Google Spam Update: What It Is and How to Check Whether It Hit You

Since the August 2025 update I have covered every spam update here with its own reports. That is five updates by now. On 24.09.2026, for the fifth, I opened all twelve spam update entries in Google’s release history one by one and put each announcement next to the measured runtime. It struck me that every one of these reports has to answer the same basic questions about spam updates from scratch before it gets to the actual news.

What is a spam update in the first place? How does it differ from a core update and from a manual action? And how do you check whether it hit you? These answers do not change with every update. That is why they are now here, together with the full history and a measurement based on my own Search Console data.

The definition is on Google’s page on spam updates: “While Google’s automated systems to detect search spam are constantly operating, we occasionally make notable improvements to how they work. When we do, we refer to this as a spam update”. As an example, Google names its AI-based system SpamBrain, which is improved from time to time.

That sentence holds the most important point, because the detection itself is not switched on and off and only what changes noticeably in it gets announced. John Mueller described it like this in August in a comment under Google’s LinkedIn announcement: “We have lots of systems that update continuously (most of them update automatically), but when we make bigger / broader changes to the systems, we try to announce them.” Search Engine Journal also reported the reply.

That gives you a useful rule for any fluctuation: if your rankings move on a day without an announced update, the ongoing systems are a candidate. Or the cause is not at Google at all.

The three terms are close to each other. Google’s own documentation separates them clearly:

Sources: Google documentation on spam updates, core updates and manual actions, retrieved on 24.09.2026. “No documented notification” means: the two update pages name none.

On manual actions Google writes: “Google issues a manual action against a site when a human reviewer at Google has determined that pages on the site are not compliant with Google’s spam policies.” Both paths therefore run through the same policies. The difference lies in who decides and whether you find out.

How a core update can be checked is something I worked through with my own data in the guide to Google core updates. I use the same method for spam updates further down.

The table contains every entry Google lists as a spam update in the ranking release history. Start and end come from the individual entries, the announcement is in Google’s wording.

Source: own analysis of the publicly available entries in the Google Search Status Dashboard (ranking release history and incident entries), read on 24.09.2026 and for the September entry and all runtimes again on 08.10.2026. All times US/Pacific. Google only added the three entries from 2021 and 2022 in March 2023. Linked are the updates that have their own article here.

I read three things from the list. First, the duration varies a lot. The median of all twelve updates is 7 days and 13.5 hours. But the range runs from 19.5 hours to 29 days. Anyone who draws a conclusion on day two of an update therefore does not yet know whether they are at the start of a long rollout or already at the end of a short one.

Second, from March 2024 to September 2026 Google gave an expected duration for every spam update. So far the measured runtime roughly matched that order of magnitude. Where Google wrote “up to”, however, the reported duration was slightly longer three times, by one, two and 21 hours. The fourth time, in September 2026, it stayed about eight hours below the announced two weeks.

Third, the gaps are getting shorter: for 2024 the list shows three and for 2026 so far four spam updates against at most two in any other year. Google does not say why. I am not going to guess.

One statement from Google itself fits this list poorly. On October 2, 2026, Gary Illyes showed slides on how long crawling, indexing and serving take at Search Central Live Deep Dive in Barcelona. Search Engine Roundtable published photos of them. For spam updates the slide says “1 to 2 days (rollout)”. Search Engine Roundtable renders this as a rollout duration of one to two days. The list above does not support that. Only two of the twelve completed spam updates were done after at most two days, the September update took 13 days and 16 hours. On the slide, however, the value sits at the start of the time axis, at the minimum. Read that way, it is the lower bound and not the typical case. Google itself calls the values “estimations based on internal analysis”. Illyes later called the slides an exercise in a comment on LinkedIn.

What a spam update can respond to is set out in Google’s spam policies. The page was last updated on 28.08.2026. As of 25.09.2026 it groups the violations into these 16 sections, plus a section on other practices such as legal removals:

• Cloaking · Doorway abuse · Expired domain abuse · Hacked content
• Hidden text and link abuse · Keyword stuffing · Link spam · Machine-generated traffic
• Malicious practices · Misleading functionality · Scaled content abuse · Scraping
• Site reputation policy · Sneaky redirects · Thin affiliation · User-generated spam

The entries in the release history from October 2023 to September 2026 name none of these targets. From December 2024 on, the core of each entry is the same: global and in all languages with an expected duration. If third parties write that a specific spam update targeted AI texts or bought links, that is their interpretation and not a Google statement. It is not confirmed.

Google has linked the policies to an update itself once. With the updates of March 2024, Google announced new rules, among them expired domain abuse and a stricter version of scaled content abuse. Here the reason was public for once.

