On September 24 Google posted a short notice on the Search Status Dashboard: the September 2026 spam update has been released, applies globally and to all languages, and the rollout may take up to two weeks. It is the fourth spam update of 2026 — the previous one came in August. That is where the official information ends.
What Google Said — and What It Did Not
The announcement says nothing about which types of spam the update targets. Google did not mention AI content, links, doorways, or percentage losses. Everything published in industry media about a "strike on AI content farms" and "up to 90% of rankings" is the authors' interpretation, not Google's statement. We stress this because decisions about your sites are better made on what is known, not on what sounds loud.
What is known for certain: spam updates are updates to Google's automated anti-spam systems (Google calls them SpamBrain), and they enforce the existing spam policies. Those policies are published, and they make it clear what is always at risk.
Which Google Spam Policies Matter for Affiliates
The policies did not change for this update — but they are exactly what the system re-applies. For affiliate sites the relevant points are:
- Scaled content abuse. Many pages generated for ranking purposes with no value to users — regardless of whether they were made by an AI model, a script or by hand. Google states directly that the method does not matter, the result does.
- Doorways. Sites or pages built for specific queries to funnel users elsewhere. Hundreds of "city" pages for one offer is the classic example.
- Expired domain abuse. A purchased dropped domain with history loaded with irrelevant content to exploit inherited authority.
- Site reputation abuse. Placing third-party content on an authoritative domain to ride its rankings — what the market calls "parasite SEO".
- Sneaky redirects and cloaking. Showing Google one thing and users another.
- Scraped content. Rewrites and aggregation of others' material without added value.
If a site is built on anything from this list, it is at risk with every spam update, and September is no exception.
What to Do During the Rollout
No sudden moves for the first two weeks. While the update rolls out, rankings fluctuate and cannot be used to judge the outcome. Google itself recommends assessing impact after the rollout completes — the "completed" status will appear on the Search Status Dashboard.
Tell algorithmic from manual. If the Manual Actions section in Search Console is empty, the drop is algorithmic. Manual actions arrive with a notification and are lifted via a reconsideration request; algorithmic ones only through fixing the site and waiting for re-evaluation.
Check your assets against the list above. Pre-landers on separate domains, article-site networks for traffic funnelling, geo-query pages, dropped domains — these deserve the first review. Not because Google announced anything, but because these formats are exposed by definition.
If rankings drop after the rollout completes — Google says sites that fix violations can recover, but it takes time: the system needs to re-crawl and re-evaluate pages, and that usually means months, not days. There is no quick return after a spam filter.
What This Means for Affiliates
For those running paid traffic, spam updates barely touch ad campaigns: Google Ads and organic search are different systems. But they do touch infrastructure: if pre-landers or white pages live on domains that also try to collect organic traffic, a drop can hit the domain's trust overall.
For those working with SEO traffic to offers, this is another reminder that the "thousand generated pages for long-tail queries → redirect to the network" model lives from update to update. Google restarts its spam filters four times a year; each pass catches what the previous one missed.
Observations from your own sites during the rollout — what dropped, what rose — are being collected in the general questions section of the forum. In two weeks, when Google marks the update complete, we will return to it with data rather than guesses.