How to Measure Visibility in AI Search

You rank on Google, your reports look healthy, and yet a customer tells you they asked ChatGPT for a recommendation and your business never came up. That gap between traditional rankings and AI visibility is real, and right now most business owners have no way to measure it. This article gives you a practical method to track whether AI tools mention your business, and to understand what the numbers can and cannot tell you.

Why Traditional Rank Tracking Falls Short

Rank trackers were built for a world of ten blue links. They check your position for a keyword on a results page. AI answers do not have positions. A customer asks a conversational question, the AI generates a unique response, and your business is either mentioned in that response or it is not. There is no position 3 to climb to, and the same question asked twice can produce two different answers.

This means you cannot buy a standard tool, plug in keywords, and get an AI visibility score the way you get rankings. Anyone selling you a single definitive number for AI visibility is oversimplifying. What you can do is build a repeatable testing routine that shows trends over time: are you mentioned more often, less often, or not at all, and in what context.

Step 1: Build Your Prompt Set

Start with the questions your customers actually ask before buying. A good prompt set has 15 to 25 prompts across three types:

• Direct service questions. “Who is the best plumber in Austin?” or “What does local SEO cost for a small business?” These mirror high-intent searches.
• Comparison and recommendation questions. “Compare the top HVAC companies near me” or “Which CRM is best for a 10-person sales team?” AI answers lean heavily on comparisons.
• Problem-first questions. “My website gets traffic but no calls, what should I check?” These reveal whether your educational content gets cited.

Write the prompts exactly as a customer would phrase them, not as a marketer would. Include your service areas and business category so the tests reflect real buying situations.

Step 2: Choose Your Platforms

Test on the tools your customers actually use. For most businesses this means ChatGPT, Google’s AI Overviews, and at least one of Perplexity, Gemini, or Copilot. Google’s AI Overviews matter because they sit directly on top of traditional search results, while ChatGPT and Perplexity matter because customers use them as standalone advisors. You do not need to test everywhere. Two or three platforms, tested consistently, beats six platforms tested once.

Step 3: Run Repeatable Tests

Repeatability is what turns casual checking into measurement. Follow these rules:

• Ask the exact same prompts each round, word for word.
• Test monthly, on a fixed schedule, and log the date of every round.
• Use a clean session where possible, since personalization changes answers.
• Record the full answer or at least the relevant excerpt, not just your impression of it.

AI answers vary from run to run, so never judge from a single test. One mention proves little. A pattern across three monthly rounds tells you something real.

Step 4: Record Mentions and Citations

For each prompt, record four things in a simple spreadsheet:

• Mentioned? Is your business named in the answer, yes or no.
• Cited? Does the answer link to or reference your website or a specific page.
• Context. Are you recommended, listed neutrally among options, or mentioned negatively.
• Competitors. Which competitors appear, so you can see who owns the conversation.

Over time this gives you a mention rate per prompt category, which is the closest thing AI visibility has to a ranking. If your mention rate climbs after you publish better content and earn reviews, your work is paying off.

Limitations You Should Know

Be honest with yourself about what this method cannot do. AI models update quietly, so a drop in mentions may reflect a model change rather than anything you did. Personalization means your test results will differ from what a real customer sees. There is no official API or Search Console equivalent for most AI platforms, so all measurement is observational. And mention does not equal business: being named in an answer only matters if it leads to visits, calls, or enquiries, which is why you should always pair AI visibility tracking with a “How did you hear about us?” field on your enquiry forms.

Common Mistakes to Avoid

• Testing once and declaring victory or defeat. Single runs are noise. Trends across rounds are signal.
• Using marketer phrasing. “Best enterprise SaaS SEO solutions” is not how customers ask. Real phrasing gets real answers.
• Ignoring the competition. Knowing which rivals get cited tells you what kind of content and reputation the AI rewards.
• Chasing mentions without fixing the basics. If your website is vague and your reviews are thin, no measurement routine will fix that. Measurement tells you where you stand; content and reputation work moves the needle.

What to Track on One Page

Keep a monthly scorecard with five rows: mention rate across your prompt set, citation rate (answers linking to your site), competitor mention share, referral visits from AI platforms in your analytics, and enquiries attributed to AI discovery. Review it alongside your normal SEO reports. When mention rates rise in the same months you publish substantive content and earn genuine reviews, you have evidence your AI visibility work is compounding.

The Bottom Line

You cannot measure AI visibility the way you measure rankings, but you can measure it well enough to make decisions. A fixed prompt set, tested monthly across the platforms your customers use, recorded honestly with all its limitations, will tell you whether your business is becoming the name AI recommends or remaining invisible. That trend line is worth more than any single score.

Request an AI Search Visibility Review and we will assess how your business currently appears across AI search platforms and where the gaps are.