How to measure AI visibility

Why rank tracking isn't enough

A rank tracker checks where your pages appear in a list of results. AI assistants do not return a list. They write an answer that may name several brands, link to a few sources, and change from one run to the next. None of that shows up in a position report.

So AI visibility needs different questions:

• When buyers ask about your category, are you named?
• Where in the answer, and next to which competitors?
• Is the description accurate and favourable?
• Which sources does the answer cite, and are any of them yours?
• Does any of this turn into visits and leads?

The metrics that matter

Build your question set

Your question set is the foundation. Keep it fixed, so changes over time mean something.

1. Collect real questions. Use sales calls, support tickets, site search and "People also ask" boxes. Write them the way buyers speak.
2. Cover the journey. Include discovery ("best tools for..."), problem ("how do I..."), comparison ("A vs B"), and brand questions ("is [brand] good for...").
3. Size it sensibly. Around 30 to 100 questions is enough for most brands. Fewer is noisy; many more is hard to act on.
4. Tag each question by topic and intent, so you can see where you are strong or weak.

Running the checks

• Use each assistant the way buyers do. Test the ones your audience uses, with web search on where that is the default.
• Start fresh each time. Use a new chat without history, so earlier questions do not steer later answers.
• Repeat important questions. Answers vary between runs. Two or three runs per question give a fairer picture than one.
• Record everything. Save the full answer, the brands named in order, and every cited URL, with the date.
• Check on a schedule. Weekly works for most brands. Look at the trend over several weeks, not a single run.

Doing this by hand is fine for a small question set. At larger scale, the assistants' APIs can run the same questions automatically, though API answers can differ slightly from the consumer apps.

Tracking AI referral traffic in GA4

People who click a link in an AI answer usually arrive as referral traffic from the assistant's domain. In Google Analytics 4 you can group them into their own channel:

1. Go to Admin, Data display, Channel groups and create a new channel group (copy the default one).
2. Add a channel called AI assistants and place it above Referral, so it is matched first.
3. Set its condition to Source matches regex, using a pattern like this:

(chatgpt\.com|chat\.openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai|grok\.com)

Some visits from assistants arrive without a referrer and show up as direct traffic, so treat this number as a floor, not the full picture.

Search Console and AI features

Google includes clicks and impressions from AI features such as AI Overviews and AI Mode in Search Console's Performance report, counted with regular web results rather than reported separately, at the time of writing. Useful signals:

• Impressions rising while clicks fall on the same queries often means an AI answer is now appearing above the links.
• Growth in branded queries can indicate that people are discovering you in AI answers, then searching for you by name.

A simple weekly report

The figures above are illustrative, to show the format.