By Elena Vasquez

ChatGPT-attributed sessions in Google Analytics 4 fell between 22% and 57% starting Oct. 9, according to Graphite CEO Ethan Smith, who estimated the drop at roughly 45% below the previous baseline. Independent measurement from LLM Pulse found sessions from ChatGPT down about 46% during Oct. 9–11 on its monitored sites, while total traffic on those sites stayed flat.

The cause appears to be tracking, not demand. Smith says ChatGPT began opening links in answer text without the utm_source=chatgpt.com tag (a URL parameter that marks a visit's origin) and without a referrer. Such visits then land in GA4's Direct channel. Smith called the change a bug on OpenAI's side and wrote that he expects OpenAI to fix it "very soon."

Where did the visits go?

Adobe senior director of product management Cedric Huesler compared Oct. 9–13 with the prior week across roughly 500 pages where ChatGPT drove at least 5% of visits. ChatGPT visits fell by about 31,000 a day. Direct visits rose by about 26,000 a day. Search traffic fell 7%, in line with search trends elsewhere.

Direct rose on two of every three pages that lost ChatGPT traffic, versus about half the week before. Huesler says the new Direct visitors also behave like ChatGPT visitors. On mobile, the share of Direct visitors who saw the consent banner — his proxy for a first visit — rose from 36% to 44%, moving toward the 55% he observes among ChatGPT visitors. He calls the findings early, based on four days of sampled data.

What changed in ChatGPT's links?

Huesler ties the shift to the GPT-6 rollout, which reached paid plans on Oct. 7 and Free and Go users on Oct. 8. After the rollout, ChatGPT opened links in a new tab with a browser setting that withholds the referrer, and dropped the utm_source tag that ChatGPT citations had carried since its search launched in 2024.

Smith added a key detail:

• Inline links in answer text lost the tag.
• Links in the Sources panel kept it.
• Shopping links kept it.
• 95% of mobile ChatGPT visits to Graphite's sites relied on the tag alone, making app traffic the most exposed.

Profound connects the change to a new answer format. Among ChatGPT answers with citations that it collected, the share with at least one tagged source link fell from about 98% on Oct. 8 to 66% by Oct. 14. Over the same period, Intelligent UI — ChatGPT's format for charts, comparisons and small tools — went from 2% of collected answers to about 35%. Links reached through a product or information panel still carried tags.

Did ChatGPT also link differently?

Tomek Rudzki at Peec AI ran the same 1 million prompts from Oct. 2–7 and again from Oct. 9–13. He found ChatGPT placed a direct URL for a webpage in 42% of responses, up from a previous rate of about 8–10%, and that 54% of those URLs point to home pages. Those links carry no tags or referrers. The figures count links displayed by ChatGPT, not actual user click-throughs.

What the data does not settle

The LLM Pulse team reported no detectable increase in Direct traffic, because Direct is much larger than ChatGPT and a small session shift would be hard to spot. The team also stated that aggregated data cannot distinguish traffic that was lost from traffic no longer labeled as ChatGPT.

Confirmation came on Oct. 14, when Huesler said OpenAI confirmed to him the change was a bug. He then reported that a fix began rolling out early on Oct. 15, and that by 9 a.m. PT tagged ChatGPT clicks were back to roughly 95%–100% of the same hour a week earlier across thousands of domains he tracks. SEO consultant Glenn Gabe replied that he had not seen the change yet on paid or free accounts. Profound says it has seen no official word on a fix, and OpenAI's ChatGPT release notes for Oct. 1–15 do not mention link tracking. Graphite, LLM Pulse, Profound and Peec AI sell SEO or AI search services, and the LLM Pulse and Profound posts give no sample sizes.

Why this matters for marketers

A 50% decline in GA4 ChatGPT referrals in one week likely signals a reporting change, not a collapse in visitor interest. Smith and Huesler both advise annotating Oct. 8 — Huesler now frames the window as Oct. 8–15 — before anyone reads the data as lost traffic. Until the fix lands everywhere, review ChatGPT and Direct traffic side by side for each page ChatGPT referred visitors to, and check the homepage, where Rudzki found more than half of direct links point.

UTM parameters (custom URL tags) that marketers set on clicks from ChatGPT Ads should be unaffected, according to Smith. The open question now is how quickly tags return for every account, since OpenAI has not mentioned the fix in its release notes — and restoring tags will not recover attribution already lost. Gemini moved the opposite direction two weeks ago, adding UTM parameters to its links.