Doorway pages and scaled content abuse: what they mean for links

A doorway page is a page built to rank for one of a group of similar searches and then pass the visitor on to somewhere else. Scaled content abuse is the mass production of pages, by software or by people, mainly to manipulate rankings. Both are named in Google’s spam policies. They matter for off-page work in two ways. Many of the sites that sell links are built from exactly this kind of content, and many of the pages people want to build links to, such as sets of city or keyword pages, are close to the doorway definition. A link cannot rescue either. So the policies tell you something about where a link is worth placing and something about which of your own pages deserve one.

What Google’s policies say

Google’s spam policies page, as read in October 2026, defines doorway abuse as sites or pages created to rank for specific, similar search queries, which lead users to intermediate pages that are not as useful as the final destination. It gives four examples. The first is having multiple websites with slight variations to the URL and home page to maximise reach for a query. The second is having multiple domain names or pages targeted at specific regions or cities that funnel users to one page. The third is generating pages to funnel visitors into the usable or relevant part of a site. The fourth is creating substantially similar pages that are closer to search results than to a clearly defined, browseable hierarchy.

The idea is old. In a Search Central blog post dated 16 March 2015, Brian White of Google’s webspam team announced a ranking adjustment aimed at doorway campaigns and described the harm in plain terms: a searcher clicks one result, dislikes it, tries the next result and lands on the same site. The post gave site owners questions to ask of their own pages. Among them were whether the pages are an integral part of the site’s user experience or exist to funnel visitors, whether they duplicate useful aggregations that already exist on the site, and whether they exist as an “island” that is difficult or impossible to reach from other parts of the site. That last question is still the quickest test available. A page that no menu, category or internal link leads to was not built for the people already on the site.

Scaled content abuse is newer as a name. The same policies page defines it as generating many pages for the primary purpose of manipulating search rankings and not helping users, and says the practice is typically focused on large amounts of unoriginal content that provides little to no value, no matter how it is created. The examples include using generative AI or similar tools to generate many pages without adding value, scraping feeds or search results to generate pages, stitching together content from different pages without adding value, and creating multiple sites to hide the scaled nature of the content. The page also states the consequence for every policy on it in one sentence: sites that violate the policies may rank lower in results or not appear in results at all.

What changed in March 2024

On 5 March 2024 Google announced three new spam policies alongside a core update: expired domain abuse, scaled content abuse and site reputation abuse. The Search Central post published that day says the scaled content policy builds on the previous policy about automatically generated content, so that Google can act no matter whether content is produced through automation, human efforts, or some combination of the two. Before that date the written policy was framed around automation. An operator paying writers a few pounds an article to turn out hundreds of thin pages could argue that nothing was automatically generated. After that date the argument no longer has anything to stand on, because the wording is about purpose and scale and says nothing about the tool.

The same post addresses AI directly. It says Google’s long-standing position is that the use of automation, including generative AI, is spam if the primary purpose is manipulating ranking in search results, and that the updated policy is in the same spirit. That cuts both ways. A site is not in breach because it used a language model, and a site is not safe because a person typed every word. In a companion post on Google’s company blog the same day, Elizabeth Tucker, a director of product management for Search, said Google expected the changes, combined with its earlier work, to reduce low-quality, unoriginal content in search results by 40 percent. That is Google’s own forecast of its own results, and no independent measurement of it is cited here.

The three policies announced that day are close relatives. Site reputation abuse covers third-party pages placed on a strong host, and expired domain abuse covers low-value content placed on a domain bought for its history, which our page on expired domains explains. All three describe ways of getting content to rank on the back of something other than its own worth, and a single operation can breach more than one.

How both show up on sites that sell links

A site that earns its income from placements has a strong reason to publish as many pages as it can, on as many subjects as buyers ask for, at the lowest cost per page. Each new article is a unit of stock. The result is a familiar kind of publisher: hundreds or thousands of articles of similar length, on unrelated topics, written to a pattern, each carrying one or two outbound links to commercial sites. Whether the articles were generated or commissioned makes no difference to the policy. The purpose test is what applies, and a page whose reason for existing is the link inside it was not made to help a reader. Google’s link spam policy has a matching line, which lists creating low-value content primarily for the purposes of manipulating linking and ranking signals. Our lesson on the link spam policy goes through that list example by example.

