iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Reports published on October 14, 2026 indicate that links in some ChatGPT answers no longer include the utm_source=chatgpt.com parameter and may no longer send an HTTP referrer. That can make a ChatGPT click appear as “Direct” in analytics even when the visitor came from ChatGPT. It does not, by itself, show that ChatGPT stopped sending visitors. OpenAI has not confirmed whether the behavior is intentional, a bug, or permanent.
What appears to have changed
Search Engine Roundtable and SEO Daily described a change noticed after GPT-6 began rolling out: links in the main portions of some ChatGPT responses appeared without the UTM tag that previously helped publishers identify ChatGPT-originated sessions. Cedric Huesler of Adobe also reported that GPT-6 opened links with window.open(url, "_blank", "noopener,noreferrer"), removing both the tracking query string and the referrer information that analytics platforms normally use.
When a URL arrives without a campaign tag or recognizable referrer, many analytics systems classify the session as direct traffic. The resulting report can therefore show fewer ChatGPT referrals and more direct visits without any equivalent fall in total visits.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesWhy GPT-6 timing matters—but does not prove causation
OpenAI announced GPT-6 on October 7, 2026. The global ChatGPT rollout started with Plus, Pro, Business, and Enterprise tiers, followed by Free and Go expansion on October 8. Paid users receive GPT-6 Sol, while Free and Go users receive GPT-6 Luna. OpenAI said the models powering Work and Codex were not changing in that release; its safety update described GPT-6 Sol and GPT-6 Luna as replacements for GPT-5.6 Sol and GPT-5.6 Luna inside ChatGPT, while earlier versions remained in Codex and ChatGPT Work.
#1 Best Overall
The apparent attribution change began around the same period, but timing alone cannot establish that GPT-6 intentionally removed tracking. OpenAI has made no public statement confirming a UTM or referrer policy change, and independent tests have not produced identical results.
What the available observations show
Huesler said he compared approximately 500 pages that had received at least 5% of their visits from ChatGPT. In that sample, he reported about 31,000 fewer ChatGPT visits per day, 26,000 more direct visits per day on the same pages, and a 7% decline in search traffic. He also said direct traffic increased on two out of three pages that lost ChatGPT traffic. These are reported observations from a defined sample, not an independently verified, industry-wide measurement.
Rank #2
Testing has been inconsistent. SEO Daily reported that Glenn Gabe still saw UTM parameters in ChatGPT’s More section but not on the main sources. That difference could indicate a partial rollout, a link-generation bug, or separate handling for different result surfaces.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteHow attribution can differ by link type
| Comparison | Observed or expected behavior | What it means for analytics |
|---|---|---|
| UTM parameter | utm_source=chatgpt.com may be absent from some links |
Campaign-based ChatGPT reporting can undercount visits |
| HTTP referrer | Some links may open with no referrer because of noopener,noreferrer |
Sessions can be classified as direct instead of referral |
| Main result links | Reports indicate missing tags in some primary answer links | High-volume links may be affected first |
| More-section links | At least one tester still observed parameters there | Attribution may vary within the same answer |
| Analytics totals | Reported referral losses were accompanied by direct-traffic gains | A channel shift is not proof of fewer total visitors |
| OpenAI status | No confirmation of intent, scope, or permanence | Treat the behavior as unconfirmed until an official statement appears |
How to check whether your traffic really dropped
- Choose matching dates. Compare the period after the observed change with the same number of days immediately before it, and annotate October 7–14, 2026 as the GPT-6 rollout window.
- Measure total sessions first. Check landing-page sessions, users, conversions, and revenue across all channels before looking only at the ChatGPT referral line.
- Combine channels. Add known ChatGPT referrals to direct traffic for the same landing pages. A referral decline paired with a similar direct increase is consistent with reclassification.
- Segment by landing page. Huesler’s observation concerned pages with substantial ChatGPT share. Compare the exact URLs that previously received ChatGPT traffic rather than relying on a site-wide average.
- Inspect raw request data. If your server or edge logs retain query strings and referrer headers, check whether new sessions arrive without both values. Analytics interfaces may apply their own channel rules.
- Test different ChatGPT surfaces. Open a link from a main answer and from the More section, on both desktop and mobile where possible. Record the full URL, query string, and referrer visible to your measurement tools.
- Watch conversions. If direct traffic rises while conversions from the affected landing pages remain proportionate, attribution loss is more likely than a sudden audience collapse.
What publishers should do now
Preserve a pre-change baseline
Export daily ChatGPT referral, direct, search, session, and conversion figures before analytics platforms overwrite detailed histories. Keep the date range, landing-page list, device split, and channel definitions with the export.
Do not force a new channel rule prematurely
Manually labeling all direct traffic as ChatGPT would overstate the source because direct sessions also include bookmarks, typed URLs, privacy tools, and other unidentifiable visits. Use a separate “possible ChatGPT” analysis only when you can explain the method and its limits.
Document link-surface differences
Record whether a test came from a primary answer, a More section, or another ChatGPT interface. The conflicting observations indicate that one test cannot represent every link ChatGPT generates.
Communicate uncertainty internally
Report the finding as a measurement change under investigation, not as proof that GPT-6 reduced demand. Include total traffic and business outcomes alongside channel counts.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
What remains unknown
- Whether the missing UTM tags affect every ChatGPT tier, device, language, or interface.
- Whether the behavior is an intentional privacy or product decision, a temporary bug, or an experiment.
- Whether OpenAI will restore tags, provide another attribution method, or leave links untagged.
- How many publishers are affected outside the reported sample of approximately 500 pages.
ChatGPT reaches more than 1.2 billion people each week, according to OpenAI’s 2026 announcement, so even a partial attribution change can materially alter channel reports. Until OpenAI clarifies the implementation, the safest interpretation is that some referral metadata may be missing—not that ChatGPT traffic has disappeared.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
