The most repeated claim about this policy is that enforcement became fully algorithmic in August 2025. As far as anyone can document, it didn't — and Google has confirmed to industry press that neither the March nor June 2026 spam updates targeted site reputation abuse at all.
That matters, because a great deal of current guidance is built on an inference presented as a fact. Two years into the policy, here's what's actually documented, what's genuinely disputed, and the one change from late 2024 that closed the defence most publishers were relying on.
The timeline, with the disputed part marked
| When | What happened |
|---|---|
| March 2024 | Policy introduced alongside scaled content abuse |
| May 2024 | Enforcement begins — manual actions only |
| November 2024 | Policy updated: first-party involvement no longer a mitigating factor |
| August 2025 | Spam update, 27 days, SpamBrain — targeted content-level spam, not this policy |
| November 2025 | European Commission proceedings opened |
| March 2026 | Spam update, under 20 hours — Google confirmed it did not target this policy |
| May 2026 | Spam policy definition rewritten to cover AI Overviews and AI Mode |
| June 2026 | Spam update, ~2 days — first pass after the May redefinition |
The claim that August 2025 flipped a switch is doing a lot of work in the current literature. What Google has said is that it intends to make enforcement algorithmic. What it has published is a policy enforced through manual actions since launch, and confirmations that the recent spam updates targeted something else.
How to hold this honestly Treat the shift toward algorithmic enforcement as a direction, not a documented event. The practical effect is identical either way: the authority you used to inherit is no longer reliably there.
The reason to be precise about it is practical rather than pedantic. If enforcement is still manual, you get a notification in Search Console and a defined recovery path. If it's algorithmic, you get neither — you just lose visibility with no explanation. Those are very different situations to plan for, and guides that assert the second may be sending people to look for a signal that doesn't exist.
The change that closed the main defence
The most consequential detail in the whole policy, and it post-dates most of the articles still ranking for this topic.
In November 2024, Google updated the policy to remove the level of first-party involvement as a mitigating factor. The current position: no amount of oversight changes the third-party nature of content if its primary purpose is exploiting the host domain's ranking signals.
That closed the defence almost everyone was preparing. The argument had been "we have an in-house editor reviewing these pages, so it's not really third-party content." After November 2024, that argument does nothing. The test is purpose, not process.
Which also means there's no recovery path through editorial improvement. You cannot fix a site reputation abuse problem by making the content better, because quality was never what the policy was measuring.
What recovery actually requires
Three things, and the first surprises people.
Remove the content, don't relocate it. Moving the section to a different subfolder, subdomain or URL structure doesn't address the issue — the arrangement is the violation. Nor does improving the pages.
File a reconsideration request documenting what was removed. Specifics, not intentions. What existed, what was taken down, what has changed about how you'll operate.
Expect it to take a while. Practitioner-reported manual action recovery timelines generally run to a couple of months, though these figures come from agencies rather than from Google and should be treated as indicative. If the loss turns out to be algorithmic rather than a manual action, there is nothing to file and the wait is longer — closer to the pattern in what recovered and what didn't after the 2026 core updates.
That last step is worth doing precisely because it's the report almost nobody opens — one of the ones worth checking only when something is wrong, as covered in the Search Console reports that matter.
Nightwatch
Accurate daily rank tracking tool with beautiful visual reporting dashboards
Best for: Daily Rank Tracking
The scope kept widening
The policy was written about rented subdomains on news sites. What it now catches is broader, and several categories affected weren't in anyone's original reading.
- Publisher commerce sections. The well-documented cases — affiliate and deals subfolders on major news and business publications — were the first and most visible.
- Guest post arrangements. A guest article that sits outside the host's subject area and exists mainly to rank fits the test, whatever it's called. This connects directly to the shifting economics in the state of link building in 2026.
- Affiliate content without genuine oversight, including product reviews with no evidence the reviewer handled the product — no original photographs, no measurements, no testing.
- Self-serving rankings. Brands publishing "best X" listicles that place themselves first without substantiation are exposed under a related reading of the same principles.
