There's a traffic source sending visitors to your store that you almost certainly aren't tracking. It doesn't show up clearly in your channel reports. It probably looks like Direct traffic. And it's growing faster than any other acquisition channel right now.
It's ChatGPT. And Perplexity. And Gemini. And a half-dozen other AI tools that hundreds of millions of people use every day to ask questions like "what's the best skincare brand for dry skin" or "where can I buy minimalist home goods online."
When those tools recommend your store and a user clicks through, that session lands in your GA4. You're just not seeing it for what it is.
How Big Is AI Referral Traffic Actually?
The numbers are striking once you see them together.
According to Adobe's 2025 Digital Economy report, AI referral traffic to US retail websites grew 10x between July 2024 and February 2025. That's not a typo — ten times in seven months.
A study by SE Ranking tracking 400+ websites found AI-driven traffic grew 527% in the first five months of 2025 alone. ChatGPT accounts for roughly 77–87% of all AI referral traffic. Perplexity comes in second at around 15%. Google's Gemini is third and growing fast.
The kicker from Digiday: ChatGPT generated 243.8 million referral visits in April 2025 alone.
And AI-referred traffic converts well. ChatGPT-referred sessions convert at 15.9% — higher than typical organic search. Perplexity-referred sessions convert at 10.5%. These are people who asked a question, got a recommendation, and clicked through with intent.
This traffic is real. It's growing. And most store owners are either missing it entirely or watching it vanish into their Direct bucket.
Why It Shows Up as Direct (or Not at All)
The referral attribution problem with AI traffic comes from a few different mechanisms:
The copy-paste problem. ChatGPT often mentions a brand or product without hyperlinking it. The user reads the recommendation, types the URL directly, and GA4 records it as Direct. GA4 has no way to know that a recommendation was the catalyst.
Mobile app behavior. When someone uses the ChatGPT mobile app and taps a link, it often opens in the device's native browser rather than an in-app browser. The Referer HTTP header — the signal GA4 uses to identify where traffic came from — is typically stripped in that transition. GA4 sees no referrer. It records the session as Direct.
Free tier limitations. ChatGPT's base free tier didn't consistently pass referrer data for most of its history. The paid tier with Browse mode (and the newer Atlas browsing agent) is more reliable. As of February 2025, ChatGPT Search became free for all users globally — which is when referral traffic started being more consistently trackable.
HTTPS redirect stripping. If there's a redirect chain anywhere on your site (especially HTTP to HTTPS), browsers drop the referrer header per the web spec. The traffic arrives with no origin signal.
The practical result: GA4 significantly undercounts AI referral traffic. What you can see and attribute to chatgpt.com / referral is a fraction of the actual AI-influenced traffic hitting your store.
This is the same problem as Direct traffic more broadly — if you've struggled to figure out why your Direct number is so high, AI traffic is almost certainly part of it. Our guide on why Shopify traffic shows as Direct in Google Analytics covers the full picture of what's hiding in that bucket.
How to Find AI Traffic in GA4 Right Now
Quick check:
- Go to Reports > Acquisition > Traffic Acquisition
- Set the primary dimension to Session source / medium
- Search for:
chatgpt.com / referralchat.openai.com / referral(older sessions)perplexity.ai / referralclaude.ai / referralgemini.google.com / referral
If you're seeing any sessions from these sources — even a handful — you're getting AI traffic. Most store owners who do this check for the first time are surprised at what they find.
What to look at once you find it:
Don't just look at the session count. Look at the quality metrics alongside it — engagement rate, average session duration, conversions, revenue. Compare these against your other acquisition channels. AI referral traffic often shows strong engagement numbers because users arrive with specific intent from a direct recommendation.
How to Set Up a Proper AI Traffic Channel Group
The problem with finding AI traffic by scanning source/medium manually is that it's fragmented across a growing list of AI tools, and new ones appear regularly. A better approach is to group them all into a single channel in GA4.
