AI search visibility: how to measure it without fooling yourself

AI search visibility is how often your firm is named or cited when people ask AI tools the questions your buyers ask. You measure it with four sources: a fixed panel of buyer questions tested monthly across ChatGPT, Google AI Mode and others; Bing Webmaster Tools' AI Performance report; referral traffic tagged from ChatGPT; and, most important, what new clients tell you about how they found you. No single score captures it, and any tool that claims to should be treated with care.

I say that as someone who wants a clean number as much as anyone. The problem is that AI answers change every time you ask.

Why is AI search visibility hard to measure?

Three reasons.

Answers are not stable. Ahrefs asked ChatGPT the same prompts repeatedly and found less than a 1 in 100 chance of getting the same list of brands twice. A single check tells you almost nothing.

Google does not separate it out. Google's documentation on AI features says traffic from AI Overviews and AI Mode is included in the "Web" search type in Search Console's Performance report. You cannot filter it on its own.

People see you without clicking. Pew Research found that when a Google AI summary appeared, users clicked a link inside it in just 1% of visits. If your name is in the summary, that exposure mostly will not appear in your analytics.

So you have to measure it the way you would measure word of mouth: from several angles, with honest caveats.

What should you actually measure?

The first four tell you whether you are visible. The last two tell you whether it is worth anything.

How do you build a prompt panel?

This is the core of AI search visibility tracking, and you can do it with a spreadsheet.

1. Write 20 to 40 real buyer questions. Take them from enquiry emails, sales calls and Search Console queries. Mix broad ("best employment solicitors in Bristol") with specific ("can my employer change my contract without consent?").
2. Fix the list. If you change the questions every month you cannot compare.
3. Run each question in each tool. ChatGPT with search on, Google AI Mode, Google search (to see if an AI Overview appears), Perplexity and Copilot.
4. Run each more than once. Because answers vary, three runs per question per tool gives a truer picture than one.
5. Record three things. Were you named? Was your site cited? Who else was named?
6. Calculate share of answers. Times you were named, divided by total answers. Track it by tool and by service line.

Tracking tools automate this at larger scale, and several are decent. But I would build the manual version first. You learn more from reading 60 AI answers about your market than from any dashboard.

What free data do the platforms give you?

Bing Webmaster Tools

Microsoft launched an AI Performance report in public preview in February 2026. It covers Microsoft Copilot, AI summaries in Bing and some partner integrations, and shows total citations, the average number of your pages cited per day, the grounding queries the AI used, and citation counts for individual URLs. It is the closest thing to first-party AI visibility data I know of, and many firms have never opened it.

Google Search Console

No AI-specific filter, but still useful. Watch impressions for the pages you are trying to get cited, and look for long, question-style queries. Google has said early AI Mode users asked questions two or three times longer than traditional searches, so growth in long queries is a reasonable sign.

Google Analytics

OpenAI's publisher FAQ says ChatGPT automatically adds utm_source=chatgpt.com to referral links. Set up a channel group or saved report that groups AI referrers together: chatgpt.com, perplexity.ai, copilot and gemini referrals. Volumes will be small. That is fine.

How do you connect visibility to revenue?

This is where most reporting stops, and where it should start.

Add a required "How did you hear about us?" field to every enquiry form, with "ChatGPT or another AI assistant" as an option, plus a free-text box. Ask the same question on the first call. Then, once a quarter, match won clients back to their source with your finance team.

This is how we report to clients. On one B2B services engagement, the client's own finance team counted 18 new accounts in eight months from enquiries we generated, worth about £40,000 invoiced. That number came from their books, not our dashboard. It is the only kind of number I would put in front of a board.

From our work

One US business law firm we reviewed in September 2026 had gone from appearing for 776 searches in February to 283 in August, and its estimated visits fell from 564 to 104. Nobody at the firm had noticed, because nobody was measuring it. If you only look at traffic once a quarter, a drop like that has already cost you six months of enquiries by the time you see it. That is why we track a fixed set of buyer questions every week.

What we'd do in your first 30 days

• Week 1. Build the question panel with you. Verify Bing Webmaster Tools and open the AI Performance report. Check robots.txt allows OAI-SearchBot, Bingbot and Googlebot.
• Week 2. Run the first full panel, three runs per question per tool. Produce a baseline share of answers by service and by tool, and a list of competitors being named instead of you.
• Week 3. Add AI referral grouping in Google Analytics and the "how did you hear about us?" field everywhere enquiries arrive.
• Week 4. Agree with your finance team how a won client is attributed and counted. Set the monthly reporting rhythm: customers and revenue first, enquiries second, visibility third.

Then the real work begins: improving the pages and third-party mentions that decide whether you get named. Our AI search and ChatGPT SEO pages cover that side.

Checklist

• Fixed list of 20 to 40 buyer questions
• Monthly runs across ChatGPT, AI Mode, AI Overviews, Perplexity and Copilot
• At least three runs per question per tool
• Bing Webmaster Tools verified, AI Performance report checked
• AI referrals grouped in Google Analytics
• "How did you hear about us?" on every enquiry route
• Quarterly revenue match with finance

FAQ

Is there a Google Search Console report for AI Overviews?

Not at the time of writing. Google counts AI feature traffic inside the normal "Web" search type.

Which AI search visibility tool should I use?

Any reputable tool that lets you fix your own prompts, runs them repeatedly and exports raw answers. Avoid tools that give a single score without showing you the answers behind it.

How many prompts do I need?

For a firm with a few service lines, 20 to 40 is enough to spot trends. More prompts add precision but also noise if they are not questions real buyers ask.

My AI referral traffic is tiny. Does that mean it isn't working?

Not on its own. Most people who see your name in an AI answer do not click. Look at share of answers, branded searches and what new clients tell you.

Get a baseline

Want more revenue from Google and ChatGPT? Enter your website at our free Search Check. Our team reviews every site, including how you show up in AI answers, and you hear back within one working day.