Changelog
Dated changes to link attribution, newest first. Shipped improvements come with a thank-you. Reported entries come from third parties and are not yet verified by us.
ChatGPT restored the utm_source=chatgpt.com tag
On October 14, OpenAI confirmed the missing tag was a known bug and that a fix was coming that week. The fix started rolling out around 2am PT on October 15, in stages, and by 9am PT tagged ChatGPT clicks were back to about 95–100% of the same hour a week earlier, across thousands of domains in AEM Operational Telemetry. The referrer is still missing on most ChatGPT clicks, so the tag is the signal sites can rely on.
Thank you to the OpenAI team for confirming within hours and shipping the fix within a day.
Verified · Source
ChatGPT links lost the UTM tag after the GPT-6 rollout
Around 2pm PT on October 8, ChatGPT stopped adding utm_source=chatgpt.com to its links. With no tag and no referrer, ChatGPT visits that sites could attribute fell to about a third of normal within hours and showed up as Direct instead. The visits were never lost, only hidden. The drop was reported publicly on October 13 by Cedric Huesler. We share it as a common lesson: link handling is easy to change by accident in a fast UI cycle, and a regression test per link location catches it. If you report on this period, annotate October 8–15 as one event, not a trend.
Verified · Source
Gemini stopped tagging links with utm_source=gemini
On September 29 the utm_source=gemini tag stopped appearing on Gemini links, after 11 days. Visits that arrive without a referrer, mostly on mobile, are counted as Direct again. Our October 20 web test found no tag on any Gemini link.
Verified · Source
Gemini tagged its links with utm_source=gemini for 11 days
From September 18 to 28, Gemini links carried utm_source=gemini (geminiforbusiness for Workspace; earlier partial sightings from September 10). It gave sites a rare measurement: across 1,000+ domains in AEM Operational Telemetry (real-user visits, bots excluded), real Gemini clicks were 2.1× what the referrer alone shows, about 2.5× once untagged links are accounted for. Most of the gap was mobile visits with no referrer, normally counted as Direct.
Thank you to the Gemini team. Eleven days of tagging showed the size of the gap more clearly than any estimate.
Verified · Source