Generative AI has made this cheaper. A seller can now fill a new domain with a year’s worth of apparent publishing history in an afternoon, and the lesson on AI in link building describes how that changes the supply of placements. The policy example about creating multiple sites to hide the scaled nature of the content describes the next step, where one operator runs dozens of such sites under different names and sells them as separate publishers. The shared templates, hosting and writing patterns that tie those sites together are covered in the lesson on link footprints, so they are not repeated here.

Doorway structures appear on the selling side as well. Some publishers keep thin city or keyword pages that exist to hold sponsored links, and in high-risk markets practitioners describe satellite networks: many small sites, each aimed at a narrow group of queries, all passing visitors and links to one commercial site. That is the first example in Google’s doorway definition almost word for word. For a buyer the point is that the label on the listing, whether “guest post”, “niche blog” or “local news site”, tells you nothing about which of these structures you are paying to appear on.

What it means when you choose where to place a link

No published study measures what a link from a scaled content site is worth, and Google does not publish how it treats outbound links from a site it has classed as spam. What can be said rests on the policy wording and on ordinary reasoning. A site that may rank lower or not appear at all is a site whose pages may not be in the index, and a page that is not indexed passes nothing. Even where the page stays indexed, a publisher whose every article links out to buyers presents the pattern that link spam systems are built to discount. The likely outcome is the quiet one described in our lesson on link penalties and recovery: the link stays live, the report looks fine, and nothing moves. Treat that as a working inference, because nobody outside Google can show it link by link.

The practical checks follow from the definitions. Read five recent articles and ask whether any of them would exist if nobody had paid for the link inside. Look at how many unrelated subjects the site covers, how fast it publishes, and whether the articles share a length and a structure. Check whether the site earns search traffic at all, since a site with thousands of pages and almost no visitors has usually been judged already. Then look at the single page you would be buying. The publisher page checker shows how many sites a page links out to, how those links are marked and whether the site advertises placements. The fuller routine, with thresholds, is in how to vet a publisher.

Your own placement matters too. An article written only to carry your link, on a site that publishes anything, adds one more page to the pile. If you supply the content, as with paid guest posts, write something the publisher’s readers would want without the link in it.

What it means for the pages your links point to

Links are usually bought or earned for commercial pages, and the commercial pages most at risk under the doorway policy are sets: one page per city, one page per keyword variation, one page per product colour. Picture an invented removals firm with a single depot that publishes forty pages, each titled with a different town, each with the same paragraphs and the town name swapped in, and each sending the visitor to the same quote form. That is Google’s second doorway example. Pointing links at those pages does not change what they are. Links add to a page’s ability to rank when the page is otherwise acceptable, and they do nothing to answer a judgement about the content. Money spent building links to a doorway set is spent on pages that may be demoted or removed together.

So the order of work matters. Before a set of pages goes into a link plan, apply the 2015 questions to it. Check that each page can be reached through the site’s own navigation and that it holds something the others do not, such as a real address, named staff, local prices or work done in that place. Where the honest answer is that the pages differ only by a place name, merge them into one strong page and build links to that. Our guide for local businesses covers what a location page needs, and which pages need links covers how to pick targets once the pages themselves are sound.

The same reasoning applies to content produced in bulk as link bait or as a home for internal links. A hundred generated articles published to make a new site look established fall under the scaled content definition as plainly as a seller’s inventory does, and any links built to them inherit the risk. Fewer pages, each of which a person in the market would choose to read, give links something to work on.

Where the lines are unclear

Scale alone is not the offence. Large retailers, directories and job boards publish many similar pages from templates, and Google’s wording targets pages generated for the primary purpose of manipulating rankings with little value to users. Purpose and value are judgements, and Google does not publish a threshold for either. Nobody outside Google can say how many near-identical pages turn a site section into a doorway campaign, or what share of thin articles turns a publisher into a scaled content site. The useful habit is to ask the question the policy asks, which is whether the page would be worth publishing if search engines did not exist.

Enforcement is also only partly visible. The 2024 announcement says a site hit by a spam manual action is told through Search Console and can ask for reconsideration. Algorithmic demotion comes with no notice, and a link buyer is never told that a publisher has been demoted. That is why the checks above have to be made before each order and repeated for publishers you return to, because a site that was sound a year ago may have changed owner and filled up since.