Why your ChatGPT traffic dropped in GA4 in October 2026

Key takeaways

• Graphite saw ChatGPT referral sessions fall about 45% from October 9, and LLM Pulse saw about 46% fewer while total traffic on their sites held steady.
• The links in ChatGPT's answer text lost their tracking tag and their referrer, so GA4 counts those clicks as Direct.
• Links in ChatGPT's Sources panel and shopping links kept their tag, and Graphite's Ethan Smith expects advertisers' own UTMs on ads to be unaffected.
• A fix began rolling out on October 15. When I tested logged out that evening, none of the 25 answer links I checked carried the tag.
• Annotate October 8 to 15 and compare ChatGPT plus Direct per landing page before changing anything.

What happened

On October 9, sites that get real traffic from ChatGPT saw their ChatGPT sessions in GA4 drop sharply. Graphite CEO Ethan Smith reported drops of 22% to 57% across the sites Graphite tracks, about 45% overall, in a LinkedIn post covered by Search Engine Journal. LLM Pulse measured about 46% fewer ChatGPT sessions over October 9 to 11 on the sites they monitor, and total traffic on those sites held steady.

The timing lines up with GPT-6. OpenAI started rolling it out to paid plans on October 7 and to Free and Go users on October 8, per the same SEJ report. Profound tied the change to Intelligent UI, ChatGPT's newer format for charts, comparisons, and small tools. In Profound's data, about 98% of answers with citations had at least one tagged link on October 8. That fell to 75% on October 9 and 66% by October 14. Over the same days Intelligent UI went from 2% of the answers they collected to about 35%.

Smith says the inline links in the answer text lost the tag, while links in the Sources panel and shopping links kept it. So the same answer can send you one tagged click and one untagged click depending on where the person clicks.

What I saw when I tested it

On the evening of October 15, around 6:30 pm Eastern and hours after the fix was reported, I tested it myself. Logged out, in a regular browser with web search on, I asked ChatGPT 3 questions: project management tools for agencies, changes to GA4 attribution, and the best CRM for a 20-person SaaS startup. Then I pulled every outbound link out of all 3 answers.

• 25 links pointed to real sources like Asana, HubSpot, Pipedrive, and Google's own Analytics help pages. None of them had utm_source=chatgpt.com.
• Every one of the 25 carried rel="noreferrer", which tells the browser to drop the referrer when the page opens.
• The only links with tracking on them were 3 that looked like ads, each with campaign IDs in the URL. The one I checked was labeled Ad and carried Attio's own utm_source=chatgpt_ads.

Out of all 28 links, the only one I confirmed would tell GA4 it came from ChatGPT was the Attio ad, through Attio's own tag. My test was logged out on desktop, and the fix may not have reached logged-out sessions yet. SEO consultant Glenn Gabe said he wasn't seeing the fix on paid or free accounts either.

Why ChatGPT traffic shows as Direct in GA4

GA4 has 2 ways to know a visit came from ChatGPT. The URL carries a source tag, or the browser passes a referrer. Google added an AI Assistant channel to GA4 on May 13, 2026, and per Google's own release notes it is assigned when the referrer matches a recognized AI assistant. Take away both the tag and the referrer and GA4 has nothing to go on, so the session lands in Direct.

Huesler's early numbers point that way. Huesler, a senior director of product management at Adobe, compared October 9 to 13 with the week before on about 500 pages where ChatGPT drove at least 5% of visits. ChatGPT visits fell by about 31,000 a day and Direct rose by about 26,000 a day. He also looked at how those new Direct visitors behave. On mobile, the share who saw the consent banner (his sign of a first visit) went from 36% to 44%, closer to the 55% he sees for ChatGPT visitors.

Smith said 95% of mobile ChatGPT visits to Graphite's sites relied on the tag alone, so mobile traffic is the most exposed.

Is any of it a real drop?

