The lowest quality pages
The rater guidelines reserve their lowest rating for three families of pages: harmful, untrustworthy, or spammy. Any single criterion is enough to justify it. That rating punishes nothing, it describes. The practices it names do have a documented consequence, but it lives elsewhere, in the spam policies.
A screening grid, not a scale
Part four of the rater guidelines carries a blunt title, Lowest Quality Pages. It describes what someone paid to evaluate results must recognise as the bottom of the scale, and it opens with an inversion worth reading slowly.
Rating a page begins with two screening steps: establish its true purpose, then assess its potential to cause harm. A page whose purpose is harmful or deceptive receives the lowest rating before the quality of its content is examined at all.
officialSource : Google, “General Guidelines”, version of 11 September 2025, section 4.0
The quality of the text therefore comes afterwards. A page can be documented, illustrated, fast and well written: if its real purpose is to deceive about what it is, the evaluation stops before reaching the text. That order explains why so many checklists miss the subject. They start with the tags.
Any one of the lowest-rating criteria is justification for that rating on its own: the document presents its grid as a list of independent grounds, not as an accumulation of points.
officialSource : Google, “General Guidelines”, version of 11 September 2025, section 4.0
Nothing offsets anything here. A page does not buy back one ground with its qualities elsewhere, and that is what separates this grid in kind from the composite scores tools sell.
Three families, and the ones that actually concern a publisher
The document sorts lowest quality pages into three families: harmful, untrustworthy, spammy. The first covers content that encourages harm to people or to specified groups. It does not concern the vast majority of sites, and this guide does not dwell on it.
That leaves the other two, which hold the practices search marketing argues about every day: missing information about the author where trust is required, a site with a very negative reputation, a page whose main content is deliberately obscured, and above all the three abuses named below.
Scaled content abuse
Scaled content abuse means generating many pages for the purpose of primarily benefiting the website owner rather than helping visitors, no matter how those pages were created.
officialSource : Google, “General Guidelines”, version of 11 September 2025, section 4.6.5
The examples in the document are precise: producing many pages with automated tools, scraping feeds or search results, stitching together pieces of other pages, opening several sites to hide the real scale of the operation.
Two wordings coexist, and the gap between them teaches something. The rater guidelines speak of pages produced primarily to benefit the site owner; the spam policies speak of pages generated for the primary purpose of manipulating rankings. The first text describes what a reader observes, the second names a violation. Not the same sentence, and not the same consequences.
The use of generative AI tools alone does not determine the level of effort or the page quality rating: the document writes that such tools serve to create high quality content as readily as lowest quality content.
officialSource : Google, “General Guidelines”, version of 11 September 2025, section 4.6.6
Here is the most misquoted sentence in the whole document. It does not say AI is risk-free, and it does not say AI disqualifies a page either: it moves the question from the tool to the result, judged on three axes the text repeats, effort, originality and added value for the visitor.
Google now penalises AI-written content. Our rewriting service humanises your texts and removes the risk of a ranking drop.
Site reputation abuse
Site reputation abuse is publishing third-party content on a host site mainly because that host already holds ranking signals earned by its own first-party content. Hosting third-party content is not abuse in itself: what defines it is the reason for publishing.
officialSource : Google Search Central, Google Search spam policies
This is the practice the market sometimes calls parasite hosting. A solid site rents part of its addresses to pages that have no business being there, because they will rank better there than anywhere else. The published examples leave no room for doubt: loan reviews on an educational site, a casino page on a medical site, white-label coupons on a news site.
The list of excluded cases matters just as much, and gets quoted far less: wire service copy, syndication between publications, forums and contribution spaces, columns and opinion pieces, affiliate links treated appropriately. What characterises the abuse is neither payment nor the third-party origin of the text, but the reason for publishing it.
The repurposed expired domain
Expired domain abuse means buying an expired domain name to host content of little or no value to visitors, counting on the reputation of the former site. The document cites affiliate content placed on the former address of a government agency.
officialSource : Google, “General Guidelines”, version of 11 September 2025, section 4.6.3
The document describes the manoeuvre without hedging. The former site was often a government agency, a charity or a school, and the buyer counts on that reputation. It even gives the verification it expects from a rater, and it is within any reader’s reach: find the earlier version of the site in a web archive, then observe the change.
What the text does not say, on the other hand, is that an expired domain is burnt. A takeover coherent with the previous activity matches no line of the definition.
What this rating triggers, and what it does not
A low rating given by a rater has no effect on the site rated: the documented consequences of content produced at scale belong to the spam policies, which are a different text with a different mechanism.
hypothesis
The status is that of a reading, not a quotation. The document holds no sentence saying a low rating leaves the rated site untouched; it holds, in its introduction, the sentence saying no rating moves a result, and it describes no circuit anywhere between a rating and a site. We draw a conclusion, and it is displayed as one.
The documented consequence does exist, and it lives elsewhere. The three abuses described above appear in the spam policies, and breaching those can lead to a manual action, notified with its reason. Confusing the two texts means fearing an imaginary mechanism while ignoring the real one.
Rereading your pages with the screening grid rather than a technical checklist
hypothesis
- Effort
- moderate
- Time to effect
- quarters
- Risk
- none
- Durability
- structural
The effort stays modest, because the grid fits into two questions. The time-to-effect axis is the longest on this site, for an owned-up reason: the rereading produces no effect by itself, it changes what gets published next. The risk is nil, none of these steps being a technique. Durability is good, and that can be checked rather than trusted: this document has been reissued under the same title for years, with a date and a version number, which makes editions comparable.
What does not work, or no longer works
Having a mass-produced corpus rewritten by hand, first. The manufacturing method appears in none of the definitions; what is named is the ratio between the number of pages and the value they bring. Three hundred rewritten pages remain three hundred pages without a reason to exist.
Then opening a section far from the site’s subject because the domain is well established. That is the definition of site reputation abuse, and it targets the motive, not the writing quality of the result.
Finally, looking in this document for a rating you could aim at. It contains none. It contains grounds for refusal, which is the exact opposite of a target.
Sources
- General Guidelines, version of 11 September 2025, 182 pages — Googleaccessed on
- Google Search spam policies — Google Search Centralaccessed on
Block revalidated on
Frequently asked questions
- Is a “lowest” rating a penalty?
- No. It is produced by a person paid to evaluate results, and the same document writes that no rating moves a result on the page. What does punish, where it exists, is called a manual action, is notified in the console with a reason, and rests on the spam policies rather than on this grid.
- Is content written with AI rated lowest quality?
- Not because of the tool. The document says so plainly: the use of generative tools alone determines neither effort nor rating. What counts is the result, judged on three named axes, effort, originality and added value for the visitor. A human text lacking all three falls in the same box.
- What is site reputation abuse?
- Publishing third-party content on a host site mainly because that site already ranks well. The policies give examples: sponsored payday loan reviews on an educational site, a best-casinos page on a medical site. Opinion columns, user contributions and wire service copy are explicitly excluded.
- Is a repurchased expired domain doomed?
- No: what is named is buying one in order to exploit the former site's reputation while hosting content of no value. Buying a domain and continuing a coherent activity matches no line of the definition. The check offered to raters is to compare the current content with what a web archive holds.
- How do I know whether my page falls into one of these families?
- By taking the document's two screening steps in order: what is this page's true purpose, and who does it benefit first? A page that exists mainly for the site rather than for its reader answers the first question badly, and no improvement in form changes that answer.