At its usual pace, Google's core update ends on Black Friday

Google released its November 2026 core update on 10 November at 10:06 Pacific time, according to its Search Status Dashboard. The update applies globally, in all languages, and Google says the rollout “may take up to 2 weeks”. Two weeks from 10 November is 24 November, three days before Black Friday. Google’s own past rollouts say to plan for later than that.

What Google’s own history says

The dashboard keeps a history of every ranking update and how long it ran. These are the nine core updates from 2024 to 2026, and the day this one ends if it runs as long as each of them did:

Core update	Duration	This update would end (Pacific)
December 2024	6 days 4 h	Mon 16 Nov
May 2026	11 days 21 h	Sun 22 Nov
March 2026	12 days 4 h	Sun 22 Nov
March 2025	13 days 21 h	Tue 24 Nov
June 2025 (median)	16 days 18 h	Fri 27 Nov, Black Friday
December 2025	18 days 2 h	Sat 28 Nov
August 2024	19 days 4 h	Sun 29 Nov
November 2024	23 days 13 h	Thu 3 Dec
March 2024	45 days	Fri 25 Dec

Five of the nine ran longer than two weeks. At the median length, this rollout ends on the morning of Black Friday. At the length of the November 2024 update, it ends after Cyber Monday. Only one past length, December 2024’s, leaves a full week between the end and Black Friday. The 2026 updates were shorter than the earlier ones, so “up to 2 weeks” may hold this time. The history says not to count on it.

The read: Google’s check cannot run before the peak

Google’s core updates guide gives a method for checking what an update did to your site. Confirm on the dashboard that the rollout has finished. Then wait at least a full week before you analyse the site in Search Console, and compare that week with a week before the update started.

Run that method on this update. If the rollout takes the full two weeks, it ends on 24 November, and the week you wait through runs to 1 December. That week contains Black Friday and Cyber Monday, and it is the week Google’s comparison sets against 3 to 9 November, an ordinary week. At the median length, the wait ends on 4 December. In eight of the nine cases in the table, the method gives no read before Black Friday, and the read it gives after includes the busiest shopping week of the year.

So any change in organic traffic from 10 November to early December has two causes that you cannot separate in a click or session count: the new rankings and the season. This affects more than SEO reports:

• Year-over-year organic comparisons for the holiday period compare two different sets of rankings.
• Paid search reads that use organic as a baseline, such as a brand holdout or a geo test that runs across these weeks, have a baseline that moves under them.
• Blended numbers, such as total revenue over total ad spend, move with organic revenue that the ads did not cause.

AI features do not get their own line. Google’s AI features guide says clicks and impressions from AI Overviews and AI Mode are reported inside the “Web” search type in Search Console. Google’s guide recommends that you analyse search types separately. That separates Web from Images, Video and News. It does not separate classic results from AI Overviews, so a change in “Web” can be either one.

What to do

• Annotate 10 November in Google Analytics and in any dashboard that shows organic traffic. Add the end date when the dashboard posts it, then add seven days.
• Judge the update by position, not by clicks. Clicks move with demand, and demand peaks in this window. The position of each of a fixed set of your top queries says where you rank, whatever the number of searches. Google’s guide uses position too: a drop from 2 to 4 is small, a drop from 4 to 29 is large.
• Do not change pages that perform well during the rollout. Google’s guide says so, and a change you make now cannot be told apart from the update in the data later.

Some teams will try to measure the season with a model instead. A model that reads weekly totals sees a ranking change and Black Friday in the same weeks. How many such events can it take into account before two years of data are too few?