Some of it might be, and the data can't settle it yet. LLM Pulse saw no rise in Direct on the sites they monitor, though they said Direct is so much bigger than ChatGPT that a modest rise would be hard to spot. They also said their aggregated data couldn't separate lost visits from relabeled ones. Huesler calls his own findings early, based on 4 days of sampled data. Graphite, LLM Pulse, Profound, and Peec AI all sell AI search tools, and SEJ notes that LLM Pulse and Profound didn't publish sample sizes.

Tomek Rudzki at Peec AI ran the same 1 million prompts before and after the rollout. ChatGPT put a plain URL into 42% of responses, up from 8% to 10% before. Of those URLs, 54% pointed at home pages. Those links had no tag or referrer either. That counts links shown, not clicks, but it means more of your untagged ChatGPT traffic may be landing on your homepage.

What to do this week

1. Add an annotation for October 8 to 15 in GA4 and in any dashboard you share, so nobody reads the dip as lost demand. Huesler now puts the window at those dates.
2. For each page ChatGPT used to send visitors to, compare ChatGPT and Direct side by side by day. If ChatGPT fell and Direct rose by a similar amount, that's your answer.
3. Check your homepage Direct traffic on its own, since that's where Rudzki's untagged links pointed most.
4. Look at conversions and pipeline for the same window. If sign-ups and demo requests held steady, the channel didn't break.
5. If you run ChatGPT ads, keep your own UTM parameters on them. Smith says those should be unaffected, and my test matched.
6. Ask new leads how they found you. A free-text "how did you hear about us" field on your demo form picks up ChatGPT whether or not analytics does.

Profound's advice is the right default here: if the ChatGPT referral number is the only thing that fell, don't change your strategy because of it.

How I measure this for clients

Referral traffic only ever showed a small part of what ChatGPT does for a brand. A lot of it happens inside the answer, when a buyer asks for options and reads a short list of names before clicking anything. A click that arrives untagged still came from that answer, and plenty of buyers never click at all.

Ante is an SEO, AEO, and GEO agency for B2B SaaS and vertical software companies, and I run every account myself. So I measure the answers directly: a fixed set of real buyer questions asked across ChatGPT, Claude, Perplexity, Gemini, and Google AI Mode every week, recording whether the brand was named, how it was described, and which pages were cited. I put that next to what new leads say on the form. A tracking change like this one doesn't touch either of them.

If you want to see where you stand in ChatGPT's answers right now, the free AEO report runs that same measurement once.

Frequently asked questions

Did ChatGPT traffic actually drop in October 2026?

Graphite estimated a roughly 45% fall in ChatGPT referral sessions in GA4 starting October 9, 2026, and most of that looks like a tracking change. Links in ChatGPT's answers stopped carrying the utm_source=chatgpt.com tag and the referrer, so GA4 counted many of those visits as Direct. LLM Pulse found total traffic on the sites they monitor held steady.

Did OpenAI confirm the missing UTM tags were a bug?

I haven't found a public statement from OpenAI. Adobe's Cedric Huesler wrote that OpenAI confirmed to him on October 14 that it was a bug, and SEJ reported that ChatGPT's release notes for October 1 to 15 don't mention link tracking.

Is the ChatGPT UTM tracking fixed?

A fix started rolling out on October 15, 2026, and Huesler reported tagged clicks back to roughly 95% to 100% of normal by 9 am Pacific on the domains he tracks. In my own logged-out test that evening, none of the 25 answer links I checked had the tag, so check your own data before assuming it's over.

Why is utm_source=chatgpt.com missing from ChatGPT links?

After the GPT-6 rollout, links in ChatGPT's answer text began opening without the tag and with rel="noreferrer". Profound found that links reached through a product or information panel still carried the tag.

Why does ChatGPT traffic show up as Direct in GA4?

GA4 needs either a UTM source tag or a referrer to know a visit came from ChatGPT. When a link opens with neither, GA4 has no source to record and files the session under Direct.

Do ChatGPT ads still carry tracking parameters?

They should. Smith said advertisers' own UTM parameters on ChatGPT ads should be unaffected, and in my test the Attio ad carried utm_source=chatgpt_ads while none of the organic links had a tag.