Google Manual Actions: Reading the Notice and Fixing It

A manual action notice means a person on Google's search quality team reviewed your site and decided it violates one of the Spam Policies badly enough to act directly, rather than letting the ranking algorithms sort it out on their own. It appears in the Security & Manual Actions report inside Search Console, and it usually shows up as a sudden, unexplained drop in visibility for a section of the site or the whole domain — the kind of drop that doesn't line up with any algorithm update on the public timeline.

The notice itself is short: an action type, a short note, sometimes two or three example URLs, and a scope. It does not hand you a full list of the pages or links that caused it, and it does not tell you how to fix it — that part is on you. Getting it lifted means identifying exactly which manual action type you're dealing with, since "unnatural links" and "thin content" call for completely different remediation, then filing a reconsideration request specific and evidenced enough that a reviewer can verify your fix in minutes rather than guess at it.

What the Manual Actions Report Actually Tells You

Each entry in the report is built from four parts worth reading slowly instead of skimming past. The match tells you the scope — a specific URL, a directory prefix, or the entire site. The action type is one of a fixed set of spam-policy categories Google maintains, not a description written fresh for your case. The note is a short, human-written explanation that sometimes includes two or three sample URLs. The status shows whether the action is active or already resolved, along with the date it was applied.

What the report does not give you matters just as much. Google does not publish the complete list of URLs flagged for thin content, nor the complete list of backlinks flagged as unnatural. The example URLs in the note are a sample of a pattern, not an inventory of the problem. Treat the notice as a description of a policy violation, not a bug report with a reproducible list of line items — fix only the two or three URLs it names and leave the pattern they represent untouched elsewhere on the site, and the same action gets reapplied the next time someone reviews it.

Unnatural Links: To Your Site vs. From Your Site

"Unnatural links to your site" means Google believes links pointing at your domain were built, bought, or exchanged specifically to manipulate PageRank, rather than earned because someone found the content useful enough to reference. Typical triggers are paid guest posts with keyword-rich anchor text and no disclosure, large-scale link exchanges or three-way link schemes, widget or badge links embedded across many unrelated sites, and links from low-quality directories or article-mill sites that exist mainly to host outbound links.

"Unnatural links from your site" is the mirror image: your own pages are passing manipulative links outward, whether you built them that way or someone else did. Common causes include sponsored or affiliate links published without a nofollow or sponsored attribute, widget code you distribute to other sites that embeds a followed link back to you, guest posts you accepted in exchange for payment without marking the links, and comment or forum sections on your own domain that spammers filled with followed links, sometimes without anyone on your team noticing.

The fixes diverge from there. For links pointing at you, the realistic path is contacting webmasters to request removal where you can identify who controls the site, and disavowing the rest — a disavow file with no removal attempts behind it reads as an unreviewed link dump, not remediation. For links leaving your site, the fix is marking them sponsored or nofollow, or deleting the links outright, and auditing every place your site accepts user-generated content, since that's usually where the outbound spam accumulated without editorial review.

Thin Content, Pure Spam, and Structured Data Issues

"Thin content with little or no added value" targets pages that exist mainly to attract search traffic without giving the visitor anything a search results page hadn't already told them. The usual sources are templated location or category pages that only swap a place name into an otherwise identical paragraph, doorway pages built purely to funnel traffic toward one destination, content scraped or lightly reworded from another site, and affiliate pages that just reformat a manufacturer's product description with an outbound link attached. The fix is either a substantive rewrite that adds real, page-specific information a reader can't get from the search results themselves, or removing, redirecting, or noindexing the pages that can't clear that bar — consolidating ten thin pages into one genuinely useful one usually outperforms trying to save all ten.

"Pure spam" is reserved for the most aggressive violations: cloaked pages showing search engines and visitors completely different content, auto-generated gibberish stitched together purely to rank, or sites that exist as a funnel toward a scam. Recovery from a pure spam action is a different order of difficulty than the others, because the underlying judgment is that the site has no redeeming purpose, not that one section needs cleanup. A reconsideration request here usually needs to demonstrate a genuine change in what the site is and who runs it, not a patch to a few pages.

"Structured data issues" means the markup on the page — schema.org, JSON-LD, or microdata — describes something the page doesn't actually show or doesn't actually have: review-star markup with no visible reviews, inflated ratings, availability or pricing markup that doesn't match what a visitor sees, or event and job-posting markup applied to pages that aren't really events or jobs. Run the Rich Results Test against a sample of URLs from each template on the site, not just the flagged examples, and either correct the markup to match the visible content exactly or remove it from templates that can't.

Site Reputation Abuse: Google's Newest Manual Action

"Site reputation abuse" targets a specific and newer pattern: an established site with its own real ranking signals hosts content from a third party — on a subdirectory, a subdomain, or a syndicated section — mainly so that third party can borrow the host's authority, with little or no editorial oversight from the host and little value for the host's actual readers. The pattern most often cited is a coupon operator, a loan-comparison service, or a casino-review outfit paying a news outlet or established blog to publish under its domain.

