ChatGPT Stopped Sending UTM Parameters in October 2026: What It Means for Shopify Attribution

Starting around October 9, 2026, Shopify merchants and other site operators began reporting a sharp drop in ChatGPT referral sessions, sometimes by a third or more, with no matching drop in overall site traffic. If you track AI-referred revenue and your "ChatGPT" line item fell off a cliff that week while total sessions held steady, this is almost certainly why: ChatGPT changed how it passes data to the sites it links to, not how often it links to them.

What changed on October 9, 2026

The timing lines up with OpenAI's GPT-6 rollout. Prior to this release, links ChatGPT generated in response to shopping and research queries carried a utm_source=chatgpt.com parameter and passed a standard referrer header when a user clicked through. Both of those are how GA4, Shopify Analytics, and most other tools identify a session as coming from ChatGPT.

Reports that surfaced in mid-October 2026 pointed to two specific changes in how GPT-6 opens outbound links:

• The utm_source=chatgpt.com parameter is no longer appended to links.
• Links now open using noopener,noreferrer, which strips the referrer header before the browser ever sends the request to your site.

Neither change affects whether the user clicks the link. Both affect whether your analytics can tell that they did.

Why this looks like a traffic crash but probably isn't

When a session carries no UTM and no referrer, GA4 and Shopify Analytics don't drop it. They bucket it as Direct traffic, the same category reserved for typed URLs and bookmarks. One analysis of several hundred pages that previously received meaningful ChatGPT referral volume found daily ChatGPT-attributed visits falling by tens of thousands, with Direct traffic on those same pages rising by a comparable amount on roughly two-thirds of them.

That pattern is the tell. A real falloff in buyer interest would show up as a drop in total sessions, not a shift between two referral categories. If your ChatGPT number fell sharply in the days after October 9 but your Direct traffic rose by something close to the same amount, and your total sessions stayed flat, the underlying visits are still happening. What's gone is the label.

How to check whether this happened to your store

Three quick checks, in order of how fast they are to run:

1. Compare total sessions before and after October 9. If total traffic is roughly flat week over week, you're looking at a relabeling problem, not a referral collapse.
2. Overlay your Direct traffic trend against your ChatGPT referral trend for the same date range. A visible rise in Direct that mirrors the drop in ChatGPT is the strongest signal this is the UTM change, not a real decline.
3. Check mobile session share within Direct traffic. ChatGPT's mobile app and in-app browser sessions were already prone to referrer stripping before this change, so a disproportionate mobile share inside your "new" Direct traffic is another consistent marker.

None of these checks require new tooling. GA4's existing session-source reports and Shopify's native Acquisition report both have the data; you're just looking at it with a different question in mind.

Is this only ChatGPT?

As of mid-October 2026, yes, at least based on what's been reported. Perplexity, Claude, Gemini, and Copilot haven't shown the same referrer-stripping behavior, and their outbound links still carry identifiable referral data in most contexts.

Platform	UTM parameter on outbound links	Referrer header passed
ChatGPT (post-GPT-6)	No	No (noopener,noreferrer)
Perplexity	Yes	Yes
Claude	Yes	Partial (desktop app has long stripped referrers)
Gemini	Yes	Yes
Copilot	Yes	Yes

That Claude caveat isn't new. Claude's desktop app has never reliably passed a referrer header, which is a separate and older problem from what just happened with ChatGPT. The difference now is that ChatGPT, previously the most reliably tracked of the five, has moved toward the same blind spot.

What this means going forward

Referrer and UTM data were never guaranteed. They're conventions that a platform can change with a single release, and nothing requires OpenAI, Google, or Microsoft to keep passing them in a format your analytics tools expect. This isn't the first time a browser or app has stripped referrer data, and it won't be the last.

For merchants who need a durable read on AI-referred revenue regardless of what any single platform's links do next, the alternative is attribution that doesn't depend on a UTM parameter surviving the trip: order-level tagging based on session behavior rather than referral headers alone. That's a different problem from the one this post is about, but it's the direct consequence of relying on a tracking method that one product update can quietly turn off.