ChatGPT Links Lost Their UTM Tags After the GPT-6 Release

After the GPT-6 release, OpenAI appears to have removed the UTM parameters previously included in links in ChatGPT responses. As Search Engine Roundtable reported on October 14, 2026, site owners now have a harder time tracking visits from ChatGPT separately: some may show up as direct traffic. The publication links the change to the GPT-6 release a week earlier but does not know whether it was intentional. If your ChatGPT traffic has fallen, check whether direct visits to the same pages have risen.

What Changed in ChatGPT Links

In June, OpenAI added UTM parameters to links in ChatGPT. The tag in the page URL helped analytics tools identify visitors who had clicked a link in ChatGPT. Now, observers say, the utm_source=chatgpt.com parameter has disappeared from those links. That removes one of the signals site owners could use to distinguish these visits from other traffic.

Cédric Hüsler of Adobe reported another change. In his observations, GPT-6 opens links in a new tab in a way that prevents the destination site from receiving the address of the page the visitor came from. Along with the UTM tag, that clue about the traffic source disappears. As a result, analytics may classify a visit as direct even though the user followed a link from ChatGPT.

The distinction matters: a change in how visits appear in a report does not necessarily mean the number of people clicking links has changed by the same amount. But a single metric cannot establish the opposite, either. Once a visit is classified as direct, that record alone cannot tell you with confidence whether it came from ChatGPT. It remains unclear whether this way of opening links will stay in place or whether OpenAI intended to change how those visits are tracked.

What the Traffic Comparison Showed

Hüsler examined around 500 pages where ChatGPT had accounted for at least 5% of visits before the change. He compared October 9–13 with the previous week. Across those pages, visits identified as coming from ChatGPT fell by roughly 31,000 per day. Direct visits rose by around 26,000 per day over the same period.

Direct visits increased on two out of three pages where the ChatGPT figure fell. A week earlier, the two trends coincided in about half of cases. According to Hüsler, search traffic fell by 7%—the same as traffic overall. These figures warrant a closer look at how visits are classified, but they do not justify attributing the entire increase in direct traffic to ChatGPT.

There is also an indirect observation about mobile visitors. Among those classified as direct traffic, the share who saw the consent banner as new visitors to the site rose from 36% to 44%. Among visitors from ChatGPT, that share was 55%. Hüsler believes the new direct visits resemble ChatGPT visits in their behavior. But similarity between groups does not reveal the source of each individual visit.

Other participants in the discussion have also reported a drop in ChatGPT traffic. Sébastien Billard, for example, wrote about a decline in visits from people and in ChatGPT-User activity, alongside a rise in activity from OpenAI’s search bot. These reports do not prove that the disappearance of the UTM parameter caused every change. They show why week-to-week comparisons need to account for a possible change in how source data is passed along.

What This Means for Site Owners and Marketers

A drop in the ChatGPT metric alone is not enough to conclude that actual visits have been lost. Without the former tag and source information, some visits may end up classified as direct traffic. A practical step is to compare changes in ChatGPT and direct visits to the same pages over the periods in question. That comparison may help reveal a shift in reporting, but it will not give you an exact count of visits whose source can no longer be identified.

For reporting, this is primarily a measurement limitation. If links used to pass a tag and no longer do, you cannot compare two periods as though the tracking method had stayed the same. Significant differences between ChatGPT and direct traffic figures call for particular care in interpretation: the data points to a possible connection, not the origin of every visit.

My Take

I wouldn’t conclude that a store has lost customers based on one ChatGPT chart. When you update a product catalog, you want to know where people find your products. But if the tracking method itself has changed, it makes more sense to calmly compare the figures for the same pages first, then decide what to do about promotion.

I wouldn’t write off search because of this, either: buyers still come from there. For a small store, it would be especially frustrating to spend time and budget fixing a “drop” that turns out to be a change in reporting rather than a decline in interest in its products.

Sources

Where the news comes from. The text is a retelling in the author’s own words; the facts come from the source, the opinion is the author’s.

1. OpenAI Drops UTM Parameters From ChatGPT Results (GPT-6) www.seroundtable.com

About the author

Leila V. Seidakhmetova

owner of a home goods shop

I sell tableware and textiles online, and I’m currently updating the catalog and working on search traffic. I like it when a contractor explains a solution without technical fog.

All posts by the author