How to Track AI-Driven Traffic & Sales in GA4

If you run an ecommerce brand or manage paid media, you’re probably keeping one eye on AI – and rightly so. The integration between OpenAI’s ChatGPT and Shopify let people browse and buy products straight from a chat window: no ads, no scrolling through Instagram, no search engine in the middle. Just a prompt like, “Find me the best men’s waterproof walking boots under £100,” and the AI hands back a recommendation – sometimes with a buy button attached.

Here’s the part most brands miss. If an AI is already sending you buyers, you probably can’t see them. That traffic hides in “Direct,” gets misattributed, or vanishes entirely. So this guide leads with the fix – how to track AI-driven traffic and sales in GA4 – then covers how to feed the machines so there’s something worth tracking in the first place.

How to Track AI-Driven Traffic and Sales in GA4

Optimising for AI is only half the job. The other half is visibility. If ChatGPT, Perplexity, Gemini or Copilot is driving traffic and conversions, you need to see it in your reports – not guess at it. Here’s the setup we use for clients at HOC.

How to Detect Each AI Traffic Source

Before you can track AI traffic, you need to know what it looks like in your data. Two signals matter: the referrer domain a real visitor arrives from, and the crawler user-agent that shows up in your server logs when an AI reads your site. They tell you different things – one is a customer arriving, the other is the machine deciding whether to recommend you at all.

AI platform	Referrer domain	Crawler / user-agent	Recommended UTM tag
ChatGPT	chatgpt.com (older: chat.openai.com)	OAI-SearchBot/1.4 (search), ChatGPT-User/1.0 (user click-throughs), GPTBot/1.4 (training)	?utm_source=chatgpt&utm_medium=ai_referral
Perplexity	perplexity.ai	PerplexityBot/1.0 (indexing), Perplexity-User/1.0 (user visits)	?utm_source=perplexity&utm_medium=ai_referral
Gemini	gemini.google.com (AI Overviews / AI Mode can surface as a google.com referral)	Google-Extended (robots.txt control token; pages are fetched by Googlebot)	?utm_source=gemini&utm_medium=ai_referral
Microsoft Copilot	copilot.microsoft.com (and bing.com)	Bingbot (Copilot draws on Bing’s index; there is no separate Copilot crawler)	?utm_source=copilot&utm_medium=ai_referral

One caveat worth stating plainly: referrer coverage is patchy. The user-agent tokens above are published by OpenAI and Perplexity; the referrer domains are the ones GA4’s own AI Assistant channel now recognises. If you spot a domain in your logs that isn’t listed here, treat it as “verify in your own data” and add it, rather than assume.

Set Up an “AI Assistants” Channel in GA4

Two ways to do this, and you’ll probably want both.

Use the native channel. In May 2026, Google added an AI Assistant channel to GA4’s default channel group. When a session arrives with a referrer that matches a recognised assistant – ChatGPT, Gemini, Copilot, Claude, Perplexity – GA4 tags it with the medium ai-assistant and files it under AI Assistant automatically. Head to Reports > Acquisition > Traffic acquisition and switch the dimension to Session default channel group to see it.

Build a custom channel group for control. The native channel is a floor, not a ceiling. To catch platforms Google hasn’t added yet, create a custom channel group (Admin > Data display > Channel groups), add a channel called “AI Assistants,” and match Session source with matches partial regex:

chatgpt\.com|chat\.openai\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai

Keep that string updated as new tools appear. Our sibling guide on tracking LLM referral traffic in GA4 walks through the full report build.

Tag every link you control. Wherever you can add a URL to an AI plugin, action or feed, append UTM parameters so the source is unambiguous:

?utm_source=chatgpt&utm_medium=ai_referral&utm_campaign=product_recommendation

Pick one utm_medium value (ai_referral works well) and reuse it across platforms so your reporting stays consistent. While you’re in there, check Admin > Data streams > Configure tag settings > List unwanted referrals to be sure you aren’t accidentally excluding these AI domains from attribution.

