How to Track AI and ChatGPT Referral Traffic in GA4

As more people begin their research inside ChatGPT, Perplexity, Gemini and Copilot rather than a traditional search box, a new question is landing on marketers' desks: are these tools actually sending us visitors, and how would we even know. The answer matters, because AI assistants increasingly cite and link to sources, and those citations can drive genuinely high-intent traffic that your reporting may currently be lumping in with everything else.

This guide explains, in practical terms, how to find and measure AI referral traffic in Google Analytics 4. It covers where this traffic shows up by default, how to build a view that isolates it, the limits of what you can and cannot see, and how to judge its quality rather than just its volume. It is deliberately honest about the gaps, because measuring AI visibility well means understanding what the data can and cannot tell you.

Where AI Referral Traffic Actually Shows Up

When someone clicks a citation or link inside an AI assistant and lands on your site, GA4 records it like any other referral, attributing the session to the source domain of the tool. In practice that means you will see hostnames associated with the major assistants appearing as referral sources, for example the domains used by ChatGPT, Perplexity, Gemini and Copilot. The exact domains evolve as the products change, so the reliable approach is to look for the pattern rather than memorise a fixed list.

The catch is that, out of the box, GA4 does not have an AI channel. This traffic falls into the general Referral channel alongside every other site that links to you, which is why so many businesses assume they are getting no AI traffic when in fact it is simply uncategorised. To see it, you need to go looking in the Traffic acquisition report, filter the session source or referrer for the assistants' domains, and observe how much is already arriving.

Start, then, by auditing what is already there. In Traffic acquisition or an exploration, add session source or session referrer as a dimension and search for the AI domains you care about. Even a small, steady trickle is worth knowing about, because it tells you which tools are surfacing your content and, when you look at the landing pages, which of your pages they are citing. That single view often reframes the conversation from whether AI sends traffic to which pages are earning it.

Group AI Referrers Into One Clear Segment

Once you know AI traffic is arriving, the next step is to stop it hiding inside general referrals. The cleanest way to do this in GA4 is a custom channel group, which lets you define a new channel, for example AI Assistants, using rules that match the referrer or source against the assistants' domains. From then on, that traffic reports as its own channel in your acquisition reports, so you can trend it over time without rebuilding a filter each session.

If you would rather not change your channel grouping, an exploration achieves much the same result on demand. Build a free-form exploration with session source or referrer as the row dimension, add sessions, engagement and your key conversions as metrics, and apply a filter that includes only the AI domains. Save it, and you have a reusable AI referral dashboard you can open whenever you need it, alongside a landing-page breakdown that shows which content is being cited.

Whichever route you choose, define the segment by matching the family of AI domains rather than a single hard-coded hostname, because the products rename and restructure their linking over time. Review the definition periodically so it keeps capturing new or changed domains. The goal is a durable, low-maintenance view that answers three questions at a glance: how much AI referral traffic you get, which tools it comes from, and which pages earn it.

Create a custom channel group with an AI Assistants channel matching the tools' domains
Or build a saved exploration filtered to AI referrers for an on-demand view
Match a family of domains, not one hostname, so renames do not break your tracking
Always pair the source view with a landing-page breakdown to see what is being cited

Understand What GA4 Can and Cannot Tell You

It is essential to be clear about the limits, because AI referral data is only part of the picture. GA4 records a visit when someone clicks through to your site from an AI tool. It does not, and cannot, record the many occasions when an assistant mentions your brand, summarises your content or answers using your information without the user ever clicking a link. Much of the value of AI visibility is this influence without a click, which sits outside your analytics entirely.

There are further practical caveats. Referrer information can be missing or stripped in some contexts, so a share of genuine AI-assisted visits may arrive without a clear source and blend into direct traffic. The specific domains used by each assistant change as the products evolve. And volumes are often modest today compared with organic search, which is normal for an emerging channel and not a reason to dismiss it, especially given how qualified the visitors tend to be.

Treat GA4, therefore, as one instrument among several. It gives you a reliable, trending measure of click-through AI referrals and the pages that earn them, which is genuinely useful. To understand the wider picture of whether AI systems are citing you at all, you complement it with direct testing of the assistants on your priority questions and, where available, Search Console data on AI-influenced surfaces. No single tool captures AI visibility completely, so the honest approach is to combine them.

