Google Analytics 4 has stopped hiding your AI traffic. Sessions that arrive from ChatGPT, Gemini, Claude, Copilot, Deepseek and Grok now land in a dedicated AI Assistant channel instead of getting dumped into the Referral bucket with every random blog and forum link. Practical Ecommerce published the first data-backed operator walkthrough of the channel on July 13, after sitting on two months of live data before writing about it.
That walkthrough matters more than the feature announcement did, because it exposes what the feature actually does and does not capture. If you run a high-ticket store, this is the first time you can open a report and see how many people asked an AI assistant about a $4,000 product, clicked through to your site, and started a quote. Here at Ecommerce Paradise I have been telling clients for a year that AI answers were quietly feeding their phone calls. Now there is a number attached to it.
There are three catches, and they are big enough that most store owners are going to misread their own data on day one. The channel only counts forward from May 13, 2026. Perplexity is not in it. And clicks from Google’s own AI Overviews and AI Mode are not in it either, because Google counts those as Organic Search.
Below is what changed, why it took Google this long, what it means for a store selling $2,000 to $10,000 products on a six-week consideration cycle, and the exact reports to build this week.
Google can now tell you which AI sent your buyer. Your LLC filing still tells anyone who looks where you sleep at night. Northwest puts their address on your public filings instead of yours →
GA4 Now Reports ChatGPT and Gemini Traffic as AI Assistant
Google added the AI Assistant channel to Default Channel Group reports on May 13, 2026, according to Search Engine Journal. It reached most GA4 properties by early June. Nobody had to configure anything, and most store owners never noticed it appear.
The mechanic is simple. When GA4 sees a referrer header from a recognized AI assistant, it stamps the session with a medium value of ai-assistant and files it under the AI Assistant channel. Google’s launch post named ChatGPT, Gemini and Claude. The live Default Channel Group documentation now also lists Deepseek, Copilot and Grok.
You find it in Reports, then Acquisition, then Traffic acquisition, under the Session default channel group dimension. It sits right next to Direct, Organic Search and Paid Search. Per Practical Ecommerce, the report gives you engagement rate, events per session, average time per session and the rest of the standard metric set, which means you can finally compare an AI-assistant session against an organic session on quality rather than guessing.
Perplexity Is Still Filed as Referral
Perplexity is absent from Google’s official channel definition. Those sessions still land in Referral. That is not a small omission for a high-ticket store, because Perplexity’s shopping layer has been one of the more aggressive AI buying surfaces this year, and it skews toward exactly the kind of comparison-shopping behavior that precedes a large purchase.
So the AI Assistant number in your report is not your AI number. It is a subset. Treat it as a floor.
AI Overviews and AI Mode Are Counted as Organic Search
This one trips people up. Google does not include AI Overviews or AI Mode clicks in the AI Assistant channel. Per Practical Ecommerce, those clicks are reported as Organic Search. So if you were expecting the new channel to show you the damage or the upside from Google’s own AI answers, it will not. That data lives in Search Console’s Generative AI section, not GA4.
Referrer-Less AI Traffic Still Lands in Direct
The channel is entirely dependent on the referrer header. AI traffic that arrives without one, which happens with in-app browsers, mobile apps, and anyone who copies a link out of a chat and pastes it into a fresh tab, still gets counted as Direct. On a high-ticket product, copy-and-paste is common behavior. Somebody asks Claude which brands make a commercial-grade unit, gets three names and a link, and pastes it into a browser because they are on their work laptop.
Those buyers are real. They are just not in the report.
Why AI Chatbot Clicks Hid Inside GA4’s Referral Bucket for a Year
Until May, every ChatGPT click looked identical to a click from a random directory. GA4 had no bucket for it, so it swept those sessions into Referral, where they sat next to spam, forums, and the one guy who linked to you from a Reddit thread in 2023. If you wanted to isolate them, you had to write a regular expression against session source and medium yourself.
Plenty of operators did exactly that, and the regex approach still works. Practical Ecommerce publishes a working pattern that catches chatgpt.com, perplexity, copilot.microsoft.com, openai.com, gemini.google.com, claude.ai, grok.x.ai and edgepilot. You paste it into a Session source/medium filter with a “matches regex” match type and you get a platform-by-platform breakdown, which is more than the AI Assistant channel gives you, since the channel does not tell you which assistant it lumped together.
