Get In Touch
I'll be honest. When I first looked at my GA4 reports in early 2025, something felt off. Traffic was growing, but the sources didn't add up. Direct traffic had this unexplained bump I couldn't pin to any campaign. Turns out, ChatGPT and a handful of other AI tools had been quietly sending visitors to my clients' sites for months, and GA4 was scattering that data across three different buckets like it didn't know what to do with it.
Because it didn't. GA4 has no built-in channel for AI traffic. Visits from ChatGPT get tossed into "Referral." Clicks from Perplexity might land in "Direct." Meanwhile, a study that analyzed over 3.3 billion sessions found that AI referral traffic now accounts for more than 1% of all web visits, with ChatGPT responsible for roughly 87% of it.
So if you're running any kind of SEO or generative engine optimization strategy in 2026, you need to track LLM traffic in Google Analytics 4 separately. This guide is essentially LLM traffic tracking explained in simple terms. Here, I’ve covered three methods to track AI traffic in Google Analytics: custom channel groups, exploration reports, and advanced GTM setups. I'll walk through each one, give you the regex patterns, show you how to track LLM traffic in GA4, and explain what your data actually means once you can finally see it.
Before we fix the problem, it helps to understand why it exists.
Every time someone lands on your site, GA4 looks at two things: the source (where they came from) and the medium (how they got there). These values get matched against a set of rules called "default channel groups," which sort traffic into familiar categories like Organic Search, Direct, Social, and Referral.
Here's the issue. GA4's default channel definitions were built before ChatGPT existed. The system doesn't have rules for chatgpt.com, perplexity.ai, or claude.ai. So when someone clicks a link inside an AI chat and arrives on your website, GA4 checks its rulebook, finds no match for the source, and dumps the visit into either Referral or (worse) Unassigned. That's why Google Analytics 4 AI traffic tracking requires manual configuration. It's not that GA4 is broken. It just hasn't caught up.
It gets messier. Not all AI traffic even sends a referrer header. When a free-tier ChatGPT user clicks a link inside the mobile app, the browser often doesn't tell your site where the click came from. Same thing with in-app browsers and certain API integrations. Without a referrer, GA4 defaults to "Direct," which is the analytics equivalent of a shrug.
Reports suggest that 20-40% of AI-generated clicks arrive without referrer data. That's not a small margin of error. It means that any attempt to track ChatGPT traffic in Google Analytics 4 will always undercount the real volume.
Google has acknowledged that a dedicated AI traffic channel is coming to GA4. But "coming" is doing a lot of heavy lifting in that sentence. As of early 2026, there's still no native channel. Google AI Overviews traffic is still blended with organic data in Search Console. And while Google has been rolling out AI Mode and other AI search features, the analytics side hasn't kept pace.
So for now, you build your own tracking, which isn't that hard once you know the steps.
Before you can meaningfully track AI traffic in Google Analytics, you need context. Let me share some numbers, because this is where the conversation gets interesting for anyone doing LLM optimization or confused about an AEO vs SEO strategy.
An analysis of nearly 64,000 websites across 250 countries found AI traffic grew roughly 7x year-over-year, jumping from 0.02% in early 2024 to about 0.15% by April 2025. By late 2025, Conductor's benchmark pegged it at 1.08% across 13,770 domains.
One percent doesn't sound like much until you consider two things. First, it's growing faster than any other referral channel. Second, for industries such as IT and consumer products, LLM referral traffic in GA4 is already at 2-3% of the share. If your competitors are tracking this and you're not, they're seeing something you can't.
ChatGPT dominates. There's really no other way to put it. It drives roughly 77-87% of all AI referral traffic, depending on which study you look at. Once you learn how to identify ChatGPT traffic in GA4, you'll likely find it's your single largest AI source. If you're interested in getting your brand visible inside that platform specifically, I've written a separate guide on how to get your business showing up in ChatGPT.
Perplexity comes in as a distant second, accounting for roughly 15% of AI referrals. It's smaller but punches above its weight in certain B2B verticals.
But the fastest growth story belongs to Google Gemini. Gemini referral traffic grew 388% year-over-year between September and November 2025, compared to ChatGPT's 52% growth in the same period. Gemini's monthly active users climbed about 30% to 346 million. If that trajectory holds, ranking in Gemini's answers becomes a real priority by mid-2026.
Copilot, Claude, and DeepSeek round out the list with single-digit shares, but they're worth including in your tracking regex as their user bases grow.
This surprised me when I first segmented the data. A study analyzing 973 e-commerce websites over 12 months, found that ChatGPT traffic had lower bounce rates than traditional search traffic but also lower conversion rates and revenue per session.
The researchers described it as traffic that's "clearly relevant" but doesn't yet translate into comparable sales outcomes. In plain terms: AI sends you curious, engaged visitors who aren't quite ready to buy. They're researching and validating. That's worth knowing because it should change how you think about the landing pages receiving this traffic.
