You diagnose this by watching for a cluster-level pattern across Google Search Console’s Performance and Page Indexing (Index Coverage) reports: geo-modifier pages that ranked and then collectively declined together, pages getting consolidated or canonicalized to a single URL despite being built as separate location pages, and impressions/clicks concentrating on far fewer geo variants than were published, rather than any single explicit signal that says “doorway page” anywhere in GSC. There is no disclosed algorithmic doorway score visible in any Google tool; this is inference from indexing and ranking behavior, evaluated against Google’s own doorway pages policy criteria, not a metric you can look up directly.
What GSC actually shows before enforcement
Google’s Search Console doesn’t have a “doorway pages detected” flag. What it does have is Performance report filtering by URL pattern (regex or contains-match on a URL path segment), which lets you isolate the entire set of geo-modifier pages (e.g., everything matching a /locations/ or /[city]-[service]/ pattern) and look at their aggregate and per-URL trends over time, and the Page Indexing report (formerly Index Coverage), which shows why individual URLs are or aren’t indexed, including a “Duplicate, Google chose different canonical than user” status.
The pattern worth watching for across these two reports:
Cluster-wide ranking decline that moves together. If a set of geo pages ranked adequately for a period and then declined as a group, roughly simultaneously, rather than individually based on each page’s own content quality or backlink profile, that’s more consistent with an algorithmic reassessment of the whole template/pattern than with normal page-by-page ranking fluctuation. Filter Performance by the URL path pattern for the geo-page set, segment by query, and look at whether impressions and average position moved in a correlated way across the set on a similar timeline, rather than a few individual pages losing ground for page-specific reasons (like a competitor outranking one particular city page while the rest hold steady).
Google choosing a different canonical than the one specified. The Page Indexing report’s “Duplicate, Google chose different canonical than user” status is one of the more direct signals available. If a meaningful share of the geo-modifier URL set shows this status, it means Google’s own systems have judged the pages similar enough to each other (or to a hub page) that they don’t merit separate indexing, which is functionally adjacent to what the doorway pages policy is concerned with: multiple pages built to rank for close-variant queries without offering genuinely distinct value to justify separate URLs. This isn’t proof of a doorway-pages classification specifically, but it’s evidence Google’s algorithmic systems (not a manual reviewer) are already treating the set as low-differentiation.
A widening gap between published/indexed pages and pages that actually earn impressions. Compare the count of geo-modifier URLs that are indexed (via the Page Indexing report) against how many of those URLs show any impressions at all in the Performance report over a recent date range. A large and growing share of indexed-but-zero-impression geo pages, especially if that share is increasing over time rather than reflecting normal long-tail thinness, suggests Google’s ranking systems are declining to surface most of the set even though they remain technically indexed, which again lines up with how a doorway-style pattern would behave under algorithmic (not manual) suppression.
Ranking convergence toward a single “best” page instead of many location-specific ones. Search a sample of the target geo+service queries (ideally from a clean/incognito session, or check Performance’s query-level data filtered to the geo page set) and see whether one or two of the location pages (or the hub/service page) increasingly show up for queries that were originally meant to be served by many individual city variants. Google’s ranking systems doing this consolidation on their own, promoting one representative page over the long tail of near-duplicate geo pages, is a strong behavioral echo of the doorway pages policy’s core concern.
Grounding this in Google’s actual doorway pages policy
Google’s Search Central documentation on doorway pages describes them as pages (or sets of pages) created to rank for specific, often overly similar, search queries, funneling users to a page that could otherwise be reached directly, without the individual pages offering substantively different or additional value for the searcher. Common examples Google gives include multiple pages targeting city/region name variations with otherwise near-identical content, and pages built primarily to capture search traffic and redirect or funnel it elsewhere rather than to serve the specific query well on their own. A set of geo-modifier pages that differ mainly by a swapped city name, with the same service description, same structure, and minimal location-specific substance (no genuinely distinct local information, local case studies, local pricing/availability, or other content that only makes sense for that specific place) is the canonical shape of what this policy targets.
The diagnostic signals above are meaningful precisely because they’re consistent with how Google’s algorithmic systems would behave if they’ve assessed a page set against these criteria and decided to suppress or consolidate it, even without an explicit manual action ever being issued. Manual actions (visible in GSC’s Manual Actions report) tend to arrive after a pattern is severe or persistent enough to warrant human review; the algorithmic version of this suppression can and often does happen well before that point, silently.
A hypothetical illustration
As a hypothetical illustration: imagine a hypothetical HVAC company called Acme Heating and Air that has published 40 city-specific landing pages, one for each town in its service area, each following the same template: a short intro swapping in the city name, a generic service list identical across all 40 pages, and a call-to-action. Suppose that six months after publishing, Acme’s team filters GSC’s Performance report by the /service-area/ URL path and notices impressions across the entire set declining together over the same eight-week window, rather than a few individual pages losing ground independently.
Checking the Page Indexing report, hypothetically, Acme finds that 28 of the 40 URLs now show “Duplicate, Google chose different canonical than user,” with Google consolidating them toward the three or four city pages that happen to have the most distinct content. Cross-referencing impressions, only those same three or four pages are receiving meaningful search visibility; the other 36 are indexed but essentially dormant. In this scenario, Acme would be looking at the cluster-wide decline, canonical override, and impression concentration pattern described above, converging evidence that Google’s algorithmic systems have already started treating the templated city pages as low-differentiation, well before any manual action would show up in the Manual Actions report.
What to do with this diagnosis
If the pattern above shows up, the corrective path is to genuinely differentiate the geo pages that are worth keeping (adding substantive location-specific content: local service availability, area-specific pricing or considerations, genuinely local proof points) and consolidate or noindex/redirect the pages that can’t be meaningfully differentiated, funneling that intent to a stronger hub or regional page instead. Treat the GSC signals described here (canonical override, impression concentration, cluster-wide decline) as an ongoing monitoring routine for any geo-modifier page strategy, not a one-time check, since algorithmic reassessment of a page pattern can continue evolving well before, or entirely without, a formal manual action ever being issued.
