If your ChatGPT referral traffic fell off a cliff starting October 9, 2026, your traffic probably did not go anywhere. ChatGPT stopped tagging its links, so those visits landed in your analytics as direct traffic instead. OpenAI confirmed the issue on October 14 and a fix started rolling out October 15. Here is what to do with the week of data that is now wrong.
What Broke
Since June, ChatGPT had been adding a utm_source=chatgpt.com parameter to the links in its answers. That parameter is how analytics tools knew a visit came from ChatGPT.
After GPT-6 rolled out in early October, it stopped appearing. According to Cedric Huesler of Adobe, ChatGPT also began opening links in a new tab using a browser setting that withholds referrer information. With no tag and no referrer, your analytics has nothing to work with, so the visit gets filed as direct.
Ethan Smith of Graphite reported that inline links in the answer text lost the tag while Sources panel and shopping links kept it. That explains why the drop looks different from site to site.
The Timeline
- October 7. GPT-6 rolled out to paid plans. No drop observed that day.
- October 8. GPT-6 reached Free and Go users.
- October 9. The decline in ChatGPT-attributed sessions begins.
- October 14. OpenAI reportedly confirmed the issue to Huesler.
- October 15. A fix began rolling out. By 9 a.m. Pacific, tagged ChatGPT clicks were back to roughly 95 to 100 percent of the same hour a week earlier, per Huesler's tracking across thousands of domains.
Keep in mind that Glenn Gabe reported not yet seeing the fix on paid or free accounts, so the rollout was not instant everywhere.
How Big The Gap Was
Several analysts measured this independently, and the numbers are not small.
Ethan Smith at Graphite found GA4 ChatGPT-attributed sessions fell between 22 and 57 percent across tracked sites, estimating roughly a 45 percent drop from baseline. LLM Pulse measured ChatGPT sessions about 46 percent lower on October 9 through 11 than the same days in the prior two weeks, while total site traffic stayed roughly flat.
Huesler's analysis covered about 500 pages where ChatGPT drove at least 5 percent of visits. Comparing October 9 to 13 against the prior week, ChatGPT visits fell about 31,000 per day while direct visits rose about 26,000 per day. Search traffic fell 7 percent, in line with trends elsewhere.
That last pairing is the whole story. The traffic did not disappear. It moved to a different column.
Read These Numbers Carefully
Search Engine Journal pointed out that several sources here sell SEO or AI search services, and that the LLM Pulse and Profound posts did not publish sample sizes. Huesler based his on four days of sampled data. LLM Pulse did not observe a rise in direct traffic at all, though direct is a much larger bucket, so a modest increase would be hard to spot.
Aggregated data cannot cleanly separate visits that were truly lost from visits that are simply no longer labeled. Treat the direction as solid and the exact percentages as estimates.
This is the same caution we applied when Google started blocking rank trackers. When the measurement layer breaks, the first instinct is to believe the chart. The chart is the thing that broke.
What To Do About Your Reports
- Annotate October 8 through 15 in your analytics. Huesler is using that window. Anyone reading a traffic report for this period later needs to see the note.
- Compare ChatGPT and direct side by side, page by page. If direct rose as ChatGPT fell on the same pages, you have your answer.
- Count the UTM tag, not the referrer. Huesler's point is that most ChatGPT clicks arrive without a referrer anyway, so the tag was doing the work all along.
- Do not change strategy off this week. If you were about to cut effort on content ChatGPT was citing, hold off. That is exactly the wrong read.
- Check your GA4 setup while you are in there. If you are not confident what is feeding your reports, the Google Analytics 4 install is worth reviewing.
The restored tags help future reports. They do not recover the attribution already lost, so that week is a hole you explain rather than fill.
Smith noted that UTM parameters you add to ChatGPT Ads clicks are not affected, since those are your tags, not OpenAI's.
If this is making you nervous about relying on one platform's tagging, that is the right instinct, and it is a tracking approach worth building properly.
Question to Answer:
Did your direct traffic rise by roughly the same amount your ChatGPT referrals fell between October 9 and October 15?
In Summary
ChatGPT stopped adding its UTM tag to links after GPT-6 rolled out, so visits that used to show up as ChatGPT referrals showed up as direct traffic instead starting October 9. OpenAI confirmed it on October 14 and a fix began rolling out October 15, with tagged clicks back to roughly normal levels the same day for most sites.
Your job now is cleanup, not strategy. Annotate the window, compare ChatGPT against direct on the same pages before you conclude anything, and do not make a content decision based on a week where the measurement was broken. The traffic was there. The label was not.
0 comments