How to Track AI Search Traffic in GA4

2. What Counts as AI Search Traffic?

Quick Answer: AI search traffic is any visit that lands on your site after someone uses an AI tool — ChatGPT, Perplexity, Gemini, Copilot or Claude — instead of clicking a normal search result. It behaves like a referral: the AI names your site, the user taps the citation, and they arrive already informed. It is the traffic side of generative engine optimization.

There are two distinct flavours, and they show up differently in your data. The first is a true referral click, where someone reads an AI answer and clicks a linked source to your site. The second is a click from Google’s own AI Overview, which still arrives as ordinary Google organic, not as a separate AI source.

The AI engines that send Malaysian businesses the most referral traffic today are a short, stable list:

• ChatGPT. The largest single source, arriving from chatgpt.com when a user clicks a cited link.
• Perplexity. A search-first AI that shows sources prominently, so it clicks through more than most.
• Google Gemini and AI Mode. Gemini referrals plus AI answers blended into Google search.
• Microsoft Copilot and Claude. Smaller but rising, especially among office and research users.

If the difference between ranking and being cited is still fuzzy, our explainer on SEO, AEO and GEO lays out where each one fits.

3. Why AI Traffic Is Hard to See in GA4 by Default

Quick Answer: GA4 has no built-in “AI” channel. Visits from ChatGPT or Perplexity drop into the generic “Referral” bucket, and when an AI app strips the referrer, they vanish into “Direct”. The traffic is in your data — it is just mislabelled and scattered, which is why AI clicks feel invisible even as click-through behaviour shifts.

Three things conspire to hide AI traffic from a standard report:

• No native AI channel. GA4’s default channel groups were built before AI search existed, so nothing labels these sources as “AI”.
• Referrer stripping. Many AI tools open links inside an app or strip the referrer, so the visit lands as “Direct” with no source at all.
• AI Overviews look organic. A click from Google’s AI answer is still tagged as Google organic search, blending in with every other Google visit.

The result is an undercount. Your real AI traffic is higher than the “Referral” line suggests, because a chunk of it has scattered into “Direct” and into organic. Fixing the labelling is what turns a vague hunch into a number you can report.

4. How AI Platforms Show Up in GA4

Quick Answer: Each AI platform arrives under a different source name, and most default to “Referral”. Knowing the exact referrer domains is the key to capturing them all with one rule. The table below maps the main AI engines to how they appear in your GA4 reports, the same way they surface when you get cited in AI answers.

Build your tracking around these source names. They change slowly, so one well-made rule lasts for months.

5. How to Build an AI Channel Group in GA4

Quick Answer: The cleanest way to track AI search traffic is a custom channel group that catches every AI source with one regex rule. It takes about five minutes in GA4 Admin, and once saved it works across your reports and Explore. Pairing it with a focused SEO plan, like our SEO service, turns the numbers into a growth lever.

How to set up an AI channel group in GA4

Follow these steps inside your GA4 property. You need Editor access to save a channel group.

1. Open Admin then Channel groups. Click the Admin gear at the bottom left, then find Channel groups under Data display.
2. Create a new channel group. Click Create new channel group and name it something clear, like “AI / LLM traffic”.
3. Add a channel and a condition. Add a new channel called “AI”, then set its rule to Source matches regex.
4. Paste an AI regex. Use a pattern that covers the main engines, for example: chatgpt|openai|perplexity|gemini|copilot|claude|deepseek
5. Order it above Referral. Drag the AI channel above Referral so it catches these sources before the generic bucket does.
6. Save and apply. Save the group. It now appears as a “Session custom channel group” dimension across your reports.

One extra habit helps a lot. When you paste your own links into an AI tool, add UTM tags such as utm_source=chatgpt so those clicks self-label even if the referrer is stripped.

6. How Fast AI Referral Traffic Is Growing

Quick Answer: For most Malaysian SMEs that publish helpful, well-structured content, AI referral traffic starts small and compounds month over month. The illustrative ramp below shows the shape we see across client accounts: quiet at first, then a steady climb as more pages get cited and the question of whether SEO is dead answers itself.

The numbers below are a modelled example for a site starting consistent AI-visibility work in month one. The exact figures vary by industry, but the shape is what matters.

7. AI Referral Visitors vs Google Organic: How They Behave

Quick Answer: AI referral visitors are not the same as Google organic visitors. In our client data they arrive in smaller numbers but more pre-qualified — they have already read an AI summary, so they spend longer and convert at a higher rate. That is the upside hiding behind the zero-click search shift.

When you compare the two channels side by side, the pattern is consistent: fewer AI visits, but better ones. The figures below are an illustrative view from accounts we manage.