The common thread isn't the commercial arrangement. It's whether a section of a domain exists to serve that domain's audience or to borrow its ranking signals for somebody else's benefit.
The policy is itself under challenge
An uncertainty almost nothing written on this topic mentions.
European Commission proceedings opened in November 2025. The question is not whether Google may have such a policy, but whether its specific application aligns with the obligations the Digital Markets Act places on gatekeepers — essentially, whether a dominant search provider can remove publisher sections from results in this way.
Google reportedly offered remedies including adjusted application for news domains and increased transparency. Press reporting in May 2026 indicated the Commission considered that offer insufficient, and Google's public position has been that the priority is protecting users from deceptive practices.
The outcome is unresolved. Which means anyone planning around this policy in the EU is planning around something that could change — worth monitoring through Commission primary sources rather than trade coverage, since this is exactly the kind of story where secondary reporting runs ahead of the record.
What changed in May 2026, and why it matters more
Arguably the more significant development of the past year, and it isn't specifically about site reputation abuse.
On 15 May 2026, Google rewrote its spam policy definition to cover AI Overviews and AI Mode. Manipulating AI-generated answers is now explicitly spam, and the June 2026 update was the first enforcement pass after that change.
That extends the same logic to a new surface. The tactics that grew up around borrowed authority in conventional results have obvious analogues in generated answers — and the policy now says so. Anyone building a strategy around getting cited in AI answers should read that as a boundary rather than a technicality, alongside the legitimate approaches in generative engine optimisation, what AI Mode means for organic traffic and what's working in AI Overviews.
Where the borrowed authority went instead
Worth noting, because demand for third-party visibility didn't disappear when the subdomain arbitrage closed.
It moved toward platforms that were never rentable in the first place. Community sites, forums and video now occupy a substantial share of commercial results, which is the subject of where Google is sending clicks now.
The important distinction is that participating in those communities as a genuine contributor is not the same activity as leasing a subfolder. One earns visibility from an audience that can reject you; the other buys it from a domain that can't. The first is durable and slow; the second was fast and is now largely closed.
Which is the honest strategic read. The tactic that produced quick results is gone, and what's left is the slower work of being a recognised source in your own domain — structuring content for authority, building a strategy that compounds, and earning mentions rather than renting them. That was always the better answer; it's now the only one.
What publishers should take from it
Three things, if you host third-party content commercially.
The revenue was real and the risk is now priced in. Publisher commerce sections funded newsrooms. Pretending that arrangement was costless was always optimistic; pretending it's still available is worse. The question is what replaces the revenue, which is a business model question rather than an SEO one.
Partial measures don't work. Tightening editorial oversight on a section that exists to rank for terms unrelated to your publication doesn't change its classification — the November 2024 update settled that.
Your main domain isn't automatically safe. The enforcement pattern has been to act on the section rather than the whole site, which is better than the alternative but means a subfolder can be removed from results while the rest continues. Monitor sections independently rather than watching only sitewide traffic, or you'll see the decline as a gentle slope rather than as one section going to zero.
What this means if you were never doing it
Most readers of this weren't leasing subdomains. The policy still changes something for you, in two ways worth naming.
The competitors who outranked you may have been renting. If a category you compete in was dominated by publisher subfolders and affiliate sections, some of that competition has been removed — and the terms are now winnable by whoever has genuine depth on the subject. That's an opportunity with a shelf life, since everyone else in the category is noticing the same gap. A content audit against those specific terms is a reasonable next step.
The tactics adjacent to it got riskier. Guest posting, sponsored placement and syndication all sit closer to the line than they did in 2023. None are prohibited, and the distinguishing question is consistent: is this content serving the host's audience, or borrowing the host's signals? A guest article on a site whose readers genuinely want it is fine. The same article placed on twelve unrelated domains is the pattern the policy describes.
There's a broader read here too. Across scaled content, site reputation abuse and the May 2026 extension into AI answers, Google has been enforcing one principle in three places: content that exists primarily to rank rather than to be read is the target, and the mechanism of production or placement is beside the point. That same logic runs through the rights and provenance questions in where the copyright rules stand for AI-generated content — the direction of travel is consistently toward asking who this was actually made for.
