Doorway Pages

The distinction that matters is whether anything genuinely differs. A page for 'roofers in Phoenix' that would read identically for Tucson with the noun swapped is a doorway. A page carrying Phoenix's own measured data — which businesses AI named there, that market's openness score, its county's contractor density — is a report that happens to be programmatic.

The test Google's own documentation implies is about the visitor, not the generator: do these pages funnel users to the same destination without the intermediate page adding anything? A city page that exists only to rank and then push everyone to one generic contact form is a doorway however it was written. A city page that answers the question a local searcher had is not, however it was built.

A useful internal check is to diff siblings. Take two generated pages, strip the shared navigation and footer, and measure how much of the remaining text is common to both. If the answer is most of it, the pages are variants of one document and Google will treat them that way. We hold sibling city pages to a measured uniqueness floor and fail the build if the overlap creeps past it, because the alternative is discovering the problem months later in an indexation report.

The other half of the discipline is refusing to generate a page when there is nothing to say. Most doorway sets exist because the template could cover 300 cities and the data only supported 30. Unmeasured cities on this site return 404 rather than a template with a name swapped in — a rule that costs traffic on paper and prevents the entire set from being classified as scaled content.

That standard is what our city pages are built to, and programmatic SEO covers the legitimate version of the same technique.