Read a Core Update's Impact After It Ends

Observation: Both Times, the First Three Days Looked 35% Down

According to the Search Status Dashboard, the November 2026 core update began on November 10 at 10:06 Pacific time, applies globally and to all languages, and may take up to two weeks to complete. Right now is the stretch when rank trackers and social feeds fill up with "down" and "up."

So we went back over 2026's two previous core updates in Search Console for another site we run (a developer-focused publication; "Site A" below), in the same format. Web-search clicks are shown under three comparisons. Dates are as Search Console reports them (Pacific time), using finalized data only.

"Same weekdays" compares the first three days with the same three days one week earlier; "Google's method" compares the seven days before the start with the seven days beginning one week after completion (clicks; the last column is impressions).

Looking only at the first three days, we'd have reported a drop of over 30% both times. In fact March was roughly flat against the same weekdays, May was up, and by Google's method after completion clicks had risen both times. The same "−35%" was produced, twice, by reasons close to coincidence.

What Made the Apparent Drop: Weekdays and Pacific-Time Dates

Site A is read on weekdays. Comparing each weekday's average clicks from March 6 to June 16, 2026 with the average across all seven days gives this skew:

For a site read in Japan, the lowest days are Saturday and Friday, while Sunday barely dips. That's because Search Console's performance report splits days in Pacific time in every view except the 24-hour view. In Japan the date changes at 5 p.m. (4 p.m. during US daylight saving time), so Search Console's "Friday" is Friday evening to Saturday evening in Japan, and "Sunday" is Sunday evening to Monday evening. The weekend dip shows up half a day early, on Friday and Saturday.

The March update started on a Friday, so the first three days were Fri–Sun against Tue–Thu before. May started on a Thursday: Thu–Sat against Mon–Wed. Both times we were comparing across the line from strong days to weak days. Most of the −35% was the weekday gap, not the update.

The Yardstick: Measure Your Site's Normal Swing First

Even with weekdays aligned, week-to-week swings remain. How much movement it takes to say "the update moved it" differs by site. So from the same Site A period we removed anything touching an update (from the day before the start to seven days after completion), and compared each 7-day total with the previous seven days, sliding one day at a time — 28 pairs of windows.

• Median absolute change in clicks: 13%; top tenth: 57% or more
• Impressions: median 12%; top tenth: 47% or more
• The largest, 89%, was a window comparing Japan's long Golden Week holiday with the weeks around it

On a site with double-digit daily clicks, a week moves a bit over 10% with no update at all, and more than half once holidays are involved. Reading the table above with this yardstick, May's +8% by Google's method is below the median, and March's +19% sits within the normal range too. For Site A, the right conclusion both times was "we can't say the core update moved it."

The sliding windows overlap in dates, so they're not independent. This doesn't replace a statistical test, but it's enough to draw a line in advance: "inside this range, we don't decide." It covers three and a half months, so seasonality isn't included.

Designing the Judgment: When, Against What, and What Not to Do

Google's core update guidance says to confirm the rollout has finished, wait at least a full week after completion, and compare with a week before the start. Then look at top pages and queries: for a small drop like position 2 to 4, no drastic action (avoid changing content that's performing well); for a large drop like 4 to 29, do a deeper assessment. It also recommends analyzing search types (web, images, video, news) separately.

We add two things from the measurements above and run it like this:

The first addition is to align comparisons to Sunday–Saturday weeks. Search Console's weekly view is also split Sunday to Saturday, so using it as-is cancels the weekday skew. The second is to draw the "did it move" line from the previous section's yardstick before judging. A line drawn after seeing the result bends toward the conclusion you wanted.

Not rewriting during the rollout isn't about caution — it's about attribution. Add your own change mid-update and, if the numbers move after completion, you can no longer tell which one moved them.

Where This Design Breaks Down

The baseline week contains another update or a holiday

The March 2026 core update started right after the March spam update (March 24–25, Pacific time). The week before the start contains that spam update in full. With this comparison we can't separate whether March's "impressions −32%" came from the core update or from a different baseline after the spam update. In November, the week before the start contains a Japanese national holiday (November 3), and depending on when it completes, the comparison week may contain another (November 23). Match the number of holidays, or if you can't, note that the week includes a holiday and weaken the judgment accordingly.

The site is small and normal swings are too large

With single-digit daily clicks, normal weekly swings are even larger than Site A's, and one-week comparisons often say nothing. Switch to four weeks against four weeks, compare with the same period last year, or look at position by top query instead of the site total. If it's still inside the swing, the conclusion is to record that you held off.

Impressions and clicks move in opposite directions

In March, clicks rose 19% while impressions fell 32%. Impressions swing heavily when low-ranking queries slip in and out. Reading site-wide impressions as "rated up" or "rated down" goes wrong. Narrowing to top pages and queries and checking whether position changes look like Google's "small drop" or "large drop" is closer to a decision.

The large drop wasn't caused by the update

A forgotten noindex, a misdirected canonical, or a robots.txt change can land during an update window too. When timing overlaps, it's tempting to read it as "the update hit us." When you see a large drop, check whether the affected pages are communicated correctly to search before reviewing content. Reverse the order and you'll rewrite content while the technical fault stays in place.

What We're Doing for the November Update

On both this site (Direbase) and Site A, we're logging our own changes with dates during the rollout and waiting for the completion notice on the dashboard. The baseline is the Sunday–Saturday week before the start (noting it includes a holiday); the comparison is the Sunday–Saturday week after one full week past completion. The line for "moved" is set before judging, using this article's yardstick (57%, the top tenth, for Site A).

How we approached earlier updates is in three things Google stated about the May core update and decomposing growth during the July volatility. For verifying a change in Search Console see verifying SEO changes with Search Console, and for bots mixing into GA4 numbers see keeping GA4 bots out of decisions.

Related Tools

• SEO Score Check — rule out noindex or canonical errors on pages that dropped
• Site-Wide Diagnosis — check technical issues across many pages at once
• Competitor SEO Analysis — compare foundations with the sites that swapped places with you

Sources (checked November 11, 2026)

• Google Search Status Dashboard — November 2026 core update
• Google Search Central — Google Search's core updates and your website
• Search Console Help — Performance report (Search results): Dimensions and data groupings

Start and end times of past updates were checked in the same dashboard's history (March 2026 spam and core updates, May 2026 core update). Site A figures are finalized Search Console web-search data (March 6 – June 16, 2026, Pacific time).