How Do I Track ChatGPT and AI Traffic in GA4?

Can You Actually See ChatGPT Traffic in GA4?

Yes, partially. GA4 shows sessions where ChatGPT, Perplexity, or another AI tool passed a referrer header — these usually land in the Referral or Unassigned channel group. But a real share of AI-assisted visits arrive as Direct traffic instead, because the click happens inside an app shell (the ChatGPT app, an in-app browser) that strips the referrer entirely — so GA4 alone will always undercount your true AI traffic.

This matters because most Shopify owners land in one of two wrong places: they check once, see almost nothing, and conclude "AI doesn't send us traffic" — or they see a number and treat it as the complete picture. Neither is right. The real number sits between what GA4 reports and what's actually happening, and closing that gap is the point of this guide.

For the bigger-picture question of whether AI tools mention your store at all — not just whether they send clicks — see how to track your AI search visibility, which covers manual spot-checks and citation monitoring. This guide is narrower: it's specifically about isolating AI referral sessions inside GA4 itself.

Step 1: Know the Referrer Domains to Filter For

Each AI tool sends a different domain when it does pass a referrer, so start by filtering GA4 for the handful that matter. As of early 2026, the domains worth watching are:

• chatgpt.com — ChatGPT's consumer web app (OpenAI moved chat.openai.com traffic here; older links may still redirect)
• perplexity.ai — Perplexity's answer engine, generally reliable at passing referrer data
• bing.com and copilot.microsoft.com — Microsoft Copilot-assisted searches, often logged as a Bing referral
• gemini.google.com — Google's Gemini assistant
• www.google.com with AI Overviews — this is the tricky one: AI Overview clicks still show up as ordinary Google organic traffic in GA4. Google's own documentation on AI features in Search doesn't describe a separate referrer or channel for AI Overview clicks, so you cannot cleanly isolate them from regular Google search clicks using GA4 alone.

Do not assume ChatGPT traffic will show up cleanly. OpenAI has said ChatGPT can browse the web and cite sources in its answers, but whether a given click passes a full referrer depends on the surface (chatgpt.com web vs. the mobile app vs. a shared conversation link) and can change without notice as OpenAI ships updates. Treat referrer data as a signal, not a ledger.

Step 2: Build the GA4 Filter or Exploration

Here's the direct path — no plugins needed, just GA4's native reporting.

1. In GA4, go to Reports → Acquisition → Traffic acquisition.
2. Change the primary dimension to Session source / medium.
3. Use the search box to type chatgpt, then perplexity, then bing, checking each one against the list in Step 1.
4. For a cleaner, saved view, go to Explore → Blank exploration instead. Add Session source as a dimension, Sessions and Conversions as metrics, and set a filter where Session source contains chatgpt.com OR perplexity.ai OR gemini.google.com.
5. Save the exploration so you can revisit it monthly without rebuilding the filter.

If you want a standing segment rather than rebuilding a report each time, create a Session-scoped segment in Explore: condition = Session source matches regex chatgpt\.com|perplexity\.ai|copilot\.microsoft\.com|gemini\.google\.com. Apply that segment across any exploration going forward.

A caveat worth flagging honestly: if your sessions are in the single or low double digits, don't over-read the trend. Small-sample noise looks a lot like a real pattern when you want it to.

Step 3: Cross-Check Against "Direct" and Unassigned Traffic

Referrer-based filtering will always miss the sessions that arrive as Direct. This step is how you estimate the size of that blind spot rather than ignoring it.

• Watch your branded search volume in Google Search Console (Performance → Queries, filtered to your brand name). When AI tools mention a store, curious shoppers often search the brand name on Google right after. A rising branded-query trend that doesn't correlate with any campaign you ran is a reasonable indirect signal that something — an AI mention, a podcast, word of mouth — is driving people to look you up.
• Check landing page behavior for Direct sessions. If Direct sessions are landing disproportionately on a specific product or comparison page (rather than your homepage, which is the more typical Direct pattern), that's a hint some of it is AI-referred traffic that lost its referrer.
• Add a post-purchase or on-site survey question: "How did you hear about us?" with an "AI tool like ChatGPT" option. This is the most reliable single data point you can get, because it doesn't depend on referrer headers surviving the trip.

None of these fully close the gap. Combined, they give you a defensible estimate instead of a blind guess.

Step 4: Tag Your Own Links Where You Can

You can't control what ChatGPT or Perplexity pass as a referrer, but you can control links you post yourself that AI tools might crawl and cite — your blog, your press mentions, your directory listings.

• Add UTM parameters (utm_source=chatgpt, utm_medium=ai-referral) to any link you specifically expect an AI assistant to surface, such as a resource page you know gets cited.
• This won't help with organic AI citations of pages you don't control the link for, but it does mean any traffic that flows through your tagged links will show up cleanly in GA4 regardless of what the AI tool itself passes.
• Keep a simple naming convention so these don't collide with your existing UTM campaigns for ads or email.

What to Track Monthly, Summarized

Run this monthly rather than daily. AI referral volume is still small for most Shopify stores in early 2026, and daily noise will make you chase phantom trends. A three-to-six month view is where real patterns show up.

Why Referral Traffic Is Only Half the Story

Tracking clicks tells you AI tools are sending people. It says nothing about how often you get mentioned but not clicked — which, for a lot of buying-research questions, is where AI answer engines actually operate. Someone asks ChatGPT "what's the best [category] for [use case]," gets three brand names back, and searches your competitor's name directly instead of clicking anything. That never shows up in your GA4 referral report at all.

This is the core reason store owners underestimate their AI exposure: they're measuring the click-through, not the citation. The two are related but distinct, and a store can be getting mentioned constantly while showing near-zero referral traffic in GA4. Referral tracking answers "is AI sending clicks." It can't answer "is AI citing me at all," "am I structured well enough to be cited," or "which competitors is AI recommending instead of me." Those require checking your store's actual structured data — the kind Google's structured data documentation says helps search systems understand a page — and content, not your analytics dashboard.

If GA4 is showing you close to nothing and you're not sure whether that means "AI isn't talking about us" or "AI is talking about us but not linking," that's exactly the gap a free StoreCited scan is built to close. It checks whether your product pages and product schema are the kind of source an AI assistant can cite in the first place — independent of what your analytics currently show — and names the competitors AI mentions instead of you.

Common Mistakes to Avoid

• Filtering only for chat.openai.com. That domain has largely been superseded by chatgpt.com — an outdated filter will show you near-zero traffic even if AI referrals exist.
• Treating a low GA4 number as proof AI traffic doesn't matter. Given the Direct-traffic blind spot, a low referral count is inconclusive, not a null result.
• Confusing AI Overviews clicks with AI Overviews visibility. Ranking in Google's AI Overviews is a separate, worthwhile goal — but the clicks you get from it are invisible inside GA4's normal Google organic bucket, so GA4 can't tell you whether your AI Overview strategy is working. You'll need Search Console's Search Appearance filters for that, and even those have gaps.
• Never checking manually. Referrer data is a proxy. Asking ChatGPT or Perplexity your customers' real questions, on a recurring schedule, remains the most direct way to know if you're showing up at all.

For deeper reading on the structured-data side of getting cited in the first place, see structured data for Shopify and answer engine optimization for Shopify.