The thin-content line — when “pages at scale” becomes doorway pages.

“Pages at scale” lives a single step away from “doorway pages,” and a lot of sites have taken that step without noticing. The technique — one template, many rows — is the same on both sides. What changes is whether each page carries real value or just exists to catch a search and shove the visitor somewhere else. Google has policies for the second kind, and they’re not new: doorway pages have been against the rules for years, and in 2024 the spam policies added “scaled content abuse” to name the modern version of it directly. If you’re going to build at scale, you need to know exactly where that line sits — because the penalty for crossing it isn’t just “those pages don’t rank.” It’s worse than that.

Google’s policies, in plain English

Two policies matter here, and it’s worth knowing what they actually say rather than the folklore version.

Doorway pages. A long-standing one. A doorway page is a page (or a set of them) created to rank for specific searches and then funnel the visitor to some other destination — the “real” page, the booking form, the call. Classic examples: dozens of near-identical city pages that all push to one contact form, or pages targeting slight keyword variations that have no purpose except to be a landing spot. The tell is that the page isn’t an end in itself; it’s a turnstile. Google’s stance is that pages should be useful where the visitor lands, not waystations to somewhere useful.

Scaled content abuse. Added to the spam policies in 2024, and it’s the one most relevant to programmatic SEO done badly. Paraphrasing the substance: mass-producing pages primarily to manipulate search rankings — rather than to help people — is against the rules, regardless of how the pages are produced. By hand, by template, by generative AI, by some mix — the production method isn’t the issue. The issue is the purpose: were these pages made because there was something worth saying, or were they made because more URLs seemed like more rankings? This sits alongside Google’s “people-first content” guidance — the test of “would this page be useful if search engines didn’t exist?” If a programmatic set fails that test, it fails the policy, and it doesn’t matter how cleverly it was generated. (The overlap with AI-produced pages specifically is covered in what Google actually says about AI content — short version: same rule, the tool is irrelevant, the value is everything.)

The guardrails — what every cell has to clear

Before a page goes on the build list, it has to pass all of these. Not most. All.

• It satisfies a real search someone actually does. There’s genuine demand for it — people type this. A page targeting a search that doesn’t exist isn’t a page; it’s a permutation. Google will leave it un-indexed, and rightly. (What that looks like in practice: why aren’t my programmatic pages ranking.)
• It carries substantial unique value, not just a swapped noun. The variable layer has real, specific content true of this row and not the next one — local facts, real data, a genuine angle. If the only difference between this page and its sibling is a proper noun, you’ve built one page and published it twice. (The two-layer model: programmatic page templates.)
• It isn’t a near-duplicate of its siblings. Related to the above, but worth stating separately: a set of pages that are 90% identical is a content farm, full stop. The boilerplate can be shared; the substance can’t.
• It doesn’t exist purely to funnel to one destination. The page has to be worth landing on. It can absolutely have a CTA — every page here does — but its reason for existing can’t be “be a doorway to the contact form.” If you deleted the CTA and the page had no other purpose, it’s a doorway page.
• A person could read it and learn something they came for. The summary test. Read the page as the searcher. Did you get the thing you searched for? If yes, it clears the bar. If you got a stub with a noun in it and a button, it doesn’t.

Build the cells that pass. Noindex or skip the ones that don’t. A page count is not an achievement — a set of pages that each earn their place is.

“We eat our own cooking” — why we skip cells

This isn’t theoretical for us. We’re building a Tampa-Bay-first geo matrix the way this guide describes, and the discipline is exactly the one above: we only build the {vertical} × {city} cells where there’s a genuine local angle — a real service offered there, real demand, something specific and true to say. Where a cell doesn’t clear the bar — we don’t actually serve that area, nobody searches that combination, there’s no honest substance to put on the page — it doesn’t get made. Noindex if it’s already live and shouldn’t be; skip it if it isn’t. That’s not us being cautious for show; it’s that the alternative actively hurts. A handful of thin cells doesn’t just sit there harmlessly — it pulls down the pages around it. Same logic applies whether it’s our geo matrix, a client’s service-area set (service-area pages), or any cluster: the count is whatever number of cells genuinely clear the bar, and not one more. The programmatic SEO service is built on that rule.

How thin programmatic pages backfire

The mistake people make is thinking the worst case is “those pages don’t rank.” If that were the worst case, thin pages would be a harmless gamble — build a hundred, some stick, no downside. That’s not how it works.

Google forms a view of your site as a whole. A site that’s mostly thin, near-duplicate, template-heavy pages reads — to the algorithm — as a low-quality site, and that assessment doesn’t politely stay confined to the bad pages. It dilutes the quality signal sitewide. Your genuinely good pages — the pillar, the real service pages, the case studies — rank worse than they would have if the thin pages didn’t exist. You’ve taken a site that was small but solid and made it large but compromised. That’s the actual risk: not that the experiment fails, but that it drags down the parts that were working. It’s why “build them all and see what sticks” is a bad strategy and “build only the ones that clear the bar” is the only one — and why the move when you find thin cells is to noindex or delete them, not to leave them lying around. (More on this in is more pages always better for SEO — the same principle, applied to topical depth rather than programmatic scale.)