Which is not a comfortable answer for anyone whose programme was built on volume, and it is a genuinely good one for anyone whose wasn't. Plenty of practitioners reached the same conclusion from a different direction some time ago.
Signals worth watching
- Search Console Manual Actions, the only place a notification appears.
- Section-level visibility rather than domain-level, so a removal doesn't hide inside an average.
- Google's Search Central spam policy page, where changes are made officially — the May 2026 rewrite appeared there.
- Commission primary sources for the DMA proceedings, rather than trade summaries.
- Whether Google ever announces the algorithmic version it has said is coming. That announcement would change the recovery picture materially.
And if you commission content that gets published elsewhere, the risk assessment has changed enough to be worth revisiting — the pricing and positioning shifts in how agencies are pricing AI-assisted work reflect the same underlying move away from volume tactics. If placement was a meaningful part of your acquisition mix, replacing it is a content strategy problem now, not a media buying one.
The short version
The claim that enforcement went fully algorithmic in August 2025 appears not to be documented — Google confirmed the March 2026 update didn't target this policy, and the same was reported of June. Enforcement has been manual since launch, with Google saying an algorithmic version is intended. Treat the shift as directional. The decisive change was November 2024, when first-party involvement stopped being a mitigating factor, which removed the editorial-oversight defence entirely and means there is no recovery through improving the content. Recovery requires removal plus a documented reconsideration request; relocating a section does nothing. The scope has widened well past rented subdomains to guest posts, unsupervised affiliate content and self-serving listicles. The policy itself is under European Commission examination with remedies reportedly rejected in May 2026. And the more significant recent change may be the May 2026 rewrite extending spam policy to AI Overviews and AI Mode.
The arbitrage is closed. What replaces it is the work it was always a shortcut around, which is not a satisfying answer but is the accurate one — and it's the same conclusion reached in brand trust in the age of AI-generated everything.
Hosting third-party sections and unsure where you stand?
We audit which sections would rank on their own, then rebuild the visibility on ground you actually own.
Audit my exposure →Frequently asked questions
Is site reputation abuse enforcement algorithmic or manual?
This is widely misreported. Many guides state that enforcement became fully algorithmic in August 2025, but Google has not published such a mechanism and the policy has been enforced through manual actions since launch. Google confirmed to industry press that the March 2026 spam update did not target site reputation abuse, and the same was reported of the June 2026 update. Google has said it intends to make enforcement algorithmic; treating that as already complete is an inference rather than a documented fact.
Does having editorial oversight protect third-party content?
Not since November 2024, when Google updated the policy to remove the level of first-party involvement as a mitigating factor. The current position is that no amount of oversight changes the third-party nature of content if its primary purpose is exploiting the host domain's ranking signals. That closed the most common defence, which was to argue that an in-house editor reviewed the pages, and it post-dates a great deal of the guidance still circulating on this topic.
How do you recover from a site reputation abuse manual action?
By genuinely removing the violating content, then filing a reconsideration request through Search Console that documents what was removed and why. Editing the pages, improving them, or relocating them to a different subfolder does not resolve the issue, because the problem is the arrangement rather than the quality of any individual page. Reported recovery timelines for manual actions generally run to a couple of months, though published figures come from practitioners rather than from Google.
Is the site reputation abuse policy being challenged?
Yes. European Commission proceedings opened in November 2025, examining not whether the policy may exist but whether its specific application aligns with the Digital Markets Act obligations that apply to gatekeepers. Google reportedly offered remedies including adjusted application for news domains and increased transparency, and press reporting in May 2026 indicated the Commission considered that offer insufficient. The outcome remains unresolved and is worth monitoring through primary sources rather than trade coverage.
Does the policy affect guest posting and sponsored content?
It raises the risk profile considerably. The test Google applies is whether third-party content is being published primarily to exploit the host domain's ranking signals rather than to serve the host's own audience. Guest posts, sponsored sections and affiliate content that sit outside a site's established subject area and exist mainly to rank are exposed under that test, regardless of how the arrangement is labelled or how much review it receives.
