How I Built a Competitor Intelligence SaaS in 30 Days

The Problem

Every SaaS founder I know spends hours manually checking competitors' pricing pages, reading their changelogs, and stalking their job boards to understand where they're heading. I did the same thing — until I decided to build a tool to automate it.

TrackmyRival was born from a simple frustration: I needed competitor intelligence, and I didn't have time to do it manually.

The Architecture

The core of TrackmyRival is a crawler pipeline built with Firecrawl, which handles the heavy lifting of extracting structured data from competitor sites. The AI layer — powered by the Anthropic API — then interprets changes, extracts pricing signals, and generates human-readable summaries.

User adds competitor URL

→ Firecrawl scrapes the page

→ Claude extracts structured changes

→ Supabase stores the diff

→ User gets notified

Stack Choices

Next.js 14 with App Router for the frontend

Supabase for the database, auth, and storage

Firecrawl for web scraping that actually works

Anthropic API (Claude) for change detection and summarization

Stripe for credits-based billing

Resend for transactional emails

The QA Advantage

Here's what nobody tells you about shipping SaaS: your biggest risk isn't building the wrong thing. It's shipping a broken thing.

Coming from a QA background, I approached TrackmyRival differently:

1. I wrote integration tests before connecting the payment flow

2. I had a systematic test matrix for edge cases (competitor sites with paywalls, JavaScript-heavy SPAs, rate limits)

3. I tested failure modes — what happens when Firecrawl times out? When the Anthropic API returns an error?

This meant when I launched, I had zero critical bugs in the first week. That's not luck — that's QA discipline applied to indie hacking.

Lessons Learned

What worked:

Starting with the core crawler before building UI

Using credits-based billing (lower friction than subscriptions for early users)

Getting 10 users before writing a single line of marketing copy

What I'd do differently:

Nail down the ICP earlier (SaaS founders? Agencies? Enterprises?)

Launch a waitlist before finishing the product

Write more during the build — the blog posts you're reading came after the fact

The 30-Day Reality

The "30 days" headline is technically accurate but misleading. I had 8 prior years of engineering experience, existing Supabase/Next.js mental models, and I worked evenings and weekends.

What 30 days actually means: if you know the stack and have a clear idea, a focused solo engineer can ship a real product. Not a polished enterprise SaaS — but something people will pay for.

TrackmyRival is live. If you're building SaaS and want to stop monitoring competitors manually, check it out.