This is the most straightforward way to set up Google Analytics 4 AI traffic tracking that I recommend to most clients. It takes about 10-15 minutes, works retroactively with your historical data, and puts AI traffic front and center in every standard acquisition report.
Open your GA4 property and go to Admin (the gear icon). Under Data Display, click Channel Groups. You'll see your Default Channel Group listed.
Click Copy to create a new one. This duplicates all existing channels, so you don't lose anything. Name the copy something like "Default + AI Traffic."
Click Add new channel and name it "LLM Traffic" or "AI Traffic." Under channel conditions, select Source as the dimension and set the match type to matches regex.
Paste this into the regex value field:
^.*ai|.*\.openai.*|.*copilot.*|.*chatgpt.*|.*gemini.*|.*gpt.*|.*neeva.*|.*writesonic.*|.*nimble.*|.*outrider.*|.*perplexity.*|.*google.*bard.*|.*bard.*google.*|.*bard.*|.*edgeservices.*|.*astastic.*|.*copy.ai.*|.*bnngpt.*|.*gemini.*google.*$
This covers the major AI platforms as of early 2026. I've used escaped dots (\.) for precision on domain matches. You could use broader .*chatgpt.* wildcards, but that risks false positives. Better to be precise and add new sources as they appear.
And they will appear. New AI tools launch constantly, and existing ones rebrand (remember when Gemini was Bard?). Review and update this regex quarterly.
This trips everyone up. After saving the new channel, you have to reorder it. GA4 processes channel rules top to bottom. If Referral sits above your AI Traffic channel, visits from chatgpt.com match the Referral rule first and never reach your custom channel.
Click Reorder, drag AI Traffic above Referral, and save. Skip this, and your new channel will show zero sessions.
Go to Reports > Acquisition > Traffic Acquisition. In the data table, switch the dimension dropdown to Session custom channel group and select yours.
You should see "LLM Traffic" as its own row alongside Organic Search, Direct, and everything else. Custom channel groups are retroactive, so historical data appears immediately. If the row is empty, check your regex for typos and verify the channel order.
Custom channel groups are great for ongoing monitoring. But if you want to track ChatGPT traffic in Google Analytics 4, do ad hoc analysis, dig into specific LLM sources, or build visualizations for stakeholder reports, GA4's Explore section is where you want to be.
Go to Explore in the left navigation and click + Blank. Name it "LLM Traffic Analysis."
In the Variables panel, add these dimensions: Session source/medium, Page referrer, and Landing page. For metrics, add Sessions, Engaged sessions, Engagement rate, and Conversions. Drag the Session source/medium into Rows and Sessions into Values to create a basic traffic source table.
Click the filter icon at the top of your exploration. Select Session source as the dimension, set the match type to matches regex, and paste the same regex pattern from earlier.
Apply the filter. Your table now shows only AI-referred traffic, broken down by specific source. You can see exactly how many sessions came from ChatGPT vs. Perplexity vs. Gemini, along with engagement metrics for each.
Want to know which pages AI tools link to most? Add a landing page as a secondary dimension. This reveals which content pages are being cited in AI responses, a valuable insight for your content strategy.
Duplicate the exploration tab. Change the visualization type to a line chart. Add Date as a dimension in Rows and Sessions as the metric in Values. Apply the same regex filter.
You now have a trend line showing how AI traffic has changed over time. Switch between daily, weekly, and monthly granularity to spot patterns. AI referral traffic tends to be more volatile than organic search. Siege Media's analysis of 475 million sessions found ChatGPT referrals dropped 7.7% during the summer months, with B2B segments hit hardest.
The methods above handle 80% of what most marketers need. But for deeper event-level tracking or polished dashboards, there's more you can do. This is where Google Analytics 4 AI traffic tracking gets really granular.
Open Google Tag Manager and create a new tag with the GA4 Event tag type. Name the event “ai_referral_session”. For the trigger, create a custom trigger that fires when the page referrer matches your AI regex pattern using a Regex Table variable against the “document. referrer”.
Add parameters like “ai_source” (which LLM sent the traffic) and “landing_page_path”. These capture context that standard GA4 reports don't surface. Test in Preview mode before publishing.
After publishing your GTM tag, go to Admin > Custom Definitions in GA4 and register each parameter from your event tag (like ai_source). This makes your custom data available in reports and explorations. Without this step, GA4 collects the data but won't let you use it in analysis.
For client-facing reports, Looker Studio gives you more flexibility than GA4's built-in reports. Connect your GA4 property as a data source and use your custom channel group dimension directly.
Build a dashboard that includes: a time series chart for AI traffic trends, a breakdown table by AI source, a landing page analysis showing which content gets the most AI referrals, and a conversion comparison between AI, organic, and direct channels.
Setting up the tracking is only half the job. The interesting part is figuring out what to do with the data once you can see LLM referral traffic in GA4 as its own channel.
