Back to Blog
January 25, 20263 min read

How I Generated 70k High-Quality Pages in 48 Hours with pSEO & MCP

Harris Park
Harris ParkAuthor

SEO Analyst

How I Generated 70k High-Quality Pages in 48 Hours with pSEO & MCP

I'm a SaaS founder, not an SEO expert. Honestly, I had no idea how to run Google Ads, what to write for a blog, or even what "SEO" really meant in practice.

I just wanted users.

So instead of hiring an agency, I decided to solve it with code. I built an SEO MCP Server that connects my real data (Google Search Console + Ads) directly to my coding environment (Claude Desktop / Cursor / Antigravity).

The results? I generated 70,000+ programmatic SEO pages in 2 days. (Note: Indexing takes time, but the infrastructure is done).

Here is the exact "Vibe Coding" workflow I use. You can copy this.

The Workflow

1. Find the "Golden Keywords" (MCP + GSC)

I don't guess keywords. I ask Claude to pull my real GSC data and Google Ads metrics via MCP. "Find keywords with low competition (CPC < $2) but decent volume." The AI analyzes thousands of rows instantly and gives me the winners.

2. Spy on the Top 10 (SERP Analysis)

Once I have a target keyword (e.g., "MCP for SaaS"), I don't just write blind. I use the analyze_serp tool to fetch the actual Top 1-10 search results. I feed these into Claude context so it knows exactly what Google likes right now.

3. Analyze & Strategize

I ask Claude: "Analyze the structure of these top 10 articles. What are they missing? What is the gap?" This creates a strategy that is mathematically clearer than just "writing a good post".

4. Write Content (Opus 4.5 is King)

This is crucial. Do not generate 500-word garbage. I instruct Claude (specifically Opus 4.5 model) to write 3,000 - 5,000 words based on the research.

  • Deep research
  • Proper H1/H2/H3 structure
  • Data-backed arguments

This becomes my "Source Article".

5. Programmatic SEO (The 70k Page Trick)

This is where the magic happens. I don't just publish one article. I turn that article into a Template. I ask Claude to extract the "Variables" (Target Audience, Pain Point, Industry) and convert them into a JSON dataset.

Tip: Keep it simple (1-2 dimensions).

  • Good: "SEO for Dentists", "SEO for Lawyers" (Target change)
  • Bad: Changing 5 different variables at once (Result: generic, low-quality spam)

6. The Prompt I Use

If you want to try this with Vibe Coding tools (Claude Code / Antigravity), use this prompt:

You are an expert pSEO Architect. Based on this source article, design a 1-2 dimensional pSEO architecture.
- Limit sitemap to 50k URLs per file (split automatically if larger)
- Extract 'Target Audience' and 'Pain Point' into a JSON dataset
- Design a folder tree that is easy to maintain and extend

7. Why MCP?

The Model Context Protocol (MCP) is the cheat code here. While writing the code/content, the AI can actively query live keyword volumes or check if a specific niche actually searches for this topic. It’s not hallucinating demand; it’s verifying it.


The Result

I now have a system that scales content indefinitely without sacrificing quality (because the Source Article is deeply researched).

Quality > Quantity always. But with this setup, you get both.

I’m sharing my tool (FireSEO) and this workflow because I think "Manual SEO" is dying. The future is Agents that know your data.

Start Analyzing Your Data Free →

FireSEOFireSEO

Break the Barriers to Explosive Growth.

Start a conversation with FireSEO Agent today and turn your website into a growth engine.

Start Free Analysis

No credit card required

FireSEOFireSEO
Agentic SEO Automation Demo
UserUserUserUser

1,200+ websites connected

Data Analysis for AI, Rank Growth for You.

Start Free Analysis

No credit card required • Start in 30s