Every roofer who’s looked at a competitor’s site has seen the pattern: forty pages, one per suburb, each one identical except the town name swapped through it like a mail merge. “Roof repair in Chandler.” “Roof repair in Gilbert.” Same three paragraphs, same stock photo, same everything. And every roofer has wondered some version of: does that garbage work? The honest answer is that it used to, it mostly doesn’t now, and Google has a name for it, doorway pages, with a policy against them. But the conclusion roofers draw, “town pages are dead,” is wrong in the other direction. Town pages built on real local evidence remain one of the strongest plays in roofing search. The line between the two is specific, and it’s worth knowing exactly where it sits.
Could the page have been written about any town, by someone who’s never been there? If yes, it’s a doorway page regardless of how well it’s written. The swap test is the audit: replace the town name with a neighboring town’s name and reread it. If the page is still true, it was never about that town.
What makes a page survive the swap test is evidence that only exists because you’ve worked there. And a roofing company that’s been operating for even a few years has more of this evidence than almost any other trade; it just never gets off the job site and onto the site. You know which decades the subdivisions went up and what was nailed on then. You know the west side of town got hammered in the 2023 hail event and the east side didn’t. You know the HOA in that one development requires specific shingle colors, which streets have the worst tree exposure, what the permit office is like, and what you replaced on which street last spring. That’s the page. Nobody who hasn’t worked the town can write it, which is precisely why it ranks.
A town page that earns its slot runs something like this. It opens by answering the searcher’s implicit question, do you actually work here, with specifics in the first hundred words: how long, roughly how many roofs, response time from your shop. Then the local roof stock: “most of [town] was built between 1994 and 2008, which means three-tab shingles reaching end of life now; here’s what we usually find on those roofs.” Then weather history as you’ve experienced it professionally, the hail events, the monsoon pattern, what claims looked like. Then jobs: two or three real ones in that town, what was wrong, what it took, ideally with photos taken there and named streets or neighborhoods where the homeowner’s fine with it. Then reviews filtered to that town, pulled as quotes. And the boring blocks that do quiet work: which HOAs and permit quirks apply, and a short FAQ answering what people in that specific town ask you.
A page like that runs 800 to 1,200 words and takes a couple of hours with the right person in the room, the estimator who runs that territory, not a copywriter alone with a template. It also feeds the AI layer the exact material it needs: when someone in that town asks an assistant “how much hail damage does it take to total a roof,” the engine wants a source that demonstrably knows that town’s roofs. You wrote the only candidate.
The mail-merge sites publish forty pages because forty is easy when they’re fake. Real pages force a budget, and the budget forces the right question: which towns? Rank them by revenue, not population. Your best town is where your margins are best, your drive time shortest and your referral density highest, which is often not the biggest suburb on the map. Build five pages for the five towns that matter, one per week or two, before even thinking about the long tail. Five real pages outperform forty swapped ones, and they keep working, refresh each one after storm season with what actually happened there and the page compounds like the review profile does.
The discipline is resisting page forty-one. When the evidence for a town runs out, when you’re reaching for “conveniently located near,” the honest move is not to publish. A thin page doesn’t just fail to rank; sitewide, a pile of them teaches Google your domain is the mail-merge kind.
If you want help ranking your towns by search demand against your actual job history, that’s a thirty minute exercise: book here. NetPageTwo builds and maintains this inside the flat service, $489 a month for technical, on-page and AI search work, $1,297 with conversion and landing pages, month to month, fourteen days notice. More on the vertical: what we ship for roofing companies.
