The story that will get told about this is that Google stopped punishing parasite SEO in Europe. That reading survives about one paragraph of the actual documentation. What Google described is not the absence of a consequence. It is the substitution of one consequence for another, and the substitute is the version that quietly makes the arrangement worthless instead of loudly making it stop.
What changed in the site reputation abuse policy
The site reputation abuse policy arrived with the March 2024 core update and targets a specific arrangement: third-party content published on a strong host domain, with little oversight from that host, primarily to borrow its ranking authority. Coupon sections on news sites, casino reviews on university subfolders, sponsored commercial pages parked under a masthead that spent thirty years earning trust. Google began enforcing it in Europe in January 2025. It has been the single most consequential manual action category for large publishers ever since, and we have written about what a manual action actually costs a content operation.
On August 28, Barry Schwartz reported the documentation change at Search Engine Land. From August 30, 2026, manual actions issued under this policy will no longer impact search results for users in the European Economic Area. The trigger was a European Commission investigation into the policy itself. A Google spokesperson framed it as a concession rather than a retreat, saying European users are no less frustrated by parasite SEO and other deceptive, pay-to-play tactics that degrade search results, and that the company stands by the policy.
Four things did not change, and they matter more than the headline. Google will still issue the manual action. Site owners will still be notified in Search Console. Reconsideration requests are still available. And algorithmic actions, the automated systems that handle most of this at scale, are untouched in the EEA and everywhere else. What was removed is one specific lever: the manual, human-reviewed demotion, applied to European results only.
| MECHANISM | REST OF WORLD | EEA FROM AUG 30, 2026 |
|---|---|---|
| Manual action issued | Yes | Yes |
| Search Console notification | Yes | Yes |
| Manual action changes rankings | Yes | No |
| Affected section separated in Google's systems | Not described | Yes, over time |
| Algorithmic spam systems apply | Yes | Yes |
| Reconsideration request available | Yes | Yes |
Separation is not a softer demotion
The replacement language is the part worth reading twice. Google's documentation says the affected section of the site may be separated in its systems so that, over time, it ranks independently from the rest of the site. Strip the phrasing and you get the mechanism: the subfolder stops inheriting the host domain's authority and has to earn its position on its own.
Now consider what the arrangement was for. Nobody pays a national newspaper to host casino comparison pages because the newspaper's editorial team writes excellent casino comparisons. They pay for the domain. Authority inheritance is the entire product. A demotion suppresses that product and can be lifted on appeal. Separation deletes the product and there is nothing obvious to appeal, because from the outside it does not look like an enforcement event at all. It looks like the pages simply never took off.
“A manual action is a door being closed. Separation is the room being moved to a different building, with no sign on the old door.”
There is a real asymmetry here in how quickly each is felt. A manual action is a step function: the notification arrives, positions fall, everyone in the room knows the date. Separation is a decay curve with no announcement attached to it. Teams that monitor for penalties by watching for sudden drops will not catch this, because the shape of the signal is wrong. It resembles ordinary content fatigue, and it will get misdiagnosed as ordinary content fatigue for at least a quarter.
Illustrative shape of the two mechanisms over time, not measured traffic. The point is the shape of the curve, not the values.
Two enforcement regimes on one URL
This is the first time Google has applied a spam policy differently based on where the searcher is, which is a genuinely new thing to have to model. Until now, a page was either penalized or it was not, and geography entered the picture only through localization and language. From Sunday, a single URL under a manual action can be suppressed for a reader in Chicago and untouched for a reader in Cologne.
For any business with meaningful European organic traffic, that breaks a reporting assumption that has been safe for a decade: that a global average position describes a coherent thing. It now blends two enforcement regimes, and the blend moves whenever your traffic mix moves. A site whose European share grows will show what looks like recovery. A site whose American share grows will show what looks like a fresh penalty. Neither is real.
If you rent space on somebody else's domain
Two audiences need different advice here, and most coverage collapses them into one.
If you are the brand renting the space, your European exposure did not improve. It changed shape. Under the old regime you would have known within days whether the arrangement had been caught, because the host would have received a notification and told you, or your traffic would have fallen off a cliff. Under separation you get neither. Your pages keep ranking for a while and then stop, and by the time the pattern is legible you have spent another two quarters of budget on placements that were already severed. Ask your host publisher directly whether they have received a manual action, in writing, and put it in the contract for the next renewal. That question is now the only fast signal available to you.
If you are the publisher, you are the one carrying two regimes on one property. The commercial section that pays for the newsroom is still penalized in the United States and still generating revenue in Germany, and those two facts now have to live in the same board deck. The strategic question is unchanged and just got easier to defer: whether hosted commercial content is worth what it does to the host domain's standing. Publishers have been having that argument since the policy landed, against a backdrop of search traffic declining unevenly by publisher size, and this change removes some of the urgency without removing any of the risk.
For regulated categories the calculus is sharper still. Financial comparison, insurance and gambling are where these arrangements concentrate, and they are also the categories where algorithmic systems are most active. If you operate in fintech, assume the algorithmic layer is doing most of the work against you already, and that the manual layer was never your main exposure.
How to instrument the site reputation abuse policy split
The instrumentation problem is small and completely solvable, which is why it is worth doing this week rather than after the first confusing monthly report. You need every metric that could be affected by a site reputation manual action to be reported EEA and non-EEA separately, from Sunday forward, so that you have a clean before and after.
The ratio in step five is the useful number. If the EEA view holds while the non-EEA view falls, you are seeing the manual action working exactly as documented. If both fall together, that is the algorithmic layer, and no carve-out is going to help you. If the EEA view drifts down slowly over months while the non-EEA view stays flat at its new lower level, that is separation doing its work, and it is the case that would have been invisible without the split. Reporting this way is the same discipline we apply when we build dashboards on first-party data rather than vendor defaults.
One caution on the snapshot. Take it before Sunday if you can, because the value of a before reading falls to zero the moment the change lands. If you are reading this after August 30, take it anyway and label the gap honestly rather than quietly comparing against a period that already includes the change.
Do this before Sunday
Three moves, in order, and none of them takes longer than an afternoon. First, list every path on your domains that carries third-party or sponsored commercial content you do not editorially control. Most enterprise teams find at least one they had forgotten about, usually a legacy partnership that outlived the person who signed it. Second, take the geo-split snapshot above so that Sunday is a boundary in your data rather than a smudge. Third, if you are buying placements, get a written answer from each host on whether they are currently under a manual action, because you no longer have a reliable way to infer it from performance.
Then decide the strategic question on its own merits rather than on enforcement risk, because enforcement risk is now the least predictable input you have. Hosted commercial content either serves your readers and belongs on your domain, or it is authority arbitrage that a regulator has temporarily made cheaper in one market. The arbitrage version was always a rented position, and the last two years should have taught everyone that rented positions get repriced without notice. Building the editorial content that earns its own authority is slower and it does not get carved out by anybody's investigation.
The reasonable counterargument deserves a hearing: for a publisher whose newsroom is genuinely funded by these sections, an eighteen-month reprieve in a major market is real money, and telling them to walk away from it is easy advice to give with somebody else's balance sheet. Fair. Take the reprieve if you need it, but take it with the timer running and a written plan for what replaces the revenue, because Google was clear that it stands by the policy and the Commission's investigation will not stay open forever. The carve-out is a pause in one mechanism, in one market, under one regulator. It is not a verdict.
See where you are cited today
A free snapshot audit of your rankings and AI citations before we ever talk.
Josh leads work at the intersection of SEO and generative engines at Something Inc., helping B2B brands get ranked and cited across every major AI engine.