I built 14 million SEO pages for HonestDoor. Last week I built 250 for my own site. Here's what changed—and why it's not AI slop.
Programmatic SEO is when you turn a database into thousands (or millions) of search-optimized pages.
The formula: Data × Template = Pages
You have 10,000 products? That's 10,000 product pages. You have property data for 14 million addresses? That's 14 million pages indexed by Google.
Each page targets a specific search term. Someone searches "123 Main St Regina" and hits your page. Someone searches "Nike Air Max 90" and lands on your product page.
Database becomes search terms. Templates become traffic.
HonestDoor is a real estate platform. They have property data for 14 million addresses across Canada.
Each property gets a page. Each page is indexed by Google.
14 million pages = 14 million search terms = bulk of their organic traffic.
Someone searches for their home address. Google shows the HonestDoor page. They click. They see their property value estimate. They explore the neighborhood data.
This is traditional programmatic SEO at scale. Database-driven. Template-based. Works.
Everyone heard "AI can write blog posts" and started pumping out generic content.
The result: millions of pages that all say the same thing. No unique value. No original images. No interactive elements. Just templated text with keywords swapped in.
Google calls this spam. Users call it useless. I call it lazy.
Programmatic SEO isn't dead. Using AI to generate generic content is.
It's not using AI to write content. It's using AI at every level of the infrastructure.
AI helps source ideas for what pages to create.
Example: For my pattern library, I use Claude Code with a specialized prompt system (VIBE-CODE-MASTER.md) to suggest design patterns that non-technical founders actually search for. Not "database normalization" (nobody searches for that). Instead "what is brutalism" or "shopping cart drawer" - real search intent.
AI generates unique visual assets for each page.
Example: Each pattern in my library gets a unique 450×450px card. I use Claude Code to generate the React component, then Puppeteer to screenshot it. 250+ unique images. Not stock photos. Not AI-generated generic illustrations. Actual UI mockups that show what the pattern looks like.
AI builds interactive components that send engagement signals to Google.
Example: Every pattern page has a click-to-copy prompt. Users interact. Google sees time on site, bounce rate, engagement. This isn't static content—it's a tool people use. That's the difference between spam and value.
HighStakeHealth is a poker tournament database. Problem: tournament data is scattered across different sites.
Solution: AI-powered scrapers collect the data. Claude Code writes the scraper code. Puppeteer runs it. Database fills up.
Each tournament gets a page. But not just scraped data—proprietary analytics on top.
Field Softness ratings. ROI projections by skill tier. This is value-add. Not just regurgitated data.
Interactive elements: collapsible sections, tooltips, hover states. Signals to Google this isn't static spam.
This site you're on right now is programmatic SEO at work.
267 design pattern pages. Each one targets a specific search term: "what is brutalism," "shopping cart drawer," "product filters."
Built 250+ pages in 7 days using Claude Code and an epoch-based workflow.
Claude Code reads VIBE-CODE-MASTER.md and suggests 6 pattern ideas per epoch. Targets real search intent.
Generate 6 React components at once. Each one is a 450×450px card with educational content and click-to-copy prompts.
Puppeteer screenshots each card. Saves to /public. Each page gets a unique Open Graph image.
Short answer: Week 1. ROI that matters: 3-6 months.
People asking this question usually don't know much about SEO. They want to know when the investment pays off. Here's the truth with actual examples:
Traffic starts immediately. Super long-tail searches like "what is neobrutalism design" or "product filters sidebar."
20-30 patterns ranking. People finding the site. Low conversion—they're learning, not buying.
Compounding over years. Not my primary lead source (word of mouth + social is). But free organic traffic in a hyper-competitive space (SEO, web design).
Strategy: Backdoor approach. Can't compete for "web developer Regina" so I own "what is brutalism" instead. Builds authority over time.
Organic traffic instantly. People searching "furnace replacement Regina" finding the site. Job posting created within 7 days.
AI-generated furnace articles. Brand new domain. Less traffic than HashBuilds but higher intent—people need furnaces NOW.
Everything optimized for Google. Exact match domain. Local SEO. High-intent keywords. Pure Google search play.
Result: Week 1 conversions. Not vanity traffic—actual business inquiries. Different content type, different purpose, same AI infrastructure.
You'll see traffic. Long-tail searches. Low volume. Proves it's working.
ROI becomes clear. Pages rank higher. Authority builds. Conversions start (if high-intent).
Compounding effect. New pages rank faster. Domain authority grows. Free traffic compounds.
Both examples are brand new domains with purely AI-generated content. Both get traffic. Different strategies for different goals. Low-intent = long game. High-intent = immediate conversions. Choose based on your business model.
Not stock photos. Not reused images. Each page has custom-generated visuals that show the actual pattern/concept.
Click-to-copy prompts, filters, calculators, demo UIs. Users interact. Google sees engagement signals.
People searching "what is brutalism" find the answer + visual example + prompt to use it. That's value.
All pages match the HashBuilds aesthetic. Not templated content with keywords swapped in.
Epoch-based workflow. Review grid before publishing. Not bulk generation of 1000 pages overnight.
Users copy prompts. They share pages. They bookmark patterns. These are tools, not content.
Products, addresses, tournament results, job listings, recipes, tools, courses. If you have data, you have pages.
Not just scraped data. Add analytics, comparisons, calculations, recommendations. Make it more useful than the source.
Not "database normalization." Real searches: "how to price my product," "best CRM for dentists," "poker tournament ROI calculator."
Calculators, filters, comparison tools, copy buttons, demo UIs. Anything that makes users engage = signals to Google.
I've built this at three different scales: 14M pages, AI-powered scrapers, and 250 pages in a week. Let's talk about what makes sense for your business.
START A CONVERSATION →267 pages. Each one targets a specific search term. Each one solves a real problem. This is programmatic SEO done right.
VIEW PATTERNS →