What I learned shutting down a startup

The sequence of events

"Caspian was an analytics platform for e-commerce teams. We had 40 paying customers, a product that worked, and a Series A process that was going well until it wasn't. The round didn't close. We had three months of runway. We told the customers, processed the refunds, and shut it down. I'd been inside companies that failed before. But I'd never been there at the exact moment the decision was made. It's quieter than you'd expect."

What nobody tells you about building on a deadline

"When you have 18 months of runway you make decisions you wouldn't make with unlimited time. Some are correct. Some become visible as problems exactly when you have least capacity to fix them. We used React Context for state management. For the first year, fine. In the last six months we started spending real time tracking down re-render bugs that a more explicit state solution would have prevented. I'd reach for Zustand earlier. The counterpoint: React Context got us to 40 customers. The perfect architecture you ship in month 12 doesn't help the customer you lose in month 6."

On documentation

"We had almost none. This felt reasonable — we were two engineers who talked constantly. It became a problem when we needed to hand things off cleanly during shutdown. Write the README before you think you need it."

What I'd tell someone joining a seed-stage startup

"Ship the thing. Technical debt from moving fast is almost always recoverable. Customers lost by moving slowly are not. Know which parts of the codebase are load-bearing and write it down. Build the architecture you need for the next six months, not the next five years."

The quieter thing

"Caspian didn't fail because of anything I built. The product worked. The customers liked it. The fundraising environment was what it was. That's both easier and harder to sit with than a technical failure would be."

"Elias Ward is a full-stack developer based in Amsterdam. Currently open to work."