The lesson is to judge AI traffic on quality, not just volume. A channel that sends 200 pre-qualified visits a month can out-earn one that sends 2,000 cold ones.

8. How to Track Google AI Overviews and AI Mode Clicks

Quick Answer: Clicks from Google AI Overviews and AI Mode do not appear as a separate source — they count as normal Google organic in GA4. To see their effect, pair GA4 with Search Console and watch impression and landing-page trends on your most answer-style pages, the ones built to rank in Google AI Overviews.

Because the source is invisible, you measure this one indirectly. Three signals work well together:

• Search Console impressions vs clicks. Rising impressions with flat clicks on a query often means an AI Overview is answering it without the click.
• Landing-page trends in GA4. Watch organic sessions to pages written as direct answers; movement there hints at AI-answer pickup.
• Self-reported attribution. A simple “How did you hear about us?” field captures AI-assisted visits that never carry a referrer.

The Ahrefs walkthrough above also covers bot analytics, separating training crawlers from citation bots, which is a useful third lens once your referral and organic tracking is solid.

9. Which AI Engines Send the Most Traffic

Quick Answer: Across Malaysian SME sites we track, ChatGPT sends the largest slice of AI referral traffic, followed by Perplexity and Gemini. Knowing the split tells you where to focus, and which AI marketing tools are worth testing for your own research.

This breakdown helps you prioritise. If ChatGPT drives half your AI visits, that is where a citation is worth the most.

10. Common Mistakes When Tracking AI Search Traffic

Quick Answer: The biggest mistakes are ignoring “Direct” inflation, building a regex that misses new engines, and reporting AI traffic in isolation instead of against leads. Avoid these and your AI numbers stay trustworthy enough to guide where your SEO investment goes.

These are the slips we see most often when businesses first start measuring AI traffic:

• Treating all Direct as type-ins. A jump in Direct often hides AI clicks whose referrer was stripped. Investigate before you dismiss it.
• A brittle regex. A pattern that only lists today’s engines misses tomorrow’s. Review it every quarter and add new sources.
• No conversion tracking. Counting sessions without leads tells you nothing about value. Tie AI traffic to your key events.
• Reporting volume, not value. AI sends fewer but warmer visits, so a quality lens beats a raw-traffic one.
• Checking once, then never again. AI search is moving fast. A monthly look keeps your view current.

11. Conclusion

AI search traffic is no longer a rounding error. It is a small, fast-growing, high-intent channel that most Malaysian businesses cannot see because GA4 was not built to label it. The fix is straightforward: know the source names, build one AI channel group with a regex, and read AI Overviews indirectly through Search Console and self-reported attribution.

Do that, and you turn a blind spot into a number you can act on. You will know which engines send buyers, how those buyers behave, and whether your AI visibility is climbing month over month — the foundation for deciding where to invest next.

12. Frequently Asked Questions

1. Can GA4 track AI search traffic automatically?

Not on its own. GA4 has no built-in AI channel, so visits from ChatGPT, Perplexity or Gemini land in the generic “Referral” group, and some slip into “Direct” when the referrer is stripped. You get a clear AI view only after you build a custom channel group with a regex that matches the AI source names. It takes about five minutes and then works across all your reports.

2. Why does my AI traffic show as Direct?

Many AI tools open links inside an app or strip the referrer for privacy, so the visit arrives with no source attached. GA4 files anything without a source under “Direct”. This is why a rise in Direct traffic often hides real AI clicks. Adding UTM tags to links you share in AI tools, and using a “how did you hear about us” question, helps recover some of that lost attribution.

3. What regex should I use for an AI channel group in GA4?

Start with a pattern that matches the main engines by source, such as chatgpt|openai|perplexity|gemini|copilot|claude|deepseek. Set the channel rule to “Source matches regex”, paste the pattern, and drag the AI channel above the Referral channel so it catches these visits first. Review the pattern every few months and add new engines as they appear, since the AI landscape changes quickly.

4. Do Google AI Overviews show up in GA4?

No, not as a separate source. A click from a Google AI Overview or from AI Mode is recorded as ordinary Google organic search, so it blends in with all other Google visits. To gauge its impact, pair GA4 with Search Console, watch impressions against clicks on answer-style queries, and track organic sessions to the pages you have written as direct answers.

5. How much traffic should AI send to a Malaysian SME site?

It varies, but the shape is consistent. AI referral traffic usually starts under 1% of total sessions and compounds as more of your pages get cited, often passing 3 to 5% within six months of steady content work. The volume stays modest compared with Google organic, but the visits tend to be more pre-qualified, so judge the channel on conversions rather than raw session counts.