AI NewsGo-to-marketAnnouncement

Google Search Console undercounted AI-answer impressions for three days

Google confirmed a logging bug. It cut impressions in the Search Console Generative AI performance report for 9 to 11 November.

AI News

Editorial2 min read

LinkedInX

Why it mattersA team tracking its AI-answer visibility in Google will see a dip in those three days that is not real. Do not read it as lost reach.

A publisher watching its AI-answer visibility in Google will see a three-day drop in impressions. The dates are 9, 10 and 11 November. The drop was a logging bug. It was not a real change. Google added a note to its data-anomalies page. The note names the Generative AI performance report. It says a logging error caused a decrease in impressions for those three days.

The report in question is Search Console's Generative AI performance report. It records how often a site shows up inside Google's AI Overviews and AI Mode answers in Search. Google says the fault sits in the logging path. Clicks, position and the live search results were not affected. Only the recorded impression count on this one report was.

How the bug was found

Barry Schwartz at Search Engine Roundtable reported the note on 12 November 2026. Schwartz says Google added it to the data-anomalies page and also to an annotation visible inside the report itself. Google has not published a wider statement.

The dates line up with a second event on the same report. Google began rolling out its November 2026 core update on 10 November. A site owner looked at the Generative AI report on 11 November. The dip in impressions was easy to read as the first sign of a ranking drop from the core update. That reading is wrong on these dates. The dip is a Google counting problem.

Three jobs for the week

A team that reports AI visibility to a client or to its own leadership has three jobs this week.

First, mark 9, 10 and 11 November on every chart as blocked rather than slow. A footnote on the chart is enough. The dip is Google's, not the site's.

Second, compare weeks, not days, on the Generative AI report for the rest of the month. A weekly view smooths the three missing days and still catches a real change. A daily view is noisy for a different reason now.

Third, do not use the same three days to test a change. A content refresh or a schema edit released on 10 November cannot be measured on the AI report yet. Rerun the test after a clean week.

The tool for a wider read of AI visibility is a measured prompt set. Run it against ChatGPT, Perplexity, Gemini, Copilot and Google AI Mode. A monitor like that catches a dip in brand mention rate. Google's own report cannot show that dip on its best day. The three-day Search Console bug is a reminder. One vendor's dashboard is one vendor's reading of the world.

Source

Google Search Console data anomalies, Google Search Central.

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.

Share
LinkedInX
Start a project