How to Audit Your AI Visibility in 30 Minutes

You cannot fix what you have not measured. An AI visibility audit tells you exactly where you stand — which platforms mention you, how they describe you, and which competitors are capturing the recommendations you want. The entire audit takes 30 minutes if you follow this framework.

I have run this exact process for over 80 businesses. The average starting score was 14 out of 100. Most business owners were surprised — they assumed their SEO strength carried over to AI. It does not, as I explain in my post on GEO vs SEO. An audit gives you the real numbers. Then you can make real decisions.

What You Need Before You Start

Three things. A spreadsheet or notepad. Access to ChatGPT, Perplexity, and Google. A list of 15-20 queries your ideal customers would actually ask.

Build your query list first. Think about what a potential client would type into ChatGPT, not Google. AI queries are conversational. “Who is the best workers compensation attorney in Atlanta” is a typical AI query. “workers comp attorney Atlanta” is a typical Google search. Write both types. Include location-specific and category-specific variations.

Target 15 queries minimum. 20 is better. Fewer than 15 and your data is too thin to be reliable. More than 25 and you are overcomplicating your first audit. You can always expand later.

Step One: Test ChatGPT (10 Minutes)

Open ChatGPT with web search enabled. Enter each query one at a time. For each query, record three data points.

Appearance. Did your business appear in the response? Score 0 if absent, 1 if mentioned with caveats, 2 if mentioned favorably, 3 if you are the primary recommendation.

Description. What did ChatGPT say about you? Copy the exact text. You need this later to assess sentiment accuracy and messaging alignment.

Competitors. Which competitors appeared? Note their names. You will track these over time to measure relative position changes.

Work through all 15-20 queries. Stay focused. Do not chase tangents. Record the data and move to the next query. This should take about 10 minutes.

Step Two: Test Perplexity (10 Minutes)

Repeat the same process on Perplexity. Perplexity is valuable because it shows its sources. When it mentions a business, you can see exactly where it pulled that information from.

Record the same three data points — appearance score, description, and competitors. Add a fourth: source attribution. Which websites or pages did Perplexity cite when it mentioned your business or your competitors?

Source attribution is strategic gold. If Perplexity cites a competitor because of their Avvo profile, you know you need to strengthen yours. If it cites a press article, you know media coverage matters for your category. This data tells you exactly which corroborating sources to build.

As I detail in how ChatGPT decides which businesses to recommend, source corroboration is one of the four factors that determine AI recommendations. Perplexity shows you the sources directly. Use that information.

Step Three: Test Google AI Overviews (10 Minutes)

Google AI Overviews appear on roughly 42% of search results pages. Not every query will trigger one. Enter each query in Google and check whether an AI Overview appears at the top of the results.

If an AI Overview appears, check whether your business is mentioned. Record the same data points: appearance score, description, and competitors. Note whether Google cites your website directly or pulls information from a third-party source.

If no AI Overview appears for a given query, mark it as N/A and move on. The absence of an Overview is also useful data — it tells you which queries still route users to traditional search results where your SEO work applies.

Step Four: Calculate Your Score

Now you have raw data from three platforms. Time to calculate your baseline AI visibility score.

For each query on each platform, you have a 0-3 score. Add up all your scores. Divide by the maximum possible score (3 points times number of queries times number of platforms that returned results). Multiply by 100.

Example: 18 queries tested across 3 platforms. That is 54 total data points. Maximum score is 162. Your total is 24. Your AI visibility score is 14.8 out of 100.

I explain this scoring model in detail in my post on the AI visibility score. The number itself is less important than the baseline. Now you can measure improvement over time with a consistent methodology.

The Three Actions to Take Immediately

After your audit, you will have clarity on what to fix. Here are the three moves that produce the fastest results.

First, fix entity consistency. If your business name, category, or location varies across platforms, AI has trouble recognizing you. Standardize everything. Same name, same address, same service descriptions. Every directory, every profile, every page.

Second, build corroborating sources. Perplexity showed you which sources AI trusts. Go get listed or featured on those sources. If competitors get cited from Avvo, you need an Avvo profile. If industry associations appear as sources, join and get listed.

Third, publish specific content. Your audit revealed which query types you miss. Write content that directly answers those queries with specific, factual claims. Not marketing copy. Verifiable information AI can cite.

Then run the audit again in 30 days. Track your score. Repeat.

Frequently Asked Questions

How Often Should I Audit My AI Visibility?

Monthly for the first six months. Then quarterly once your score stabilizes above 50. AI platforms update frequently and competitors adjust too, so regular monitoring catches changes early.

What is a Good AI Visibility Score?

Above 50 means you are competitive. Above 70 means you are leading your category. Most businesses start between 10 and 25. The goal is consistent improvement, not a specific number.

Do I Need Paid Tools for This Audit?

No. The framework above uses free access to ChatGPT, Perplexity, and Google. Paid tools help you scale and automate tracking, but they are not required for a baseline audit.

Should I Include My Brand Name in the Test Queries?

No. Test queries should reflect what a potential customer who does not know you would ask. Branded queries measure brand awareness, not AI visibility. You want to see whether AI finds you organically.