The reason this got fixed in 2026 and not 2025 is that the platforms finally have a commercial reason to measure it. Amazon disclosed on its Q1 earnings call that nearly 20% of shoppers who interact with a sponsored brand prompt in Rufus keep talking about that brand, per Marketplace Pulse. Amazon is now selling its AI agent technology to other retailers. When AI answers become an ad surface, the referral data becomes a product. Reporting follows money, and it always has.
Google Search Console shipped its Generative AI section in June. Bing improved its AI performance report the same month. GA4’s AI Assistant channel is the third leg of that.
What GA4’s AI Assistant Channel Means for High-Ticket Stores
Here is the part that costs money if you get it wrong.
A high-ticket buyer does not behave like a $40 impulse buyer. On a $3,000 to $8,000 cart, you are looking at a two to six week consideration window, four to nine touchpoints, and a phone call somewhere in the middle. That is the model I teach in my complete guide to high-ticket dropshipping, and it is why last-click attribution has always lied to us.
AI assistants slot into the research phase of that window, not the buying phase. Somebody asks ChatGPT “what’s the difference between a 3-ton and a 4-ton mini split” and gets an answer that cites your buying guide. They click. They read. They leave. Three weeks later they come back through a branded Google search or type your URL directly, and they call your number and buy.
Under the old setup, that first click was a Referral session with zero revenue attached and a bounce, and the sale got credited to Direct or branded search. If you were pruning content based on revenue per page, you were deleting the exact articles feeding the AI answers that started the whole chain. I have watched clients do it.
The Number That Actually Tells You Something
Do not look at AI Assistant sessions. Look at the ratio of AI Assistant engagement rate to organic engagement rate on the same pages.
Practical Ecommerce found something worth acting on: in Ann Smarty’s tests, the top pages by organic traffic and the top pages by AI-assisted traffic did not overlap. Different pages win in each channel. If that holds on your store, and on the two client accounts I checked yesterday it did, then your AI-cited pages are a separate content asset that your organic reporting has never surfaced.
Those pages are typically specification comparisons, sizing guides, and “which model do I need” content. They are the pages that read like an answer. They are not the pages that rank for a fat commercial keyword, which is why they were invisible.
The Attribution Math on a $5,000 Cart
Rough numbers. Say you do 60 sales a month at a $5,000 average order value, which is $300,000 in revenue. If 4% of your first-touch sessions come from an AI assistant, and those sessions convert at your normal blended rate after multiple touches, you are looking at roughly two to three sales a month, $10,000 to $15,000, that had an AI answer as the origin point.
Nobody was crediting that. It was sitting in Referral looking like junk traffic. Now you can price it, and once you can price it, you can decide whether to fund more of the content that produces it. If you are under 20 SKUs and running a single niche, this is a straightforward exercise you can do in an afternoon. Over 200 SKUs across multiple brands and the reporting alone becomes a real job, which is where most operators stall out.
If you would rather have the whole measurement and content system built and run for you than spend your Saturday inside GA4, that is exactly what my team does with the turnkey done-for-you store build and management service. We set the tracking up correctly on day one, so you are not reverse-engineering six months of misattributed sessions later.
New to this and want the whole high-ticket model laid out before you touch analytics? Grab my free high-ticket dropshipping mini course →
How to Set Up GA4 AI Assistant Tracking and Regex This Week
Six things, in order. Most of them take under ten minutes.
- Open the AI Assistant channel and set your baseline. Reports, Acquisition, Traffic acquisition, Session default channel group. Pull the last 30 days. Write down sessions, engagement rate and average engagement time. That is your floor, and remember the data only starts May 13, so do not bother pulling a year.
- Filter your pages report to AI Assistant and find your cited pages. Go to Reports, Engagement, Pages and screens. Add a filter with Session default channel group as the dimension, “Exactly matches” as the match type, and AI Assistant as the value. The pages that come back are the pages AI assistants are citing. Print that list. It is your new content roadmap.
- Build the regex report so you can see Perplexity. In Traffic acquisition, add a filter on Session source/medium, match type “matches regex”, and paste in a pattern covering chatgpt.com, perplexity, copilot.microsoft.com, openai.com, gemini.google.com, claude.ai and grok.x.ai. This is the only way to get a platform-level breakdown, and it catches the Perplexity traffic the official channel misses.
