Serious Programmatic SEO: How We Scaled a 8-Page Site to 175 Pages in Two Months

Two months ago, this very site had 8 indexable pages. Today, it has 175—all with unique content, proprietary data, and rankings for specific long-tail keywords. This isn’t just scaling for the sake of scale; it’s serious programmatic SEO, and the results speak for themselves.

But before we dive into the how, let’s clear up the confusion: programmatic SEO isn’t about spinning up 500 near-identical pages with city names swapped in. That’s thin content, and Google’s been catching that for over a decade. These pages end up in the dreaded "Crawled – currently not indexed" limbo—never to rank.

So, what is programmatic SEO?

What Programmatic SEO Actually Is (And What It Isn’t)

Programmatic SEO generates hundreds (or thousands) of indexable pages by combining data dimensions. The classic example? Airbnb. One page for every city, neighborhood, and property type. This creates thousands of pages that rank for hyper-specific long-tail queries like "apartments in Palermo" or "house rentals in Bariloche with a pool."

What isn’t programmatic SEO:

• A template with a city name swapped out 500 times.
• AI-generated pages with a paragraph of fluff and three FAQs.
• A site that feels like a glorified spreadsheet.

If your "pages" could be swapped out without anyone noticing, you’re doing it wrong. Google’s algorithms have evolved far beyond that.

When to Use (and Skip) Programmatic SEO

Programmatic SEO is a high-impact, high-effort strategy—not a silver bullet. It makes sense only if:

1. Your product or service naturally has many variants (cities, models, brands, professions, or categories).
2. Each combination has real search demand (not fabricated long-tail terms—verify this in Search Console or keyword tools).
3. You can generate genuinely unique content for each combination—no copy-pasted paragraphs.

It’s a terrible fit if:

• You sell a single product or service.
• Your market is hyper-local (e.g., one city or province).
• You can’t source unique, verifiable data for each page.

In those cases, a well-structured blog with high-quality editorial content is a far better investment.

The N+M = N*M Pattern: How to Scale Without Thin Content

The secret sauce of efficient programmatic SEO? Generate content by dimension, then assemble it at build time.

For this site, we defined two core dimensions:

• Locations: 12 (CABA, Buenos Aires, La Plata, Córdoba, Rosario, Mendoza, Mar del Plata, Tucumán, Mexico, Chile, Colombia, Uruguay).
• Services: 9 (B2B SaaS, CRM, ERP, n8n automation, Cloud & DevOps, SEO, Legacy Modernization, QA, Branding).

If we wrote each combination manually, that’s 12 × 9 = 108 pages. But with the right pattern, we only needed 12 + 9 = 21 base contents, combined dynamically at build time. Each final page includes:

• A location-specific paragraph.
• A service-specific paragraph.
• Locally relevant data.
• Combined FAQs.
• Strategic internal linking.

A single prompt to an LLM (Groq or Gemini) generates all 108 pages. That’s the multiplier effect that makes programmatic SEO economically viable.

The Layer That Separates Real Programmatic SEO from Thin Content

This is the part most tutorials gloss over—and why so many fail. A page generated by AI with a generic paragraph and three FAQs won’t index. Google flags it as synthetic content and buries it.

To rank, each page needs at least one unique, verifiable data point. For example, the software development page in Córdoba includes:

Córdoba is Argentina’s second-largest IT hub, with approximately 14,000 formal software jobs. Source: OPSSI – CESSI, 2024.

This data is:

• Unique to the page (not duplicated anywhere else on the site).
• Verifiable (sourced from a reputable organization).
• Location-specific (changes per city/region).

The same logic applies across industries:

• Insurance: DNRPA vehicle registrations by model and province.
• Real estate: Property registry data by district.
• Healthcare: Hospital bed availability by region.

Without this layer, your pages are just templates—no matter how well-written they are.

Case Study: Mi Seguro de Auto (My Car Insurance)

Before Pair Programming, we applied this at Mi Seguro de Auto, a site that combines motorcycle models (Honda Wave 110, Honda Tornado, etc.) with cities. Each /insurance/{city}/{model} page includes:

• Local cost of insurance (aggregated from real quotes).
• Number of motorcycles of that model registered in the province (DNRPA-verified data).
• Common coverage options in the area.
• Model- and location-specific FAQs.

This content is impossible to duplicate elsewhere. It builds domain authority, and Google treats it as a source, not a clone of 500 identical pages.

The Technical Pipeline: Three Layers for Scalability

To make this work at scale, you need three technical layers—most tutorials miss at least one:

1. Layer 1: LLM-Generated Content Base paragraphs and FAQs for each dimension. Normalized, rate-limited, and cached. Run once, reuse forever.

2. Layer 2: Real-World Data Static JSON files from official sources (updated every 3–6 months). This is the layer 90% of guides ignore. Without it, your pages are just AI fluff.

3. Layer 3: Structured Schema.org Markup FAQPage, BreadcrumbList, Service, FinancialProduct—whatever applies. This tells Google exactly what each entity is on the page.

Skip any of these, and your pages won’t index. Do all three, and you’ve got programmatic SEO that ranks.

The Feedback Loop: Optimize, Iterate, Scale

Deploying the pages isn’t the end—it’s the beginning. The workflow:

• Weekly Search Console audits Track indexed pages, queries, and impressions. Identify low-hanging fruit: URLs ranking in positions 5–15 with high impressions but low clicks.

• Content optimization sprints Tweak titles, meta descriptions, and FAQs to push those pages into the top 3.

• Iterate based on data Use a custom pipeline (we built ours at Pair Programming) that combines Search Console data with multiple LLMs debating improvements. We started with Mi Seguro de Auto and are now rolling it out across other projects.

The Next Frontier: GEO (Generative Engine Optimization)

Programmatic SEO alone won’t cut it in 2025. The next step? Optimizing for AI-generated answers.

In our Mi Seguro de Auto project, 40% of traffic now comes from AI engines like ChatGPT and Perplexity. To appear as a cited source, you need Generative Engine Optimization (GEO).

GEO goes beyond traditional SEO:

• Structured data that AI can parse.
• Content that answers questions concisely (the "zero-click" format).
• Citations from authoritative sources.

If you want to dive deeper, check out our guide: What Is GEO and How to Get Cited in ChatGPT.

Should You Try Programmatic SEO?

If your business has multiple natural dimensions (cities, models, categories, professions) and you want to scale organic traffic without writing 200 articles by hand, programmatic SEO is worth exploring.

But it’s not for everyone. Do your homework first:

• Validate search demand for your target long-tail terms.
• Identify unique data sources for each page.
• Build a technical pipeline that combines LLMs, static data, and structured markup.

If this sounds like a fit, we’d love to help. Book a 30-minute call to see if your vertical is a good match and estimate potential page volume and traffic gains. Explore our programmatic SEO service or contact us directly.

---

By Esteban Aleart, Founder & Lead Engineer at Pair Programming. This article is a case study of a site using programmatic SEO on itself.