Doorway pages penalty — diagnose and fix gateway-page clusters

You have built dozens, hundreds, or tens of thousands of pages that each target a slight variation of the same intent — "{service} in {city}", "cheap {service} {city}", "{service} near {city}" — and every one of them hands the visitor off to the same booking form, the same contact page, or the same generic offer. The pages rarely differ beyond a swapped place name in the title and an introductory sentence. On the SERP side the giveaway is that these URLs barely surface for anything except the exact permutation in their title, and even then they sit on page two with a fragile grip. When enforcement lands it lands across the whole permutation set at once: the cluster loses impressions in a single step, not page by page. If a manual action attaches, Google Search Console shows "Doorways" under Security & Manual Actions, pointing at the spam policy at https://developers.google.com/search/docs/essentials/spam-policies. Algorithmically there is no notice — the orphaned permutation pages simply stop being indexed, the Discovered-not-indexed bucket swells, and the few that survive cannibalize each other. Internally the tell is a navigation that no human would design: a footer or sitemap dump of thousands of city links pointing into pages that exist only to catch a search and pass the user along.

Location or query permutation sets that all resolve to one destination

The defining doorway shape: a single offer wrapped in a generated page for every city, ZIP, or phrasing variant, each one routing the visitor to the same form or landing page. Google treats the whole set as one door multiplied, because the genuine destination is identical and the permutation pages add no standalone value of their own.

Orphan pages reachable only by search, not by site navigation

Doorways are frequently built to be found by Google and ignored by humans — they sit outside the real navigation, linked only from a mass footer block or an XML sitemap. When a page exists solely to intercept a query and pass the user onward, that orphan status is itself a structural signal that the page was made for engines rather than people.

Near-identical body copy with only the entity noun swapped

When the only thing that changes between two URLs is the city or keyword token while the surrounding paragraphs, headings, and calls to action are byte-for-byte the same, the cluster reads as a templated funnel. The pages are not differentiated by genuinely different information, so each one duplicates the intent of every sibling rather than serving a distinct need.

Scaling for keyword coverage instead of for distinct user needs

The pages were created to occupy a matrix of keyword cells — every service crossed with every location — rather than because each cell represents a question a real user has that no other page answers. Coverage-driven generation is precisely the intent Google's doorway policy describes, and it is the root cause that consolidation has to undo rather than disguise.

1. 1

   Check Search Console → Security & Manual Actions for an explicit "Doorways" notice — if it is there you have a defined reconsideration path and should read its wording before changing anything.

2. 2

   Run pseolint across your full sitemap and open the doorway-pattern findings first — that rule is built to flag permutation sets that share intent and a destination, which is the exact shape you are diagnosing.

3. 3

   Read the near-duplicate findings alongside it to quantify how much body copy is shared across the cluster; two URLs that differ only by a swapped noun are a doorway pair, not two pages.

4. 4

   Read the thin-content findings to confirm the permutation pages carry little standalone value beyond the entity swap and the funnel link out to the shared destination.

5. 5

   Read the template-diversity findings to measure how mechanically uniform the cluster is — low diversity across thousands of URLs is the structural fingerprint of a generated doorway set.

6. 6

   Trace internal linking: if the affected URLs are reachable only from a footer dump or the sitemap and never from real editorial navigation, mark them as orphan permutation pages destined for consolidation or removal.

7. 7

   Apply the test to every surviving URL — "would this page exist if search engines did not?" — and sort the cluster into keep-and-strengthen, merge, or remove based on the honest answer.

pseolint findings most strongly correlated with this pattern.

A home-services franchise ran 9,400 programmatic pages of the form "{trade} in {city}" across four trades and roughly 2,350 towns, every one of them ending in the same national booking widget. Organic impressions to the set were already thin when a manual action for doorways landed on February 12, 2025, wiping the cluster's residual traffic overnight. The team consolidated to 4 substantive trade hub pages plus 180 genuine service-area pages for towns where the franchise had a physical branch, real local pricing, and named technicians; the surviving pages each linked into the relevant hub and out to a town-specific (not national) contact path. The remaining 9,216 orphan permutation URLs were 410'd in a single batch. Reconsideration was filed February 27, 2025 and the manual action was lifted 21 days later. The consolidated hubs, which had previously been buried under the permutation noise, climbed into the top five for the head trade terms and recovered an estimated $58,000 of monthly booked-job revenue within four months — more than the doorway set had ever produced.

If a manual action is attached, the clock is bounded by reconsideration: requests are typically reviewed within 14 to 28 days, and the action lifts as soon as the reviewer agrees the doorway pattern is genuinely gone rather than disguised — so the gating factor is the honesty of your consolidation, not Google's queue. For an algorithmic demotion there is no human in the loop; the cluster's suppression eases only as Google re-crawls and confirms the permutation pages are removed or merged, which means planning for 30 days from fix to the first movement and 60 to 90 days to a stable new equilibrium. The orphan URLs you 410 fall out of active scoring within about 45 days, and as they leave the host's median quality rises, which can lift the surviving hubs independent of the doorway fix itself. Do not expect the freed traffic to land back on the old URLs — it consolidates onto the substantive pages, so track the hubs' impressions and head-term positions as your recovery signal rather than watching the retired permutation set.