An October tracking bug temporarily caused some ChatGPT visits to appear in analytics reports as Direct traffic instead of ChatGPT referrals. Tracking tags have since largely returned, but reports covering October 8–16 remain difficult to compare with other periods.
Before the disruption, many ChatGPT clicks could be identified because the landing URL included a ChatGPT source parameter, or because the browser sent a ChatGPT referrer. Search Engine Roundtable reported on Oct. 14 that some ChatGPT links no longer carried the utm_source=chatgpt.com parameter and were opened in a way that withholds referrer information.
Search Engine Land initially reported the missing UTM parameters on Oct. 15. In an Oct. 20 update, it said OpenAI had confirmed the issue as a bug to Adobe’s Cedric Huesler on Oct. 14, with a fix beginning to roll out the following day. Search Engine Journal cautioned that the reported referral decline did not establish an equivalent loss of visits.
OpenAI’s Oct. 7 ChatGPT release notes describe the GPT-6 and Intelligent UI rollout, including interactive answer formats, but they do not mention a link-tracking change.
Why that turns into Direct traffic
Google Analytics says (direct) / (none) represents traffic that does not have a clear referral source. Google also says a session is processed as direct when no information about the referral source is available, or when the referring source or search term has been configured to be ignored.
GA4 added an AI Assistant channel in May 2026, but Google’s own channel definition depends on recognizable signals. It sets the medium to ai-assistant and the campaign to (ai-assistant) when the referrer matches a recognized AI assistant. If a ChatGPT click arrives with neither a tracking parameter nor a referrer, GA4 has no reliable source field to classify it as ChatGPT.
The reported data points to a label problem
Search Engine Land cited an analysis from Adobe’s Cedric Huesler covering roughly 500 pages where ChatGPT had previously supplied at least 5% of visits. Comparing Oct. 9 to 13 with the week before, the analysis found ChatGPT traffic down by about 31,000 visits per day, while Direct traffic rose by about 26,000 visits per day on the same pages. Search traffic was down 7%, and Direct rose on two-thirds of the pages that lost ChatGPT referrals. Search Engine Land also cautioned that the findings did not establish that every missing referral came from the same issue.
Search Engine Journal reported several separate datasets. Graphite tracked a 22% to 57% drop for sites with meaningful AI Assistant traffic, LLM Pulse reported roughly 46% fewer ChatGPT sessions from Oct. 9 to 11 while total site traffic stayed relatively steady, and analysts tied the pattern to inline links missing the tag and referrer. Search Engine Journal also noted there was not enough data to know whether some visitation was lost.
Aloha Digital reported that, in its own sample, ChatGPT cited links carrying utm_source=chatgpt.com dropped from 99.9% from Oct. 2 to 8 to 0.75% from Oct. 11 to 15. In an Oct. 17 update, it said the tag was returning and appeared on 51.9% of cited links on Oct. 16. That suggests the issue may not have affected every ChatGPT link or account in the same way, and recovery may not have been uniform.
How to read October reports
A drop in ChatGPT referrals paired with a rise in Direct entries on the same deep pages is more consistent with lost attribution than lost interest. A drop across total visits, conversions and Direct would point to a broader traffic problem.
For the affected dates, the Direct row may contain a mix of typed visits, bookmarked visits, privacy-protected visits, app or browser handoffs, and unlabelled ChatGPT clicks. It should not be treated as proof of brand recall or proof that AI visibility is gone.
What site owners can do now
Annotate the affected dates. Mark October 8–16 as a possible attribution disruption window in GA4 and reporting dashboards, then narrow the dates based on your own data.
Review landing pages, not only source and medium. A Direct increase on the same pages that lost ChatGPT traffic is the key pattern to inspect.
Keep pre- and post-change comparisons separate. Comparing October ChatGPT referral traffic across the tracking shift can make performance look worse than it is.
Preserve UTMs through redirects. If site redirects strip query parameters, even properly tagged visits can lose attribution before GA4 records them.
Do not count crawler activity as visits. Server logs can show activity from AI crawlers or fetchers, but those requests are not the same as human sessions in analytics.
Track visibility separately from referrals. AI citation checks, landing-page trends and conversion patterns can add context when the referral label is unreliable.
Referral traffic is only one measure of AI-driven discovery. Tech Help Canada’s guide to how AI is changing search explains why impressions, branded demand, engagement and conversions also deserve attention when evaluating performance.
What remains uncertain
Independent monitoring indicates that attribution had substantially recovered by October 16. Profound reported that approximately 99% of the ChatGPT answers with cited links it collected contained at least one tagged source link by that date. Aloha Digital’s lower recovery figure measured individual cited links rather than answers containing at least one tag, so the findings are not directly comparable.
Restoring tracking does not recover source information that was missing from earlier visits. The remaining questions are how many sessions were misclassified, whether attribution has fully recovered across all link formats, and whether any genuine changes in visitor behavior occurred alongside the bug.
Get new small business insights by email
Practical ideas and useful articles to help you make better business decisions.
Tech Help Canada Staff researches, writes, and reviews practical content for business owners and professionals. Our coverage spans business, marketing, SEO, technology, and the tools and systems people use to grow and operate online. We focus on clear, useful information backed by research, hands-on experience, and editorial review. Learn more about our team and editorial standards. Need help with something? Contact Us
