Site reputation abuse: what the evidence actually shows

More than two years after Google’s first manual actions under the site reputation abuse policy (May 2024), the practice it targets, renting a trusted domain’s ranking signal to third-party content, has become a byword in SEO for something that no longer works. Whether that claim rests on evidence or on a handful of loud examples is a separate question, and one Google’s own documentation does not answer.

What Google defines#

The policy’s wording, current as of its 28 August 2026 update, is precise about mechanism, not just intent: it “applies where third-party content is published on a host site mainly because of that host’s already-established ranking signals, which it has earned primarily from its first-party content.” The examples Google gives are specific: an educational site hosting sponsored payday-loan reviews, a medical site hosting a low-quality “best casinos” page written by someone else. The policy carves out wire services, syndication, user-generated forums, editorial columns, advertorial and affiliate links used appropriately as acceptable, which is a narrower target than “any third-party content” would suggest.

Google’s announcement did not name an enforcement mechanism. Its March 2024 announcement of the new spam policies said sites that violate them may rank lower or not appear at all, with the policy taking effect from 5 May 2024, and did not say whether site reputation abuse would be enforced by manual actions, algorithms or both. Search Engine Journal later reported that Google has indicated plans for algorithmic updates, which gives later reporting something to be checked against.

What has actually been observed#

The manual actions are real and named. Search Engine Journal’s coverage in December 2024 reported manual actions against CNN Underscored and WSJ Buyside (per Glenn Gabe) and a heavily impacted Forbes Advisor (per Lily Ray), timed days before Black Friday that year. Aleyda Solis’s analysis in the same article found visibility declines, not confirmed manual actions, for USA Today’s Reviewed, Newsweek’s Vault and The Sun UK’s shopping section. Gabe and Ray described entire folders losing indexed pages. This is the strongest tier of evidence available on this topic: a named outlet, attributing specific claims to named practitioners, about specific named sites.

Algorithmic enforcement is not confirmed to have shipped. Glenn Gabe, writing on his own site in July 2024, said enforcement to that point had been manual-only, with the algorithmic component he expected still unreleased. A Capconvert retrospective published in May 2026 describes the same manual-only status as of that date. Neither source is Google itself, and no Google statement found in this research confirms or denies that the algorithmic component has since shipped inside a later, unlabeled spam or core update. That gap between what Google has indicated it will build and what independent trackers can observe is the most honest thing that can be said about current enforcement.

No published number exists for total enforcement scope. Not Google’s policy page, not its March 2024 announcement, not the August 2026 EEA enforcement change, states a count of pages, sites or manual actions. Every figure attached to this policy in public circulation, including the query-loss numbers Capconvert attributes to Forbes Advisor and Fortune Recommends, comes from third-party visibility tools reading a domain’s search performance, not from Google disclosing anything.

The selection bias in cases#

Every named example in this piece, and every other one in circulation, exists because a specific person was already watching that specific domain’s data when a drop happened, and chose to publish it. Forbes Advisor and CNN Underscored are large enough that a subfolder going to zero shows up instantly in rank-tracking tools that SEO practitioners already had open. A mid-size publisher, or a smaller site running the exact same third-party subfolder arrangement, could receive an identical manual action and never produce a public case study, simply because nobody with an audience happened to be watching. That means the set of “confirmed” cases is not a sample of enforcement, it is a sample of enforcement that was both large enough and visible enough to someone already looking. Treat the list of named publishers as a lower bound on scope, not a census of it.

What this means for publishers#

The policy’s own wording, and the named cases so far, both point the same direction: a subfolder that exists mainly to borrow a host’s authority, with little first-party editorial involvement, is the specific target, not any third-party content by definition. That much can be stated with confidence, because it comes from Google’s own documentation. What cannot be stated with confidence is how common enforcement is beyond the publishers big enough to be watched, or whether the algorithmic detection Google has indicated is planned is now running quietly. A publisher weighing a licensed-subfolder deal should read the named cases as proof the policy is real and enforced against sites with real editorial resources, and read the absence of a published enforcement count as a reason not to assume a smaller, quieter arrangement is safe simply because it has not made the SEO news cycle yet.