1. Open the “Manual actions” report. If there is a manual action, there is a documented individual decision. It does not rule out a parallel update, but it is a concrete, documented finding. Fix it first and then request a review. Only after that can the development be sensibly assessed together with possible update effects.
2. Wait for the release history. As long as it says “Information” instead of a duration, the rollout is not documented there as complete.
3. Compare two weeks. For core updates, Google recommends waiting at least one full week after completion and then comparing with a week before the start. That is written for core updates. I consider it a sensible minimum for spam updates.
4. Look at queries, not totals. New posts and seasonal queries shift the overall numbers. A single query that drops from page 1 far down the results tells you more. And a click loss is not yet a ranking loss, because clicks also fall at the same position when demand drops.
5. Check the window for disruptions. If a core update, a known counting error or a change in Search Console falls inside the window, the comparison is not clean.

The method is the same as in the core update guide. The week before is the seven days before the start day. The week after begins on the first full day that lies at least seven days after the end. Start and end come from the dashboard in US/Pacific. I retrieved the data on 24.09.2026 via the Search Console API (web search for the property https://www.seo-kreativ.de/).

As in the core update guide, I counted all queries with at least ten impressions in both weeks. I count it as a large loss when a query was on page 1 before and beyond position 20 afterwards. That threshold is my own. Precisely: an average position of at most 10.0 before and above 20.0 after. A change of at least one position counts as better or worse, anything smaller as stable. The classes are mutually exclusive: a large loss is not counted again as worse. I evaluated the query dimension across all countries and devices. The position is the average that Search Console reports per query across a week’s impressions, not a fixed position for all users, devices and locations.

The two large losses were “crawl budget optimization” after the March update, from position 2.6 to 32.7. After the August update it hit a single knowledge question about Google’s indexing system, from 7.2 to 21.3. Two single queries across three updates do not add up to a pattern for me.

The limits belong here. For the March update, the week after from 02.04.2026 to 08.04.2026 falls right inside the March core update, which began on 27.03.2026 and ended on 08.04.2026. Whatever moved there cannot be assigned to either update. For the August update, the tracker spikes of 12.08.2026 and 13.08.2026, which I checked against my own numbers at the time, fall into the week before. And everything here concerns a small website with a few hundred URLs where my own editorial and technical review has found no obvious violation of the spam policies.

What transfers is therefore not the result but the approach, because for a website with a real violation of a policy the same measurement would look different. By how much, I cannot tell from my data.

For the September 2026 spam update, the week after begins on 16.10.2026 under the same rule. I will add the measurement here once that week is complete in Search Console.

How large the interest is shows in search volume. I queried it at DataForSEO on 24.09.2026. The values are the provider’s estimates:

Demand follows the running rollout and not the month in the name, as September 2025 shows: the August update ran until 22.09.2025 and the peak falls in September.

How much rankings move is quantified by an analysis by SE Ranking in Search Engine Land of 27.08.2026. In the analysed sample, 16.71 percent of URLs previously in the top 10 fell beyond position 100 for the respective keyword during the August 2026 update. In a July comparison window without a confirmed update it was 9.2 percent. The measurement covered 100,000 keywords from 20 industries in Google’s organic results in the US. According to the report, the data does not show why these particular pages lost. Nor does it show a share of affected websites, only position movements in a US sample.

For Germany I have no comparable measurement. I pulled up the Semrush Sensor for mobile on 24.09.2026. It shows unrest across the SERPs, not in your own data. According to its own notice, Semrush recalculated the formula for it on 18.09.2026. I therefore do not read comparisons across 18.09.2026 as a finding.

I am framing this as conditions, not as a recipe.

1. Hold the policies honestly against your own website. That is Google’s only concrete recommendation: “Sites that see a change after a spam update should review our spam policies to ensure they are complying with those.”
2. Plan in weeks to months, not days. Google writes: “Making changes may help a site improve if our automated systems learn over a period of months that the site complies with our spam policies.” The “may” is part of it. The slide by Gary Illyes is more specific. For a change after a spam update it typically gives one to two weeks with the note “continuous”. In the slowest case it is months with the note “batch refreshes”. I read it this way: what the continuously running systems pick up can take effect after weeks. What depends on a recalculation in batches takes months.
3. Stay sober about links. After a link spam update, cleaning up may have no effect, because according to Google an earlier benefit from devalued links is lost. How I assess incoming spam backlinks is a different question. The policy describes links that are created mainly to manipulate rankings, bought ones for example.
4. Request a review if there is a manual action. Google describes the route in the report itself: fix, then request a review. Google’s page on spam updates names no such request. For lifting it, the same slide gives two to three days at the earliest and typically one to two weeks. In the slowest case it is four to six weeks, much longer for dormant sites.

What I do not recommend: large rebuilds while it is unclear which policy you violate. A rebuild based on a suspicion is at best ineffective.

Twelve spam updates in just under five years, four of them in 2026 alone. The list shows when Google changed something about spam detection and how long the rollout took each time, but not what changed.

That is why the order matters. First the manual actions, then the policies, then a clean measurement window. My own measurement shows how quickly the window tips: a core update three days later is enough.

I update this page with every new spam update. The reports on the individual updates are collected in the Google Updates archive.