Traffic drop after a Google update: what to do, and in what order

Step 1: rule out a technical fault

A sudden drop has two families of causes: a problem on your site (Google can no longer read your pages, or no longer wants to index them) or a change in Google’s assessment (an update to its ranking). They are treated in opposite ways: a fault is fixed in a day, an update calls for in-depth work. Always start by ruling out a fault.

1. 1
   Search Console, Pages report: did the number of indexed pages fall all at once? A sudden rise in “Excluded by ‘noindex’ tag” or “Blocked by robots.txt” points to a deployment error.
2. 2
   URL Inspection on your three most important pages: are they indexed, and does the page rendered by Google really show your content?
3. 3
   Robots.txt and tags: has a plugin, theme or hosting setting added a noindex or a block?
4. 4
   Redirects and canonicals: has a redesign or a change of address broken URLs or pointed canonicals to the home page?
5. 5
   Availability: repeated server errors (5xx) in the Crawl stats report slow down Google’s crawling.
6. 6
   Manual actions: Search Console reports any penalty applied by hand by Google in a dedicated report.

Step 2: date the drop and compare it with the updates

In Search Console, Performance report, show the last 16 months with impressions and average position ticked. Find the exact day the curve breaks. Then compare that date with the Google Search Status Dashboard, which lists core updates and spam updates with their start and end dates.

If the drop starts while an update is rolling out and hits the whole site at once, you have your explanation. Google states that its ranking systems work at page level, but that they also use site-wide signals and classifications: a weak part can weigh on the whole.

This is exactly what we saw in summer 2026. On our own sites, the median position of entire sites went from about 24 to 70 to 80, in line with update episodes, and our investigation of 22 September 2026 found no technical fault.

What Google says about core updates

A core update is not a penalty. Google reassesses all content to better surface what it considers helpful and reliable. Since March 2024, the helpful content system has been part of its core ranking systems: there is no longer a separate “helpful content update”, the whole ranking takes it into account.

Google asks for “people-first” content, created mainly for people and not to manipulate rankings. Its spam policies target scaled content abuse in particular: generating many pages with the main purpose of manipulating rankings, for example with generative AI tools, without adding value for users.

On timing, Google is honest: after improvements, it can take some time to see the effect, and some changes only take effect over several months. There is no point changing everything every week: improve, then measure over time.

What we went through on our own sites

In summer 2026, 10 of our sites lost 90% to 99% of their impressions. Every drop followed a mass release of generated pages or assembly-line publishing, with template pages 56% to 73% similar to each other. The sites that held up carried a data point of their own on every page: a calculation, a measurement, an official source.

In mid-October, of the 15 sites with more than 70 days of data, 7 were still below half of their best 28-day period and 5 below 10%. A drop like this is not made up in a few days.

What came before a recovery: 3 sites had merged their page variants with 301 redirects, keeping their content, on 22 and 23 September, just before the spam update of 24 September. All three came back. A pet site with a shop: 1,955 impressions in the 28 days before, 147,302 after, position 56.2 then 11.8. An automation templates site: 738 impressions in the 28 days before, 9,546 after, position 22.6 then 8.5. A sales page templates site: 535 impressions in the 28 days before, 4,973 after, position 62.1 then 13.0.

We cannot prove the cause, and over the same period, a wellness guide site (French and English), a plumber comparison site and a sports directory, also pruned, stayed at rock bottom. Pruning alone does not bring traffic back: on a van comparison site, removing 8,194 pages with no impressions lost nothing (104 impressions a day before, 98 after), but nothing came back either.

Step 3: the action plan, page by page

Once the fault is ruled out and the update identified, we follow this order, from the most profitable to the longest:

1. 1
   List the pages that lost the most: Search Console, Performance, Pages tab, comparing the 28 days before and after the drop.
2. 2
   Classify each page: keep (it answers a real search and has value), merge (several pages for the same intent), remove (no impressions, no content of its own).
3. 3
   Merge duplicates with 301s: move the best content to the page you keep before redirecting, never mass redirects to the home page.
4. 4
   Remove worthless pages with a 410: empty generated pages, variants with no content of their own, archives with no visits.
5. 5
   Rewrite the pages you keep: an answer from the first sentence, a data point or source specific to the page, an identified author or owner, a real update date.
6. 6
   Submit up-to-date sitemaps: only the pages you keep, with the last modification date of the content.
7. 7
   Measure every four weeks: impressions and position of the rewritten pages, without changing everything between two measurements.

What not to do

Panic reflexes cost more than the drop itself. The most common ones:

• Deleting pages at random or redirecting everything to the home page: a redirect must lead to the page that answers the same intent, otherwise neither the visitor nor Google finds what they were looking for.
• Changing the title of every page at the same time: you will no longer know what worked.
• Buying links to “climb back”: it breaks Google’s spam policies and can make things worse.
• Mass publishing generated pages to recover volume: this is precisely what the rules on scaled content abuse target.
• Drawing conclusions after a week: Google says some improvements only take effect over several months.

Meanwhile, stop depending on a single search engine

A Google drop is a reminder that a profitable site should not depend on a single source. On our selling sites, from 1 October 2026 to 19 October 2026, 52% of visits from a search engine came through Bing and the engines that use its index, and 13% of measured visits came from an AI assistant. On two guide sites that Google hardly indexes any more, every search engine visit came from this family.

Setting up Bing Webmaster Tools, opening free Google Shopping listings if you sell products, and making the site readable for AI assistants are short projects that cushion the shocks.