How to Track AI Search Traffic in Magento 2

Last updated: July 2026. AI platform behavior changes frequently — referrer policies, crawler names, and feed protocols may have evolved since publication.

You updated your robots.txt. You generated llms.txt. You fixed your Product schema. Your AEO audit score went from 25% to 85%. And then you opened GA4 — and saw nothing unusual.

This is the most common frustration after implementing AEO on a Magento 2 store. Not because the work didn’t help, but because standard analytics tools were not built to track AI referral traffic. The signals exist — they are hidden in the wrong buckets.

This guide covers several methods to surface AI search traffic in Magento 2 — from GA4’s own native classification to server-side log analysis — so you can connect your AEO work to real business outcomes.

Why AI traffic disappears in your analytics

Each AI platform handles referral data differently, which is why a single tracking method won’t cover all of them:

The practical upside: Perplexity, Gemini, and Claude together likely represent more trackable eCommerce referral traffic than ChatGPT at this stage — and all three are addressable right now with GA4’s native classification and a five-minute custom Channel Group.

Method 0: Start with this — GA4’s native “AI Assistant” channel

Effort: 0 minutes  |  Covers: ChatGPT, Gemini, Claude out of the box (list expanding)

On May 13, 2026, Google added a native “AI Assistant” channel to GA4’s Default Channel Group. This is a significant change from the setup most AEO guides — including earlier versions of this one — were written around. You no longer need to build a custom Channel Group from scratch just to see ChatGPT, Gemini, and Claude traffic separated from generic Referral.

When GA4 detects a referrer matching a recognized AI assistant, it now automatically assigns:

• Medium: ai-assistant
• Campaign: (ai-assistant)
• Default Channel Group: AI Assistant — appearing alongside Organic Search, Direct, Referral, and Paid Search in standard Acquisition reports
• Source: remains the originating AI platform’s domain (for example, chatgpt.com or gemini.google.com) — so you can still break results down by individual platform within the AI Assistant channel

No configuration is required. If your GA4 property hasn’t received the update yet, nothing is broken on your end — Google rolled this out gradually across properties, with wider availability reached over the following weeks into early June 2026.

How to verify the AI Assistant channel is working on your property

1. Reports → Acquisition → Traffic acquisition

   Open the standard Traffic acquisition report in GA4.

2. Set the primary dimension to “Session default channel group”

   This is the dimension dropdown at the top of the report table.

3. Look for a row labeled “AI Assistant”

   If it’s present, the channel is live on your property and already capturing qualifying sessions.

4. If it’s missing, check date range first

   Expand to the last 30–90 days. If it’s still absent, the rollout may not have reached your property yet, or you may genuinely have zero qualifying sessions in that window — both are normal at this stage.

Native channel vs. custom Channel Group, side by side

Method 1: Custom Channel Grouping in GA4

Effort: 5 minutes  |  Covers: Perplexity and other platforms outside GA4’s native list, plus historical trend continuity

Even with the native AI Assistant channel now live (see Method 0 above), a custom Channel Group is still worth setting up — mainly to cover platforms Google hasn’t officially added yet (Perplexity is the most common gap), and to keep a continuous trend line since the native channel doesn’t apply retroactively. Think of this as a complement to the native channel, not a replacement for it.

1. GA4 Admin → Data display → Channel groups → Create new channel group

   This is in the Property column of GA4 Admin, not the main left nav.

2. Name the group “AI Search” and set condition type to Session source

   Use contains as the operator — one rule per domain.

3. Add all known AI referrer domains from the table below

   New AI platforms appear regularly — revisit this list quarterly.

4. Save and wait 24–48 hours

   Channel groups apply to new sessions only — historical data is not reclassified.

Demo: GA4 Custom Channel Group setup — add one condition per AI domain with OR logic between them.

Method 2: UTM parameters — attributing at least part of ChatGPT traffic

Effort: 30 minutes  |  Covers: ChatGPT (partial attribution)

Since ChatGPT strips referrer headers on most link types, the most practical currently available method to attribute at least part of ChatGPT-driven traffic is adding UTM parameters to URLs exposed through AI discovery layers — such as llms.txt and experimental commerce feeds. When a user clicks a UTM-tagged link, GA4 receives the attribution string regardless of the missing referrer. This captures confirmed intentional clicks, not total AI-influenced traffic.

