Google logged the August 2026 spam update on its Search Status Dashboard at 09:27 Pacific on Tuesday, August 18, and marked it complete on August 21. Two days and sixteen hours, global, all languages. No blog post, no target list, no new policies published alongside it.
That last detail is the one worth sitting with. Nothing about what Google considers spam changed in August. Enforcement of the existing spam policies got faster and broader.
We got the question from clients within a day, and it's a fair one: we publish a lot of content, built from a query fan out map, on a fixed cadence. Does that look like the thing Google just hunted?
Short answer, no. Longer answer below, policy by policy, because "trust us" isn't an audit trail.
What the update actually was#
Three confirmed spam actions in 2026, and the rollout windows keep collapsing. What used to take Google most of a month now finishes inside three days. That's an enforcement pipeline running on classifiers, not a review queue.
Practically, it means two things for anyone publishing at volume:
- You don't get a warning window. The April back button policy came with two months of notice because it was a new rule. Enforcement passes come with none.
- Because nothing new was published, a drop you see this month traces back to a policy that's been sitting on the page for years, usually scaled content abuse or site reputation abuse.
Query fan out, and what it isn't#
Query fan out is how answer engines work, not a tactic we invented. A person asks one question, the system decomposes it into a set of related sub queries, retrieves passages for each, and synthesizes an answer. Google described this behavior in its own documentation of AI features.
Our content strategy reads that behavior backwards. We map the sub questions an engine is likely to generate around a buying decision in your category, find which of those you have no credible passage for, and commission writing that answers them properly.
So the fan out map determines which questions get answered. It doesn't determine how many pages we spin up, and it never produces a page whose only reason to exist is a phrase.
Here's the distinction that matters against the policy, stated plainly:
"Scaled content abuse isn't about volume. It's about whether the page was made for a person or for a ranking. Publishing forty useful answers a month is a content operation. Publishing forty near duplicates of one answer is the violation."
Andrew Bethel, COO, Perfectus Labs
Against the published policies, one at a time#
Scaled content abuse. Google's rule targets pages generated at scale that provide no original value, regardless of how they were produced. Our guardrails: one page per distinct question, a named human author with a real byline, subject matter review before publish, and a rule that any draft which restates an existing page gets merged into it instead of shipped beside it. Volume is capped by tier and by cadence, not by how many keyword variants a tool can produce.
Site reputation abuse. Also called parasite SEO. We don't rent, buy, or place content on third party domains to borrow their authority, and we don't sell space on yours. Everything we publish for you lives on your domain, under your brand, tied to your entity.
Keyword stuffing and doorway pages. Fan out means we write about more topics, not the same topic in more places. Location and service pages exist only where there's a real service, a real address, or a genuinely different offer behind them. If two pages would serve the same intent for the same audience, that's one page.
Cloaking and sneaky redirects. What Googlebot, GPTBot, and a person see is byte for byte the same document. The structured data we ship describes what's visibly on the page, which is also why we don't mark up reviews, prices, or FAQs that a reader can't see.
Link spam. We don't buy links, run private networks, or exchange them. Citation and link work in Cognos is monitoring and outreach, and we've never presented links as the mechanism behind AI visibility, because they aren't.
Expired domain abuse and machine generated content at scale. No domain acquisitions for authority laundering, no unreviewed model output going straight to publish. Models help with research and structure. A person owns the argument, the facts, and the byline.
What we changed after August 18#
Nothing structural, and I want to be honest about why: there was nothing new to comply with. What we did do is tighten the review pass.
- Ran a duplication sweep across every managed content library, comparing pages by intent rather than by title, and merged the near overlaps we found.
- Re-checked every structured data block against what's visible on the page.
- Pulled Search Console manual action and security reports for every account, all clear.
- Added an August enforcement note to each client's weekly report with the ranking movement in their category, so the conversation starts with data instead of a rumor.
If you saw movement this month#
Work in this order. It's the fastest path to a real answer.
- Check for a manual action. Search Console, Security and Manual Actions. If there's nothing there, this was algorithmic, and there's no reconsideration request to file.
- Separate the update from the noise. Compare August 18 to 21 against the two weeks before it, by landing page, not sitewide. A spam action usually hits a cluster, not everything.
- Audit the cluster that dropped. Ask a blunt question about each page: would a person who asked this question be glad they landed here? If the honest answer is no, that page is the problem, not the algorithm.
- Look at what you inherited. Old programmatic page sets, syndicated content, guest post campaigns, and third party subfolders are where most of these demotions live.
Our position#
We build for retrieval, and retrieval rewards the same things Google's spam policies protect: original substance, a real author, a clean technical surface, and an entity a machine can verify. That's a happy alignment rather than a compromise, and it's the reason an enforcement pass is a reporting exercise for us instead of a fire drill.
If you're not sure where your content library sits against these policies, we'll go through it with you and show you the pages we'd merge, fix, or kill.
Source: Google Search Essentials spam policies and Google's Search Status Dashboard entry for the August 2026 spam update, Aug 18-21, 2026.
