By Rahul Saini, Author at Search Counsel Co. Last updated August 2026.

Most teams either measure nothing here, or they measure the wrong thing and read it wrong. The good news: the visibility audit takes an afternoon, and the GA4 fix takes about 15 minutes. Once both are running, you can see which of your pages AI engines already cite, how many people that sends you, and whether the trend is climbing.

1) The short version

Auditing AI visibility means answering two questions, and they need two different tools.

• Are we cited? Run your buyer-intent prompts across ChatGPT, Perplexity, Gemini, Claude, and Google AI, and log where you appear. A visibility tool automates this.
• How much traffic does that send? Set up a custom channel group in GA4 with a regex filter on the AI domains, placed above the Referral channel, so AI shows up as its own row.

2) Visibility vs traffic: the distinction that trips everyone up

This is the single most common mistake, so it comes first. AI visibility is whether an engine mentions or cites your brand in its answer. AI traffic is whether a person then clicks through to your site. They’re related but separate, and they can move in opposite directions.

A big share of AI answers are zero-click: the reader gets what they need inside the answer and never visits your site. So you can have strong visibility, your brand named in dozens of answers, and modest traffic at the same time. GA4 can only ever see the clicks. It’s blind to the mentions that don’t produce one. That’s why a complete picture needs a visibility check on top of your analytics, not instead of it, and why a two-surface content strategy plans for both.

3) Part 1: audit your AI visibility (are you being cited?)

Start by finding out where you stand today. You don’t need a tool for the first pass, just a method.

• Build a prompt set. List the real buyer-intent questions your customers ask, the “best [category]”, “[category] for [use case]”, and “alternatives to [competitor]” style queries. Twenty to fifty is plenty to start.
• Run each across the engines. Ask ChatGPT (with search on), Perplexity, Gemini, Claude, and Google (watching for an AI Overview). Because AI answers vary run to run, ask each prompt a few times, not once.
• Log two things separately: whether your brand is mentioned, and whether your content is cited with a link. Note your position and which competitors appear.
• Read the sources. When a competitor gets cited, look at what got linked. It’s often a Reddit thread, a review page, or a roundup, which tells you where to work next. See why Reddit, G2 and third-party mentions drive AI citations.

To speed up the first pass, our free AI Visibility Checker runs a version of this automatically. For the rigorous, repeatable version of this method, with fixed prompts, multiple runs, and confidence intervals, see the AI Citation Index, which uses exactly this approach at scale.

4) Part 2: track AI referral traffic in GA4

By default, GA4 hides most AI traffic. When someone clicks a link inside an AI answer, the referrer is often stripped by the app or mobile browser, so GA4 files the visit under “Direct”, the same bucket as someone typing your URL. Here’s how to surface it. Our companion guide, tracking ChatGPT and Perplexity referrals in GA4, goes deeper on the reporting side.

The quick read: the native channel. In May 2026, Google added a built-in “AI Assistant” channel to GA4. To see it, open Reports, then Acquisition, then Traffic acquisition, and set the dimension to Session default channel group. The catch: it currently recognizes only ChatGPT, Gemini, and Claude, so Perplexity and Copilot traffic stays buried, and it doesn’t backfill. Treat it as a fast partial read, not the full picture.

The complete fix: a custom channel group. This takes about 15 minutes and captures every major engine.

1. Go to Admin, then Data Display, then Channel Groups.
2. Copy the Default Channel Group (or create a new one) and name it something like “Channel Group with AI”.
3. Add a new channel, name it “AI Search”, and set the condition to Session source matches regex.
4. Paste the regex below.
5. Critical step: drag the AI Search channel above the Referral channel. GA4 reads rules top to bottom, so if Referral comes first it grabs the AI traffic before your channel can.
6. Save the group, then select it in your acquisition reports to see AI as its own row.

New AI platforms appear regularly, so review this list quarterly and check your Referral source report each month for new AI domains showing up on their own. For historical analysis, build a free-form Exploration (Explore, then Free-form), add Session source as the dimension and Sessions as the metric, and apply the same regex as a filter. That gives you one table of every AI visit across the full date range GA4 holds, which you can pivot by landing page.