What separates this from a garden-variety manual action is that the violation isn't about links or thin pages in the abstract, it's about whose content this really is and how much genuine editorial control the host exercises over it. Labeling the section "sponsored content" does not resolve it if the host isn't actually reviewing, fact-checking, or standing behind what gets published there. The realistic fix is either removing or deindexing the third-party section entirely, or bringing it under real editorial ownership — the same review process the host applies to its own staff-written content, not a rubber stamp.

Partial Match vs. Site-Wide Match

The match field in the notice is not a formality. A partial match names a specific URL, a directory prefix, or a clearly bounded section, and only that scope loses visibility — the rest of the site keeps ranking on its normal signals while you work. A site-wide match suppresses the entire domain, including pages that individually did nothing wrong, because the violation Google identified is treated as a property of the site as a whole rather than of one section.

That distinction should shape your recovery plan, not just your understanding of the damage. A partial match on a link action, for instance, often traces to one campaign or one acquisition channel — a specific link-building vendor's output, or one syndication partner — and fixing that one source can be enough. A site-wide match on the same action type usually means the pattern is spread across many pages or many months of activity, and a reviewer checking your fix will sample broadly across the domain, not just re-check the handful of URLs the original notice happened to name. Fixing only the visible examples on a site-wide notice is one of the most common reasons a first reconsideration request gets rejected.

Diagnosing the Real Cause Before You Touch Anything

For a link-based action, start with Search Console's own Links report and export as much of your backlink profile as it will give you, then cross-check against any third-party backlink tool you already have access to — no single source sees every link, so triangulating matters. Look specifically for footer or sitewide links appearing across dozens of unrelated domains, clusters of low-traffic sites that link heavily outward and to each other, a pattern typical of link networks, and anchor text that reads like a keyword list rather than something a human would actually write.

For a content-based action, crawl your own site organized by template rather than by page, since thin content problems are almost always a template problem, not a one-off. Group pages by the pattern that generated them — city pages, product-variant pages, tag archive pages, auto-assembled "best X for Y" listicles — and check each template for real word count, uniqueness against your own other pages, and whether the substantive content is just a data feed with a name swapped in.

For a structured-data or reputation action, spot-check the example URLs the notice gave you first, but don't stop there: run the same check across every template or every third-party section on the site, since the notice's examples are illustrative, not exhaustive.

• Pull the Links report from Search Console and cross-reference with any backlink tool already in use
• Group flagged content by the template that generated it, not by individual URL
• Run the Rich Results Test across a sample from every template, not only the notice's examples
• Check every place the site accepts outside content — comments, guest posts, syndicated sections — for editorial oversight gaps

Writing a Reconsideration Request That Gets Accepted

A request a reviewer can act on quickly has three parts, in this order: what you found when you investigated, stated specifically in your own words rather than a vague admission that "something was wrong", exactly what you changed and when, and what you've put in place so it doesn't recur. A reviewer is deciding whether the underlying pattern is actually gone, not whether you're sorry it happened.

The mistakes that sink a first request are consistent. Disavowing an entire backlink profile without reviewing which links actually match the flagged pattern reads as an unreviewed dump, not an investigation. Submitting the request minutes after receiving the notice, before any real work happened, is easy for a reviewer to spot. Blaming "a previous SEO agency" without showing what was actually removed answers a question nobody asked. And resubmitting an almost-identical request after a rejection, without directly addressing the reason the rejection gave, wastes a review cycle that can take weeks.

Evidence carries the request, not adjectives. Keep a dated log of every link you had removed or disavowed, with URLs, rather than asserting that you "cleaned up your backlinks." Keep a before-and-after list of pages you rewrote, merged, or removed, rather than asserting that you "improved content quality." Reference that log directly in the request text, and where the fix involved outside parties, such as a link removal request sent to another webmaster, note that the request was sent even in cases where the other site never responded, since a documented attempt still counts as remediation.

• State the specific cause you found, not a general admission of wrongdoing
• List exactly what changed, with dates and URLs where relevant
• Reference removal requests you sent, even ones that got no reply
• Address the exact reason given in any prior rejection — don't resubmit the same text

After You Submit: Timeline, Rejection, and Where Black-Hat Tactics Fit In

Google states that most reconsideration requests are processed within a few weeks, though there's no fixed deadline, and site-wide or structurally complicated cases regularly take longer. The report's status only changes once a decision is made — submitting a second or third request while the first is still pending does not speed anything up, and can push your case to the back of the queue instead.

A rejection comes with a stated reason. Treat that reason literally and address exactly what it names rather than resubmitting a lightly reworded version of the same request — reviewers see repeat submissions, and a request that ignores the specific point raised the first time reads as not having understood the feedback at all.

Many manual actions originate from tactics that fall under what's commonly called black-hat SEO — private blog networks built purely to pass links, large-scale reciprocal link exchanges, buying expired domains specifically to inherit their old authority, and other techniques search engines have long treated as policy violations rather than legitimate optimization. A manual action is the enforcement response to a detected instance of one of these, not a misunderstanding to argue away, and a site with a documented history of them can expect closer scrutiny on every subsequent review, including on manual actions that look, on their face, unrelated to the earlier one.