ChatGPT Traffic in GA4: What October’s Tracking Bug Means for Small Businesses

If ChatGPT traffic in GA4 fell off a cliff around October 8 or 9, 2026, and Direct traffic rose on the same pages, you probably did not lose those visitors. For about a week, links inside ChatGPT stopped carrying the tag that tells GA4 where a visit came from. Adobe’s Cedric Huesler said OpenAI confirmed it was a known bug, and the fix began rolling out early on October 15. By October 16, tagged links were back to normal levels in the tracking data Profound published.

So the short answer is to annotate October 8 to 15 in your reports, leave your marketing plan alone, and use the scare to set up AI referral tracking that does not depend on a single URL tag.

What happened to ChatGPT traffic in GA4 between October 8 and 15

ChatGPT normally adds utm_source=chatgpt.com to the links it shows in answers. Analytics tools read that tag and file the visit under ChatGPT. On October 8, that tag started disappearing.

The timeline, from the people who tracked it:

• October 8, around 2pm PT: Huesler wrote on LinkedIn that ChatGPT, following the GPT-6 release, stopped adding the tag. Attributable ChatGPT visits in Adobe’s data dropped to about a third of normal within hours.
• October 9 to 14: Profound, which collects ChatGPT answers at scale, found the share of answers with at least one tagged source link fell from about 99% to about 75% on October 9, then to roughly 67% by October 14. The drop lined up with ChatGPT’s new answer format, which Profound calls Intelligent UI.
• October 14, evening UTC: Profound saw the new format jump to more than 90% of answers, and the tagged share fell to around 5% to 12% for about a day.
• October 14: Search Engine Roundtable reported that OpenAI appeared to have dropped the tags. Huesler said OpenAI told him that day it was a known bug with a fix coming that week.
• October 15 to 16: The fix started rolling out around 2am PT on October 15. Huesler reported tagged clicks back to roughly 95% to 100% of the prior week by 9am PT. Profound’s data recovered more slowly and reached about 99% early on October 16 (UTC).

OpenAI has not posted anything public about it. The confirmation is secondhand, through Huesler, so treat the cause as very likely rather than officially documented.

Why the visits showed up as Direct instead of disappearing

GA4 decides where a visit came from mainly two ways. It reads campaign tags on the URL (the utm_ parameters), and it reads the referrer, the header a browser sends to say which page the visitor clicked from. If both are missing, GA4 has nothing to go on and files the session as Direct, the same bucket used for someone who typed your address into the browser.

During the bug, ChatGPT dropped the tag, and according to Huesler it also opened links in a way that strips the referrer. He noted on October 15 that the referrer is still missing on most ChatGPT clicks even after the fix. That leaves the tag doing almost all of the work, which is why its absence hit reports so hard.

Huesler compared about 500 pages where ChatGPT drove at least 5% of visits. Between October 9 to 13 and the week before, ChatGPT visits fell by about 31,000 a day and Direct rose by about 26,000 a day on those same pages, while search traffic moved about the same as everywhere else. On most small business sites the numbers will be tiny by comparison, but the pattern should look the same.

How to check your own GA4 for the ChatGPT tracking gap

You can confirm this in about ten minutes, even if you are not an analytics person.

1. Open GA4 and go to Reports, then Acquisition, then Traffic acquisition.
2. Change the primary dimension to Session source / medium.
3. Set a comparison: October 1 to 7 against October 9 to 14.
4. Search the table for chatgpt. Note the sessions in each period.
5. Now search for (direct) / (none) and note the same two numbers.
6. Add Landing page as a secondary dimension and look at the pages that got ChatGPT visits in the first week. If those exact pages lost ChatGPT sessions and gained Direct sessions, you are looking at the bug.
7. Check your key events (calls, form submissions, bookings) across both periods. If leads held steady while the ChatGPT line fell, nothing real changed.

Small sites need some caution here. If ChatGPT sent you eight visits in a normal week, a drop to three is too small to read anything into by itself. In that case, look at whether Direct visits to your service pages, the pages people rarely type in by hand, went up at the same time. A rise in Direct traffic to a deep page like a pricing or service-area page is a decent sign that a tracked source lost its label.

If the ChatGPT number fell and your total visits and leads also fell, the bug may not be the whole story. Check Search Console for a search drop over the same dates before you assume it was all tagging.

Annotate October 8 to 15 and leave the strategy alone

The damage from this bug is mostly in how people read reports later. In three months, someone will pull an October report, see ChatGPT referrals dip, and suggest cutting the content work that was earning those citations. A dated note on the chart heads that off.

GA4 has an annotation feature built in. On any line chart in Reports, right-click a date and choose Add annotation. Give it a short title such as “ChatGPT UTM bug,” set the date range to October 8 through October 15, and add a one-line description. You need Analyst access or higher on the property.

Do the same in anything else you use to report on marketing: a Looker Studio dashboard, a monthly agency report, or a spreadsheet. Then tell whoever reads those reports. If you pay a marketing vendor, ask them directly whether their October numbers account for this.

Watch for the opposite mistake too. Profound warned that the jump back up as tags returned can look like growth when it is only restored labeling. Compare November to September, not to October.

Set up ChatGPT traffic tracking in GA4 that survives the next glitch

This will not be the last time an AI tool changes how it links out. A few settings make your reports much harder to break.

Know what the built-in AI Assistant channel does

Google added an AI Assistant channel to GA4’s default channel group on May 13, 2026. According to Google’s documentation, it covers visits from sources like ChatGPT, Gemini, DeepSeek, Copilot, and Grok. GA4 sets the medium to ai-assistant when the referrer matches Google’s list of AI assistants.