How to Separate AI-Assisted from AI-Direct Sales

Even with all of the above, a big chunk of AI traffic won’t carry a referrer. When someone clicks through from an AI’s in-app browser on iOS or Android, the operating system often strips the referrer, and the visit lands in Direct with no fingerprint at all. Industry measurement through 2026 suggests a large share of AI referrals arrive this way – so treat your AI channel as a floor, not the full picture.

Two things sharpen the view:

Server-side tracking. AI clients frequently don’t run client-side scripts, so cookie-based tags miss the event. Capturing conversions server-side – via server-side Google Tag Manager, Shopify’s Customer Events (web pixel) or a Conversions API – records the sale even when the browser tag never fires. Capture the order, its value, and any landing URL with UTMs intact.

Log-file analysis. Your server logs are the one place AI activity can’t hide. Filtering for the crawler user-agents in the table above tells you whether OAI-SearchBot, PerplexityBot and the rest are actually reading your product pages – the leading indicator that you’re even eligible to be recommended, long before a sale shows up.

And the simplest early warning of all: conversion rate from new users climbs, with no new campaigns to explain it. If sales start arriving from visitors with no prior engagement, there’s a good chance AI has entered the chat.

How to Prepare Your Product Feed for AI Commerce

Tracking only pays off if there’s AI traffic to track – and that depends on whether machines can understand what you sell. Your product feed is no longer just for Google Merchant Centre or Meta Commerce Manager; it’s the raw material AI uses to decide what to recommend. Get the fundamentals right:

• Specific titles and descriptions – real features and keywords, not fluff
• Detailed attributes – size, colour, material, use case, benefits
• Accurate, real-time inventory and pricing – no surprises, no stale data
• Clean, multi-angle images on plain backgrounds
• Structured tags and metafields – “sustainable material,” “best for winter,” “ideal for travel” all help AI categorise you

If you’re juggling feeds across Google, Meta, TikTok and beyond, a feed management tool – Popsixle, Feedonomics, Channable or DataFeedWatch – keeps the data clean and consistent, and increasingly that includes AI ecosystems too. Our guide to optimising product data for AI results goes deeper on getting feeds AI-ready. It’s the same principle we apply as an AI-native ads agency: make yourself easy for the algorithm to understand.

The Future of Ecommerce Is Quietly Arriving

We’re watching a shift from attention-based advertising to AI-led matchmaking. The brands that win won’t be the ones shouting loudest – they’ll be the ones giving the clearest answers, and measuring the results while everyone else is still guessing.

This isn’t science fiction. It’s the slow, steady evolution of how people shop online. The question is whether you wait and react, or set yourself up to see it coming.

FAQ

Does GA4 track ChatGPT and other AI traffic automatically?

Partly. Since May 2026, GA4 has a native AI Assistant channel that auto-detects sessions referred by ChatGPT, Gemini, Copilot, Claude and Perplexity. But it only catches sessions that arrive with an intact referrer – anything stripped by an in-app browser still lands in Direct. A custom channel group plus server-side tracking fill the gaps.

What referrer domains should I look for?

chatgpt.com (and the older chat.openai.com), perplexity.ai, gemini.google.com and copilot.microsoft.com are the main ones. Gemini-driven AI Overviews and AI Mode can also show up as a google.com referral. Add any new AI domains to your regex as you spot them in your own data.

What’s the difference between an AI referrer and an AI crawler?

A referrer domain means a real person clicked through to your site from an AI answer. A crawler user-agent – like OAI-SearchBot, PerplexityBot or Bingbot – is the machine reading your pages to decide whether to cite you. You track referrers in GA4; you track crawlers in your server logs.

Why does so much AI traffic show up as “Direct”?

Many AI apps open links in in-app browsers that don’t pass a referrer, so the session has no source and GA4 files it under Direct. Server-side tracking and log-file analysis are the most reliable ways to recover that lost visibility.

Which UTM tags should I use for AI links?

For any link you control, keep it consistent: set utm_source to the platform (chatgpt, perplexity, gemini, copilot) and use a single shared utm_medium such as ai_referral. Consistency matters more than the exact wording – it’s what keeps your reporting clean over time.