Nobody sets out to build doorway pages. They arrive as a location strategy, a landing page programme, or a scaled content project that seemed reasonable in the plan.
Where the line actually falls
The test is whether each page offers genuinely distinct value to the person landing on it. Sixty city pages with a real address, real coverage and real local detail are legitimate. Sixty pages identical except for the city name are doorway pages, whatever the internal justification was.
Scaled programmes drift across the line gradually
The first ten pages get real content because someone cared. By page forty the budget is gone and a template fills the gap. The programme was fine when approved and is a liability by the time it ships, and nobody made a decision to cross the line.
The uncomfortable audit
Take any twenty pages from a scaled set and read them side by side. If you cannot tell them apart without checking the location or product name, neither can a search engine, and neither can a customer.
Governing it before it happens
Set a rule at planning time: a page ships only if it can say something specific that no sibling page says. That constraint usually cuts a five-hundred-page plan to eighty, which is both the correct number and the one that will actually perform.