For a small business, the channel has two gaps. First, it depends on the referrer, which ChatGPT often does not send, so the AI Assistant row can undercount ChatGPT even when the tag is working. Second, GA4 Optimizer, which tracked the rollout, reports that older sessions were not moved into the new channel, so it cannot show you a trend from before May. You also cannot edit the default channel group.

Add a custom channel group based on source

A custom channel group fixes both problems. Google’s help page says custom channel groups can be applied to reports retroactively, and you can match on source, which picks up the utm_source=chatgpt.com tag whether or not a referrer came through.

1. In GA4, go to Admin, then Data display, then Channel groups.
2. Click Create new channel group. GA4 starts you with a copy of the default group.
3. Name it something like “Channels with AI.”
4. Click Add new channel, name it “AI assistants,” and set the condition to Source matches regex with a pattern such as chatgpt\.com|chat\.openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com.
5. Add a second condition with OR: Default channel group exactly matches AI Assistant. That catches sources Google recognizes that your pattern missed.
6. Click Reorder and drag “AI assistants” above Referral. GA4 puts each visit in the first channel it matches, so a rule sitting below Referral never fires for referred visits.
7. Save, then open Traffic acquisition and switch the report to your new channel group.

Standard GA4 properties allow two custom channel groups, so do not burn both on experiments. Update the pattern when a new assistant starts showing up in your source list.

Google AI Overviews and AI Mode are a separate measurement problem

GA4’s AI Assistant channel leaves out Google’s own AI results. Google’s channel documentation says it excludes AI Overviews and AI Mode, and that clicks from those features count as Organic Search. In GA4, a visit from an AI Overview looks the same as a visit from a regular blue link.

Search Console does not fill that gap yet. Its generative AI performance report, which Google says reached all websites worldwide on August 31, 2026, shows impressions from AI Overviews and AI Mode by page, country, device, and date. It does not show clicks.

On October 15, Google announced more inline links in AI Mode and AI Overviews, links placed higher in responses, and a test of a section at the top of some answers that highlights the one or two official pages needed to finish a task. When Search Engine Roundtable asked about click data for these features, Google said it would “introduce additional metrics over time,” with no date.

For a local service business, the practical result is that you can see whether Google shows your pages in AI answers, but you cannot see how many people clicked from those answers. We covered what to fix on the site itself when AI Overviews show up in local results in our post on Google testing AI Overviews in local knowledge panels. For measurement, the best move today is to watch generative AI impressions on your main service pages alongside your call and form numbers, and to stop expecting a clean click count.

Track AI referral leads in your forms and phone calls

Sessions were always a shaky way to judge AI traffic, and this bug made that obvious. A plumber, dentist, or accountant cares about booked jobs, not visit counts. A few low-tech habits give you a record that no tagging bug can erase.

Ask people how they found you. Add a short “How did you hear about us?” field to your contact and booking forms and include “ChatGPT or another AI assistant” as a choice. Have whoever answers the phone ask the same question and log it. Self-reported answers are imperfect, but they catch people who read about you in ChatGPT and then typed your name into Google or your address into the browser, a path no analytics tool can see.

Save the source with the lead. Most form tools can store URL parameters in a hidden field. If your form captures utm_source when it is present, every lead from a tagged ChatGPT link arrives with “chatgpt.com” attached, right in your inbox or CRM. Ask your web developer to set it up if your form plugin does not do it out of the box.

Track calls as events. For many service businesses the phone is the main conversion. Make sure taps on your phone number are tracked as a key event in GA4, or use a call tracking tool that records the source of each call. A week where ChatGPT sessions drop and calls hold steady tells you the visitors were still there.

Look at the pages AI tools send people to. In your custom AI channel, add Landing page as a dimension and note which pages AI tools send people to. If they are pages with specific answers (what a service costs, what areas you cover, how a process works), keep those pages accurate and current, since they are the ones doing the work.

What to do this week

• Compare ChatGPT and Direct sessions in GA4 for October 1 to 7 against October 9 to 14, with Landing page as a secondary dimension.
• Add a GA4 annotation for October 8 to 15 labeled as the ChatGPT tracking bug, and add the same note to any dashboard or monthly report.
• Create a custom channel group with an AI assistants channel that matches on source, placed above Referral.
• Open the generative AI report in Search Console and note which service pages get AI Overview or AI Mode impressions.
• Add a “How did you hear about us?” question with an AI option to your forms and phone intake.
• Confirm phone taps and form submissions are tracked as key events.

If you would rather have someone set up the tracking and read the reports with you, that is part of our SEO services for small businesses. Intuico Digital works remotely with businesses nationwide, and the first step is a look at what your current analytics can and cannot see.

Sources

• Cedric Huesler (Adobe), LinkedIn post on the ChatGPT tagging bug and fix, October 15, 2026
• Profound, ChatGPT’s missing UTM tags: what happened from October 8 to October 16
• Search Engine Roundtable, OpenAI Drops UTM Parameters From ChatGPT Results, October 14, 2026 (updated October 15)
• Search Engine Land, ChatGPT appears to drop UTM tracking, October 15, 2026
• Google Analytics Help, Default channel group and Custom channel groups
• Google Analytics Help, What’s new in Google Analytics (May 13, 2026 AI Assistant entry)
• GA4 Optimizer, GA4 AI Assistant Channel
• Google Analytics Help, About annotations
• Search Console Help, Generative AI performance report
• Google, More ways to connect with the web from Search, October 15, 2026
• Search Engine Roundtable, Google Updates Links In AI Mode & AI Overviews Again, Won’t Show Click Data, October 15, 2026