Add UTMs to llms.txt

Add UTMs to OpenAI experimental commerce feeds

If you are testing OpenAI-compatible commerce integrations or experimental shopping feeds — where your product data is submitted directly to OpenAI infrastructure — those product URLs should also carry UTM parameters. This is an evolving area and OpenAI has not published a stable, widely-documented merchant onboarding standard as of 2026, so treat any integration here as experimental:

Using distinct utm_source values (chatgpt vs chatgpt_shopping) lets you separate organic ChatGPT referrals from any feed-driven product clicks in GA4.

Method 3: Search Console Crawl Stats — AI bots as a leading indicator

Effort: 2 minutes  |  Covers: All major AI crawlers

Search Console does not show AI referral traffic directly. What it does show is AI crawler activity — and crawler visits are a leading indicator that precedes recommendation traffic by days or weeks. When GPTBot crawls a product page, it is building or updating its knowledge of your catalog.

1. Search Console → Settings → Crawl stats

   Bottom-left Settings panel, not the main navigation.

2. Click “See all crawl requests” and filter by user agent

   Look for the agents in the table below.

Demo data — illustrative crawl stats. GPTBot and OAI-SearchBot returning 200 status confirms your pages are accessible to OpenAI crawlers.

Method 4: Server-side access logs

Effort: 15 minutes  |  Covers: Everything, including ChatGPT — most accurate method

Server logs capture every HTTP request — including AI crawler visits and (where available) referrer headers from AI platforms. This is the most complete data source and the only one that can confirm GPTBot activity independently of Search Console.

Demo output — illustrative. High GPTBot frequency on product pages indicates those pages are strong candidates for ChatGPT recommendations.

For Hypernode hosting, logs are at /var/log/nginx/access.log with daily rotation. The same commands above work without modification. You can also use the Hypernode Control Panel under Logs → Access logs for a browser-based view.

Putting it together: AI traffic dashboard in GA4

1. GA4 → Explore → Blank exploration → name it “AI Search Performance”
2. Add dimensions: Session default channel group, Session source, Landing page
3. Add metrics: Sessions, Engaged sessions, Conversions, Revenue
4. Add a filter group with two conditions (OR logic): Session default channel group exactly matches “AI Assistant” OR Session default channel group exactly matches “AI Search” (your custom group). This surfaces both native and custom-tracked AI sessions in one view without double-counting, since a session can only belong to one channel group definition at a time.
5. Add date comparison: this month vs last month to track growth over time — keep in mind the native channel’s comparison will be limited until it has accumulated a few months of history

How much AI traffic should you expect? Benchmark by store type

One of the most common questions after setting up AI tracking is: “Are these numbers normal?” Here are approximate ranges based on community reports and early data from stores that have implemented AEO. Treat these as directional benchmarks, not guarantees — this space is moving quickly.

AI attribution will remain partially probabilistic

It is worth being clear about what this guide can and cannot give you. Setting up all methods described here will give you a significantly better picture of AI-driven traffic — but it will not give you perfect attribution. That gap is structural, not a setup problem.

What to do if you see no AI traffic at all

First: check the obvious AEO gaps

• robots.txt is blocking GPTBot or OAI-SearchBot — fix guide →
• llms.txt does not exist or returns 404 at yourdomain.com/llms.txt — generation guide →
• Product schema is missing offers.availability — ChatGPT Shopping skips these products — schema guide →
• Product descriptions are supplier copy — AI models cannot form a confident recommendation — description guide →
• AEO audit score above 70% and Channel Group set up — give it 4–8 weeks; AI crawl-to-traffic lag is real

Why AI traffic still shows as Direct (even after setup)

If you have confirmed AI bots are crawling your store but sessions are still landing in Direct rather than an AI channel, the issue is almost always one of these:

Frequently asked questions FAQ schema