- Reverse-engineer the prompts. Take your top five AI-cited pages and run them through Semrush to see what questions and topics they map to. Practical Ecommerce recommends the same move. Then write three more pages that answer adjacent questions in the same format, because the format is what got you cited, not the keyword. My KWFinder setup works for this too if you want a cheaper option.
- Make the AI-cited pages convert. These pages were never built to sell, because you did not know anyone was reading them. Add a quote form, add your phone number, add live chat with Tidio, and put an email capture on them through Omnisend so a researcher who is six weeks from buying does not vanish. On Shopify this is a 20-minute template edit.
- Answer the phone. AI-assistant researchers call. They arrive already educated and they want a human to confirm the spec. A dedicated business line through Grasshopper costs almost nothing and it is still the highest-converting asset in a high-ticket store. If you are drowning, a trained VA from OnlineJobs.ph can handle first-touch qualification.
If you want somebody to look at your specific numbers and tell you which of your pages are worth doubling down on, that is what one-on-one coaching is for. Bring the AI-cited page list and we will build the plan around it.
Frequently Asked Questions
Do I need to turn the AI Assistant channel on?
No. Google enabled it automatically on May 13, 2026, and it reached most properties by early June. If you do not see it, check that you are looking at Session default channel group and not a custom channel group you built yourself.
Why is my AI Assistant traffic so low?
Because it undercounts. Perplexity is still classified as Referral, AI Overviews and AI Mode clicks are counted as Organic Search, and any AI traffic without a referrer header falls into Direct. The real number is higher than what the channel shows.
Can I see historical AI traffic?
Not through the channel. It is forward-only from May 13, 2026, and Google did not reclassify old sessions. Your earlier AI traffic is still sitting in Referral, and the regex filter on Session source/medium is the way to dig it out.
Does this replace Search Console for AI visibility?
No. GA4 shows you clicks that landed on your site. Search Console’s Generative AI section shows you visibility inside Google’s own AI surfaces. You need both, and they measure different things.
Should I write content specifically for AI assistants?
Write content that answers a specific buyer question in a structure a machine can quote. Spec comparisons, sizing guides and decision frameworks get cited. Thin listicles do not. Tools like Claude and ChatGPT are useful for drafting, but the citation comes from being genuinely useful and specific.
I am just starting out. Does any of this matter yet?
Set the tracking up on day one so you have clean data in six months, then go build your product catalog and your supplier relationships. Start with my guide on finding high-ticket suppliers and pick a vertical off the high-ticket niches list.
Does my business structure affect any of this?
Not the tracking, but if you are running a store without an LLC and a registered agent you have a bigger problem than attribution. Read my breakdown on business formation for high-ticket dropshipping before you scale ad spend.
Want my private weekly breakdowns and store teardowns? Join the Patreon →
Go pull the AI Assistant report before you do anything else today. It takes four minutes and you will probably find two or three pages you had written off as dead weight that are quietly feeding your best phone leads. Subscribe to the YouTube channel for daily breakdowns. More breaking news later today.
Related Articles
If this was useful, these go deeper:
- Google Now Scores Your Store Inside AI Shopping
- AI Agents Now Buy for Shoppers. Own Your Checkout
- Google Shopping Ads Setup for High-Ticket Dropshipping
- How to Get Reviews for Your High-Ticket Dropshipping Store
- 10 Ways to Scale Your High-Ticket Dropshipping Business in 2026
Trevor Fenner is an ecommerce entrepreneur and the founder of Ecommerce Paradise, a platform focused on helping entrepreneurs build and scale profitable high-ticket ecommerce and dropshipping businesses. With over a decade of hands-on experience, Trevor specializes in high-ticket dropshipping strategy, niche and product selection, supplier recruiting and onboarding, Google & Bing Shopping ads, ecommerce SEO, and systems-driven automation and scaling. Through Ecommerce Paradise, he provides free education via in-depth guides like How to Start High-Ticket Dropshipping, advanced training through the High-Ticket Dropshipping Masterclass, and fully done-for-you turnkey ecommerce services for entrepreneurs who want a faster, more hands-off path to growth. Trevor is known for emphasizing sustainable, real-world ecommerce models over hype-driven tactics, helping store owners build scalable, sellable, and location-independent brands.