Judge Quality, Not Just Volume

Because AI referral numbers are usually small at this stage, judging them purely on volume undersells them. The more useful lens is quality. Visitors arriving from an AI assistant have often already been given a considered answer and are clicking through to verify, go deeper or take action, which can make them more engaged and closer to a decision than a casual searcher. Look at engagement time, pages per session and, crucially, conversions and enquiries from this segment, not just the session count.

Connecting AI referrals to outcomes is what turns the data into a business case. Set up your key conversions in GA4, whether that is a form submission, a call, a quote request or an audit booking, and view them by your AI channel or exploration. If a modest volume of AI referrals is producing enquiries at a healthy rate, that is a strong signal to invest in the content and structuring that earns AI citations, even while the raw traffic remains small.

Finally, use what the landing-page data tells you. The pages AI tools cite reveal the format and depth they favour: clear, well-structured, genuinely useful content that answers a question directly. Doubling down on that style, and on the specific pages already earning citations, is how you grow the channel deliberately. If you would like help setting up AI referral tracking and building content designed to be cited, SEODXB works on both measurement and answer-engine optimisation, and offers a free audit with no lock-in contracts.

Assess AI referrals on engagement and conversions, not session volume alone
Configure key conversions in GA4 and view them by your AI segment
Use the cited landing pages to learn which content format AI tools favour
Invest in the pages and structure already earning citations to grow the channel

AI assistants that cite sources do send referral traffic, and GA4 already records it, but by default it hides inside the general Referral channel under each tool's domain rather than appearing as an AI channel. To measure it, audit Traffic acquisition for the assistants' domains, then isolate them with a custom channel group called something like AI Assistants, or a saved exploration, always paired with a landing-page breakdown so you can see which content is being cited. Match a family of domains rather than one hostname so renames do not break your tracking. Be honest about the limits: GA4 captures click-throughs, not the many times an assistant mentions you without a link, and some referrers arrive unattributed. Because volumes are still modest, judge this traffic on engagement and conversions rather than session count, and connect it to real enquiries to build the case for investing in citable content. Combine GA4 with direct testing of the assistants for the fuller picture. SEODXB helps set up AI referral tracking and build content designed to be cited, with a free audit and no lock-in contracts.

Related guides and services

Frequently asked questions

Can you see ChatGPT and AI traffic in Google Analytics?

Yes, when someone clicks a link or citation inside an AI assistant and lands on your site, GA4 records it as a referral session under the tool's domain. The reason many businesses think they get no AI traffic is that GA4 does not group it as an AI channel by default; it sits inside general referrals. Filter Traffic acquisition by session source or referrer for the assistants' domains, or build a custom channel group, and the traffic becomes visible.

How do I create an AI channel in GA4?

Use a custom channel group. In GA4 admin, create a new channel group, add a channel such as AI Assistants, and define a rule that matches the session source or referrer against the domains used by ChatGPT, Perplexity, Gemini, Copilot and similar tools. From then on this traffic reports as its own channel in your acquisition reports. Match a family of domains rather than a single hostname, and review the rule periodically as the tools change their domains.

Does GA4 track every time AI mentions my brand?

No, and this is an important limit. GA4 only records a visit when someone actually clicks through to your site from an AI tool. It cannot capture the many times an assistant mentions your brand, summarises your content or answers using your information without the user clicking a link. Much of the value of AI visibility is this influence without a click, so GA4 measures the click-through slice, and you need direct testing of the assistants to understand the wider picture.

Why is my AI referral traffic so low?

Modest volumes are normal for an emerging channel and not necessarily a problem. AI-assisted research often ends without a click because the assistant answers directly, so click-through traffic is only part of your true AI visibility. Some genuine visits also arrive without a clear referrer and blend into direct traffic. Rather than judging the channel on volume, look at engagement and conversions from AI referrals, which are often high because these visitors arrive already informed and closer to a decision.

Which pages should I look at for AI referrals?

Pair your AI source view with a landing-page breakdown in the same report or exploration. The landing pages receiving AI referrals show you exactly which content the assistants are citing, which in turn reveals the format they favour: clear, well-structured, genuinely useful pages that answer a question directly. Those pages are your template. Strengthening them, and creating more content in the same style, is the most direct way to grow AI citations and the traffic that follows.

Is AI referral traffic worth optimising for in 2026?

For most businesses, yes, but with realistic expectations. The raw volume is usually still small next to organic search, so it should complement rather than replace your core SEO. What makes it worth attention is quality: AI referrals tend to be engaged and high-intent, and the share of research happening inside assistants is growing. The sensible approach is to measure it properly, connect it to enquiries, and invest in citable content, so you build the channel deliberately as it matures.