Worked 6 Months on a Perfect Side Project. Made $240.
Worked 6 Months on a Perfect Side Project. Made $240.
Six months of evenings and weekends. A beautiful UI with smooth animations. Comprehensive test coverage. Clean architecture with proper separation of concerns. Launched to crickets. Made $240 in the first three months. Meanwhile, a competitor shipped an ugly version in two weeks and made $5,000 in the same period.
Why Perfectionism Kills Side Projects
The economics of side projects are brutal. You have limited hours, no team, and no marketing budget. Every hour spent polishing the UI is an hour not spent on distribution. Every hour refactoring for "clean code" is an hour not spent talking to users.
The perfect side project fails because it optimizes for the wrong thing - code quality instead of market feedback. By the time it launches, the market has moved, your assumptions are stale, and you have invested too much to pivot.
The Ugly MVP Advantage
Ugly MVPs win because they compress the feedback loop:
- Week 1: Build the core feature with zero polish
- Week 2: Ship it and start getting feedback
- Week 3: Learn what users actually want (it is never what you assumed)
- Week 4: Build that instead
The polished approach takes six months to reach week 3. By then, you have sunk cost bias preventing you from pivoting.
What AI Agents Change
AI coding agents make ugly MVPs even faster. What used to take two weeks now takes two days. The entire value proposition of AI agents for side projects is not "better code" - it is "faster to market feedback."
Use the agent to:
- Generate a working prototype in a day
- Skip the refactoring phase entirely
- Build only the feature that tests your core hypothesis
- Throw it away and rebuild when you learn what users want
The $240 Lesson
The $240 was not a failure of the product. It was a failure of timing and distribution. The product was good - it just arrived six months late with no audience. If I had shipped the ugly version in week two and spent the remaining five months on marketing, the outcome would have been completely different.
Ship the ugly thing. Charge money. Learn. Then decide if it is worth making pretty.
Fazm is an open source macOS AI agent. Open source on GitHub.