Loading...
The playbook Moonrank's SEO Agent uses to scale content without producing hundreds of near-identical pages that dilute the site instead of extending it.
Download the methodology, see the non-negotiable rules, or run it live inside the SEO Agent.
The Programmatic SEO Strategy skill is how Moonrank's SEO Agent turns structured data into organic surface. It designs at least five scalable page models — URL pattern, template, data source, target queries, competitive moat — then ranks them in a priority matrix and lays out the first 90 days week by week.
The rules are where the value sits, because programmatic SEO fails in predictable ways. Two pages from one model sharing more than 30% of their text is a failure, not a template. Every figure needs a datable source under three years old, and a figure that cannot be sourced is left as an explicit placeholder rather than invented. And every model must be fed by data the site already holds — without that, there is no strategy, only a wish.
It's written as a SKILL.md, a plain-language instruction format that people and AI agents can both follow. Read it to design your own programme, hand it to a developer as a spec, or let the SEO Agent design it around your data.
“We have a database of 4,000 products. Can we turn that into SEO?”
Not as one model — a product-per-page template on 4,000 rows is the thin-content failure exactly. Five models instead, and the strongest is comparison pages crossing your product attributes with the use cases in your support tickets: 340 pages, each with a genuinely different body because the ticket data differs per pairing. Your catalogue alone can't do that; the tickets are the proprietary half a competitor cannot copy.
Everything structured the site holds or can generate — this is the fuel, and without it there's no plan.
URL pattern, template, data source, target queries and the moat, per model.
A matrix of pages possible, effort, SEO impact, conversion and data readiness.
Week by week: which model, how many pages, and the technical prerequisites.
Each with its architecture, template, intent mapping and competitive moat.
Effort, impact, conversion and whether the data already exists.
Ten to fifteen long-tail queries each, with intent and estimated competition.
Templates define structure, never text — the line that separates scale from thin content.
Every figure datable and under three years old, or left as an explicit placeholder.
No two models targeting the same person with similar variables.