.
We saw this happen with a client whose name was very similar to another business in the same local area. ChatGPT described the other business’s services as though they were our client’s. We checked the sources it used to see whether any had attached the wrong description to my client’s name, but they had not. In that instance, the answer itself appeared to have mixed up the two businesses. Checking the sources helped us understand the error, but it did not make the answer accurate for the customer.
Label each description as accurate, partly accurate or inaccurate, with a brief explanation. Save a copy of any response containing a significant error. This makes it easier for your team to investigate the source of the problem.
Prioritise inaccuracies that could affect a customer’s decision. If an answer says you do not serve their area when you do, that deserves attention. Update information on sources you control and request corrections from third-party sites where appropriate. Retest later to see whether the answer changes, while recognising that a correction may not appear immediately.
How to Score Your Brand’s AI Visibility
A straightforward score can help you track changes without turning the audit into a complicated reporting exercise. Give each question a score on each platform:
| Score | What you observed |
|---|---|
| 0 | Your brand and website do not appear |
| 1 | Your website is cited, but your brand is not mentioned |
| 2 | Your brand is mentioned, but not suggested as an option |
| 3 | Your brand is recommended as a relevant option |
| 4 | Your brand is recommended, described accurately and linked to a relevant page |
If you test 20 questions, the maximum score is 80 for each platform. A total of 24 would give you a score of 30% for that set of questions.
This is a score you create for your own tracking. It is not an official visibility metric from any of the platforms. Keep notes about accuracy and competitors beside it, because the number alone will not tell you what to improve.
You can also calculate separate scores for different services or stages of the buying journey. A low overall score may hide strong visibility in one valuable area and a clear gap in another.
How Often Should You Audit Your AI Visibility?
Run a full audit to establish your baseline, then repeat it every quarter. That gives you time to make changes and check whether patterns have shifted.
If AI visibility is an active priority, review a smaller set of your most valuable questions monthly. You may also want to repeat checks after launching a new service, substantially rewriting key pages or finding an inaccurate description of your business.
Keep the same core questions and scoring rules each time. You can add new questions as customer needs change, but record when you added them so comparisons remain meaningful.
Improve Your Brand’s AI Visibility With Fly High Media
An AI visibility audit shows where your business appears, which competitors are being suggested and whether the information customers see is accurate. Those findings can guide improvements to your website content, business information and wider online presence.
Fly High Media’s AI SEO approach helps businesses understand how they appear across AI search and address the gaps that matter to their customers. If your audit reveals that your brand is missing from important conversations, we can help you decide where to focus next.
