For a stretch of years, if you searched Google for a T1 line in almost any American city ("Denver T1 line," "Tampa T1 line," pick one), the top three organic results were ours. Not one result on the page. All three. Broadband Locators, Broadband Trader, and ISP Locators: three separate directory sites we built and ran, holding positions one, two, and three on the same local results page, city after city.
Nobody gets that lucky. That is what programmatic SEO looks like when the structure is right: thousands of pages, each answering one specific local question, compounding into a wall of rankings that no hand-written page, and no competitor, could match one at a time.
What programmatic SEO actually is
Programmatic SEO means generating pages from structured data instead of writing them one at a time. You build one excellent template (a city page, a provider page, a service-by-market page) and populate it from a dataset, producing hundreds or thousands of pages that each target a query real people actually type.
The directory is the natural home for this, and it is where we have done it at scale. Our business-internet directories have run thousands of pages between them, and the flagship of that family (since sold, which is its own kind of proof) ran more than 2,500 pages. Broadband Trader runs more than 1,700. ISP Locators, more than 1,000. Over five thousand pages between them, nearly all generated from structured data about providers, services, and places, and all of it dependency-light, hand-built HTML with no CMS in the way to slow it down or break it.
The math is the whole appeal. A ten-page site can rank for a handful of terms. A 2,500-page site can rank for thousands of long-tail local queries, each one small on its own and collectively a business. "Business internet in a mid-size city" is not a keyword anyone brags about winning. Winning it in every mid-size city at once is a different story.
Why most programmatic sites fail
Google has spent two decades punishing thin, auto-generated pages, and rightly so. Most programmatic projects die the same death: someone spins up ten thousand pages that all say nothing, each one a paragraph of filler wrapped around a swapped-in city name. That is a doorway-page factory, and sooner or later it gets treated like one.
The difference between a doorway page and a directory page is whether the page is useful to the human who lands on it. Our city pages worked because each one answered the actual question: which providers serve this specific market, with which services. Real data, locally distinct, genuinely worth loading. The template was the delivery mechanism, not the product.
Before you generate anything, every page in the plan needs four things:
- Real, page-specific data. If two city pages differ only by the city name in the headline, you do not have two pages; you have one page and a liability.
- A query with intent behind it. Generate pages for questions people actually ask, not for every permutation your database can produce.
- Enough substance to stand alone. Every page should be able to justify its own existence in the index without leaning on its siblings.
- An architecture that connects them. Hubs, categories, breadcrumbs, internal links. Structure beats content volume, and at five thousand pages it is the only thing holding the site together.
What the 1-2-3 result actually teaches
The positions 1-2-3 outcome was not three sites getting lucky at once. It was one structural insight applied three different ways. Each site had its own angle on the market (a marketplace, a trading exchange, a locator), its own data emphasis, and its own template. Each earned its ranking independently. Google saw three genuinely different, genuinely useful results that happened to share an owner.
The lesson is not "build three sites." The lesson is that structure was the moat. Anyone could write one good page about business internet in Denver; plenty of people did. Nobody else had built the system that produced a good page for every city at once and kept the whole thing crawlable, internally linked, and fast. Pages compete one at a time. Systems compete everywhere at once. That is also why the advantage persisted: a competitor would have needed to out-build the system, not out-write a page, and building systems is slower and rarer than writing pages.
Running the playbook in 2026
The playbook still works, and AI has made the expensive parts cheaper. Assembling datasets, drafting copy variants, QA-ing thousands of pages: work that once needed a small team now needs one disciplined operator with good tooling. That cuts both ways: your competitors can generate junk faster than ever, which makes genuinely useful data the differentiator it always was. There is also a new payoff: AI assistants answering local questions retrieve and cite exactly this kind of structured, specific page.
Start smaller than your ambition. One template, a few hundred pages, a dataset you trust. Prove the pages index, rank, and convert before you scale, and treat information architecture as the first decision rather than the last. If local queries are the target, the city-by-city side of this has its own rules; we laid them out in our local SEO playbook.
We build these systems for ourselves first, with our own money on the line, which keeps the advice honest. If you want one built for your market, that is the core of our product practice.