Creating a custom channel group:
- In GA4, go to Admin > Data Display > Channel Groups
- Click Create new channel group
- Name it "AI Referral" or "AI Traffic"
- Click Add new channel and name it "AI Chatbots"
- Set the condition: Source matches regex:
^(chatgpt\.com|chat\.openai\.com|openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|you\.com|phind\.com)$
- Drag this channel above the default "Referral" channel in the priority order — this ensures GA4 applies your rule first
- Save
Once saved, this channel group applies retroactively to your historical data. You can now use "AI Chatbots" as a channel in any report or exploration, and it will catch all the sources in your regex — including Perplexity, Claude, and Gemini alongside ChatGPT.
Update the regex as new AI tools become significant traffic sources for your store.
The OpenAI Feature That Changed Everything
Understanding why this is happening now — and why it'll keep growing — requires knowing what OpenAI actually launched.
In October 2023, OpenAI added Browse with Bing for ChatGPT Plus users — the first version of real-time web search with clickable citations. In October 2024, they launched ChatGPT Search, a dedicated search mode with sourced results designed to compete with Google directly.
The critical turning point: in February 2025, ChatGPT Search was made free for all users globally. That's when the traffic spike started showing up in data across thousands of websites simultaneously.
More recently, OpenAI's ChatGPT Atlas (their browsing agent mode) passes referrer headers more reliably than any previous version — meaning more of this traffic is now visible in GA4 as attributed referral sessions rather than Direct.
The trajectory is clear. AI chatbots are actively replacing Google as the first stop for product discovery questions. They're faster, more conversational, and increasingly accurate. The store owners who understand this shift early — and optimize for it — will have a structural advantage.
What This Means for Your Shopify Store
If you're already getting AI traffic: You have evidence that ChatGPT or Perplexity is mentioning your store in response to relevant queries. That's valuable signal. Look at which pages these sessions land on — are they hitting a product page directly? A category page? Your homepage? That tells you what the AI is recommending about you specifically.
If you're not seeing AI traffic yet: That's information too. It may mean your store isn't appearing in AI recommendations for your category — or it means the traffic is there but falling into Direct because of the attribution gaps described above.
Don't conflate zero GA4 AI traffic with zero AI influence. A customer could see your store mentioned in ChatGPT, think about it for a day, and come back directly tomorrow. GA4 records that as Direct. The real footprint of AI influence on your traffic is larger than what's trackable.
For SEO and content strategy: Being recommended by ChatGPT and Perplexity increasingly correlates with having clear, authoritative, well-structured content that these models can cite confidently. The fundamentals — clear product descriptions, customer reviews, detailed FAQs, strong about-page content — are the same things that make an AI want to recommend you.
For context on what good traffic attribution looks like once you have all your channels properly identified, see our guide on how to read the Traffic Acquisition report in Google Analytics.
The Attribution Problem Isn't Going Away
AI referral traffic is the most visible example of a broader issue: a growing percentage of your customers arrive having been influenced by something GA4 can't track. ChatGPT recommendations. Word of mouth. A podcast mention. Someone who saw your ad, didn't click, remembered you later.
GA4 does a good job tracking what it can track. But it was built for a world where almost all discovery happened through clickable links in browsers. That world is changing.
What you can do: set up the custom channel group, monitor AI traffic month over month, and treat the Direct bucket not as "people who know us" but as "people we influenced somewhere we can't see." Understanding that distinction changes how you interpret all your acquisition data.
Wardly surfaces these hidden traffic patterns automatically. Instead of manually scanning source/medium breakdowns or building custom channel groups from scratch, Wardly connects to your Google Analytics and surfaces insights like "Your AI referral traffic is up 43% this month — here's what's driving it and which pages they're landing on" directly inside your dashboard, with no digging required.
Wardly connects to your Google Analytics and gives you an interactive dashboard with ready-made reports — what went wrong, why it happened, and how to fix it.