Once you can filter by AI source and see landing pages, patterns emerge quickly. In most cases, you'll find that LLMs cite a handful of your pages repeatedly. These tend to be pages with clear definitions, structured comparisons, step-by-step instructions, or original data.
That's not a coincidence. AI models gravitate toward content that's well-organized and answers questions directly. If certain pages get heavy AI traffic, treat them like gold. Update them regularly, add schema markup, and make sure they represent your brand well since they're essentially your AI-facing front door.
Understanding the difference between AI brand mentions and AI citations matters here, too. A citation means the AI is linked to your page. A mention means it referenced your brand by name without a link. GA4 only captures citations (clicks). Mentions are invisible in analytics, but they're still building awareness.
Pull up your custom channel group report and compare metrics side by side. In most industries, AI traffic shows higher engagement rates but lower conversion rates compared to organic search. Sessions are longer, and pages per session are higher.
This tells you AI visitors are in research mode. They came because a chatbot told them your page had a good answer. They're reading and clicking around, but aren't ready to fill out a form yet. That's top-of-funnel awareness, exactly the kind of traffic that LLM seeding strategies are designed to generate.
Look for content gaps. If AI tools send traffic to your comparison pages but not your product pages, that's a signal. LLMs view you as informational but not transactional. Fix that by adding product-adjacent content like buyer's guides, use-case breakdowns, or ROI calculators.
Also, pay attention to which platforms send the highest-quality traffic. ChatGPT might send volume, but Perplexity visitors might convert better. If that's the case, focus your answer engine optimization on platforms that drive the best outcomes, not just the most sessions.
And if your content isn't getting cited at all, there's usually a reason. I've put together a guide covering why content doesn't show in AI Overviews that might help you diagnose the issue.
I'd be doing you a disservice if I made this sound like a complete solution. There are real gaps.
A significant chunk of AI-generated traffic arrives without referrer data. Free-tier ChatGPT users, mobile app clicks, and copy-paste behavior all create sessions that GA4 classifies as Direct. You can't recover that retroactively. Accept that your GA4 AI traffic numbers represent a floor, not a ceiling.
This is the biggest blind spot. When someone clicks a link from a Google AI Overview or AI Mode result, that click is reported as regular Google organic traffic in both GA4 and Search Console. If you're investing in ranking within AI Overviews, the traffic impact is real but invisible in your analytics.
Between Direct traffic masking, AI Overviews blending, and new AI tools launching constantly, what you see in your reports is an undercount. But even partial data beats no data. You can still see trends in LLM referral traffic in GA4, compare performance, and identify which content gets cited. And because the GEO versus SEO distinction is only becoming more important, having baseline data now puts you ahead.
The brands tracking AI mentions and AI referral traffic now aren't doing it because the numbers are huge today. They're doing it because the numbers are growing fast, and having six months of baseline data is infinitely more useful than starting from zero when the channel demands budget allocation. Now that you know how to track LLM traffic in Google Analytics 4, there’s no reason to delay things.
The setup is straightforward. Fifteen minutes for a custom channel group. Another thirty for Explore reports and a Looker Studio dashboard. You can track ChatGPT traffic in Google Analytics 4 and every other major LLM with the regex patterns in this guide. If the process still looks a bit complicated, getting an expert AI SEO agency on board is the way to go.
If you want to go beyond tracking and start increasing how often AI tools cite your content, we've written extensively about AI optimization as a discipline. And if you'd rather have someone handle analytics and optimization together, that's what we do at ViralChilly.
No, not currently. Clicks from AI Overviews and AI Mode are reported as standard Google organic traffic in both GA4 and Google Search Console. There's no referrer distinction that separates AI-generated results from traditional blue links within Google's own ecosystem.
I recommend a quarterly review at a minimum. Check your unfiltered Referral traffic for new domains containing "ai," "chat," or "assistant" in the source. When a new AI platform launches or an existing one rebrands, add it to your pattern promptly. Keeping your regex current is essential for accurate LLM referral traffic in GA4.
Yes. Once you create and save a custom channel group, GA4 applies it to your existing historical data immediately. You don't need to wait for new data to accumulate. This is one of the biggest advantages of figuring out how to track LLM traffic in GA4 over event-based tracking, which only captures data going forward.
This happens when the referrer header gets stripped. It's most common with ChatGPT's mobile app, free-tier browser sessions, and in-app browsers that don't pass referrer information. Without that header, GA4 has no way of knowing the visit originated from ChatGPT, so it defaults to Direct. That's why learning how to identify ChatGPT traffic in GA4 requires regex filters rather than relying on default reports.
You can do the core tracking entirely within GA4 using custom channel groups and Explore reports. GTM is only necessary if you want custom event-level tracking, like firing a specific event when an AI-referred visitor takes a particular action. For most teams, the GA4-only approach is sufficient to start.
