$0 MRR, 3 users, months of building — here's what happens when you build a SaaS and forget to market it

Hey IH — I'm Brian, building RetentionCheck solo and bootstrapped. Going to be fully honest about where things are.

THE PATTERN I CAN'T SEEM TO BREAK
Over the last year I've started multiple projects. Every single time, the same thing happens: I get excited about an idea, start building, think "just one more feature," and that turns into 10 more features. I get overwhelmed, lose momentum, step away. Come back weeks later to a codebase I barely recognize.

I don't come from a traditional coding background, which makes it worse. Every feature takes longer than it should. Every bug feels like a wall. And instead of shipping something small and seeing if anyone cares, I keep trying to build the "complete" version first.

WHAT I BUILT THIS TIME
RetentionCheck lets you paste customer feedback — cancellation reasons, support tickets, survey responses — and get AI-powered churn analysis. It ranks churn drivers by severity, gives you a Churn Health Score (0-100 with a letter grade A through F), pulls direct customer quotes, and suggests specific fixes. No integrations, no setup — paste and go.
You can try it right now with no signup: https://retentioncheck.com/try

AND THEN I DID THE THING AGAIN
I over-built it. Massively. Here's what I built before I had a single user:

5 free calculators (churn rate, LTV, quick ratio, revenue recovery, cohort retention)
125+ SEO pages
12 automated email templates
7 cron jobs running on empty databases
PDF export, trend tracking, insight annotations
A public roadmap with voting
A changelog and live stats page
An MCP server published on npm so you can run churn analysis inside Claude Code or Cursor
Dark mode

That's a product built for 10,000 users. I have 3.

I stepped away for a while after burning out on it. Life happened. The product just sat there, running on $18/month of infrastructure, serving basically nobody.

WHAT'S DIFFERENT THIS TIME
I made myself a rule: zero new features until 50 users. No more hiding behind code. The product works. Nobody knows it exists. That's a distribution problem, not a product problem.

So now it's 100% distribution. Reddit, Twitter, cold DMs, this post. I'm shipping zero features and spending every hour talking to people instead. This is the hardest part for me — putting myself out there instead of hiding behind a code editor.

HONEST NUMBERS
Users: 3
MRR: $0
Pro subscribers: 0
Cost to run: ~$18/month (Vercel + Supabase + Claude API)
Time building: months
Time marketing: almost none
Those numbers are embarrassing. But I'd rather share them than pretend I have traction.

WHAT I'VE LEARNED FROM THE DATA (THE ACTUALLY USEFUL PART)
Even with barely any users, I've analyzed enough cancellation data to spot clear patterns:

"Price" is almost never the real reason. 30%+ of cancellations mention price, but when you read the full response, they're saying "I didn't get enough value." Completely different problem, completely different fix.

Most churn happens in the first 30 days. It's almost always an onboarding problem disguised as something else.

15-20% of cancellations give zero useful feedback. Those "quiet quitters" were already gone weeks before they clicked cancel.

Most founders have never read their cancellation responses. Hundreds sitting in Stripe or Typeform that nobody checks.

PRICING
Free ($0) — 10 analyses/month, CSV/JSON export
Pro ($29/mo) — Unlimited, PDF export, trends, share links
Lifetime ($399 once) — Everything in Pro, forever (capped at 100 seats)

WHAT I'D GENUINELY LOVE FEEDBACK ON

Has anyone else been stuck in the "build too much, market too little" trap? How'd you break out?

Is the paste-and-go approach useful, or is it too simple without integrations?

Would you actually pay $29/mo for this, or is churn analysis a "do it once" thing?

If anyone wants me to analyze their cancellation data — drop it below or DM me if it's sensitive. First 5 people, I'll run the analysis and walk through what I find. No pitch, just want to show what it actually does with real data.
https://retentioncheck.com