Doorway Pages: Google’s Four Examples, and the City-Page Arithmetic That Turns Local SEO Into Spam

Doorway pages are pages, or whole sites, created to rank for a set of closely similar searches and then pass the visitor on to somewhere more useful. Google treats them as spam. Its spam policies call the practice doorway abuse, and its Manual Actions report names doorways as one of the causes of a “thin content with little or no added value” penalty. The classic modern example is a service business with one real offer and a hundred city pages that differ only by the town name. Below are Google’s definition and four examples verbatim, the test I apply to each, and the arithmetic that shows why a templated location page set is mostly duplicate text however carefully it is written.

How Google defines doorway abuse

Quoted from Google’s spam policies for web search, last updated 28 August 2026, checked today.

Doorway abuse is when sites or pages are created to rank for specific, similar search queries.

Google’s spam policies, doorway abuse

The policy then lists four examples, also verbatim:

• “Having multiple websites with slight variations to the URL and home page to maximize their reach for any specific query”
• “Having multiple domain names or pages targeted at specific regions or cities that funnel users to one page”
• “Generating pages to funnel visitors into the actual usable or relevant portion of a site”
• “Creating substantially similar pages that are closer to search results than a clearly defined, browseable hierarchy”

Two words carry the whole policy: similar and funnel. A doorway is not defined by how many pages you have or how they were produced. It is defined by pages that target near-identical searches and exist to hand the visitor on. That is also what separates it from its close relative, scaled content abuse, which the same policy page describes as “when many pages are generated for the primary purpose of manipulating search rankings and not helping users.” Plenty of doorway sets are scaled content as well, but a handful of hand-written pages can still be doorways if each one only exists to catch a variant of the same search.

The penalty it triggers

Google says it detects spam “both through automated systems and, as needed, human review that can result in a manual action”, and that sites that violate the policies “may rank lower in results or not appear in results at all.” When a reviewer is involved, doorways land under a specific notice in the Search Console Manual Actions report. Its description reads: “Google has detected low-quality pages or shallow pages on your site.” The examples listed under it are thin affiliate pages, content from other sources, and doorways. If your notice names that action, start with manual action or algorithmic demotion to confirm what you are dealing with, and with my guide to thin content, because the same notice covers both.

A test for each of Google’s four examples

The examples describe outcomes, not methods, so I turn each one into a question that can be answered from a crawl and a look at the pages. The tests and fixes are mine, not Google’s.

The fourth test is the one most sites fail without meaning to, because faceted navigation and tag archives can produce “substantially similar pages” on their own. My site architecture guide covers the hierarchy side, and index bloat covers the cleanup.

The city-page arithmetic

This is a worked model, not client data. Every input is an assumption and I have labelled them. It is here because it shows what a doorway set looks like as a dataset rather than as a page, which is how a crawler meets it.

Assume a service business with real premises in 8 towns and a service area of 120 towns. The doorway version publishes one page per town, 120 pages, each 600 words. Of those 600 words, 540 are the shared template (service description, guarantees, process, FAQ) and 60 change per page (the town name repeated, a line about the area, a postcode list).

The doorway set publishes 72,000 words, and 64,800 of them, 90 percent, are the same 540 words repeated 120 times. The consolidated version assumes 8 branch pages of 600 genuinely different words each (address, team, opening hours, the jobs actually done there) plus one 900-word service-area page that lists all 120 towns. It publishes 5,700 words, every one of them unique to the page it sits on, on 111 fewer URLs.

Look at what changes and what does not. The consolidated site still mentions all 120 towns, so nothing is lost for a reader checking whether you cover their area. What disappears is 111 URLs whose only distinct content was a town name, which is precisely the “pages targeted at specific regions or cities that funnel users to one page” in Google’s second example. The model’s limits: real templates vary more than a fixed 540 words, and Google publishes no duplicate-share threshold, so 90 percent is not a line you cross. It is a description of what the crawler is being asked to index.

When a location page is not a doorway

Location pages are not banned. A page earns its place when it carries something a visitor in that town needs and cannot get from the main service page. The checks I use:

• A real place. An address, opening hours or a named team that exists in that town.
• Local proof. Jobs, projects, reviews or photos from that location, not stock copy with the name swapped in.
• A different answer. Prices, availability, regulations or delivery times that actually differ by area.
• A place in the hierarchy. The page is linked from a locations index in the navigation, not only from the sitemap.

A page that passes two or more of these is a location page. A page that passes none is a doorway with a map embedded in it.

Recovering from a doorway penalty

Google’s recommended steps for the thin content manual action include identifying doorway pages on your site, improving the site so it delivers substantial value, and then selecting Request Review with examples of what you removed and what you added. In practice the order I follow is: inventory every templated set, decide keep, merge or remove for each page against the checks above, 301 merged pages to their nearest real equivalent, return a 404 or 410 for pages with no equivalent, and only then write the review request. Google’s own help page warns that “most reconsideration reviews can take several days or weeks.” How to write the request so it is accepted is in Google penalty removal, and the full diagnostic process is in my Google penalty recovery guide.

A common mistake is to respond to a doorway action by rewriting the 120 town pages with more filler. Longer near-duplicates are still near-duplicates. The fix is fewer pages, not bigger ones. The same logic sits behind cloaking and site reputation abuse: Google’s spam policies judge what a page is for, not how it was dressed.

Related reading

• Google Penalty Recovery: how I diagnose and reverse a traffic collapse
• Thin content: there is no word count, and there is no thin content penalty
• Manual action penalty or algorithmic demotion: how to tell which one hit you
• Google penalty removal: the reconsideration request that gets accepted
• Unnatural links: the two manual actions Google names
• Google Business Profile suspended: the appeal steps and the evidence Google accepts
• Generative engine optimization: what the research measured, and what Google says