5) The reports that matter once tracking is live

The setup is only useful if you read the right views. Three earn their place.

• Landing page by AI source. This is the most actionable report you’ll build. It shows exactly which of your pages AI engines send people to, which tells you what’s working and what to create more of.
• Conversions by channel. Compare your AI channel’s conversion rate against organic and direct. AI referrals often convert several times better, because the assistant has pre-qualified the visitor before the click.
• The trend line. One month is noise. The month-over-month direction of AI sessions is the signal that tells you whether your visibility work is paying off.

For turning these into a standing client-ready dashboard, our SEO analytics and reporting guides cover the full reporting discipline, and building an SEO report in Looker Studio covers the build. This post is the AI-specific setup; that hub is the wider measurement system.

6) The AI Overviews blind spot

One gap you can’t fully close: Google AI Overview clicks are bucketed with regular organic search in GA4, labeled google / organic, identical to a normal search click. There’s no clean way to separate them in GA4 today.

The workaround is indirect. Watch Search Console for pages where impressions stay steady but click-through rate drops, which can signal that an AI Overview is answering the query without sending the click. Our guide to fixing low-CTR pages covers how to read that pattern. Use Search Console’s AI presence data qualitatively, to see which pages appear in AI responses, alongside your custom AI channel for the engines that do pass referrers.

7) The attribution gap (why GA4 is a floor, not a ceiling)

Even with the perfect setup, your measured AI traffic is an undercount. Referrer stripping means a large share of AI visits, by some measurements 30 to 40 percent, still leaks into Direct. Analyses have found the great majority of brands can’t fully attribute their AI referral traffic. So treat the GA4 number as a floor on AI’s real influence, not the whole of it, and keep that caveat visible when you report on ROI.

The fix is triangulation. Add a “How did you hear about us?” question to your signup or checkout flow with explicit AI options (“ChatGPT”, “an AI assistant”, and so on). Teams that do this routinely find far more AI influence than their analytics showed. Reading GA4, your visibility audit, and self-reported attribution together gives you a truer picture than any one of them alone.

8) Tools for the zero-click layer

Because GA4 can’t see mentions that don’t produce a click, a dedicated visibility tool fills the gap. These run prompt sets across engines and track where your brand is cited, before any traffic happens. Options in 2026 include Semrush’s AI visibility features, Ahrefs Brand Radar, Profound, Peec, and Otterly, among a fast-growing field. Verify current coverage and pricing before committing, since the market and the rates move quickly.

You don’t need a paid tool to start. The manual audit plus the free AI Visibility Checker and your GA4 channel cover the essentials. Add a platform when you’re running measurement often enough that automation pays for itself.

9) Your AI visibility audit checklist

• Built a buyer-intent prompt set and ran it across the five engines, more than once each.
• Logged mentions and citations separately, with position and competitors.
• Created a GA4 custom channel group with the AI regex, placed above Referral.
• Built the landing-page-by-AI-source report.
• Checked conversion rate for the AI channel against organic and direct.
• Added a self-reported “how did you hear about us?” field with AI options.
• Set a reminder to re-check the regex and re-run the prompt audit quarterly.

Sources used for this guide

The GA4 steps and figures below draw on current attribution research and Google’s own channel documentation.

FAQ: auditing and tracking AI visibility

Conclusion: measure both, report both

AI visibility and AI traffic are two numbers, not one. Run your prompts across the engines to see whether AI cites you, then set up a GA4 custom channel to see the traffic those citations send, remembering it’s a floor, not the whole of AI’s influence. Add the landing-page-by-source view, watch the trend, and triangulate with self-reported data.

Do that and you stop guessing about AI search and start managing it. For the benchmark version of the visibility side, see the AI Citation Index, for the wider reporting system see our SEO analytics and reporting guides, and for the work that moves the numbers, our AI search optimization guide.