Notion, 2015: Nuke It and Restart?

You are the co-founder of an early-stage productivity startup, two years post-founding, $2M raised, four engineers including the two co-founders. You've shipped a v1 product — a tool that lets users combine notes, tasks, and databases on a single canvas. It has a few hundred power users who love it, but daily-active users are flat and signups have been steady at ~30/day for nine months.

Your fundraise runway is six months. The product has 47 features. Half of them are used by under 5% of users. The codebase is a React-on-a-custom-renderer experiment that nobody else on the team fully understands except your CTO. Two power users keep emailing about a specific bug; you can't fix it without rewriting the renderer.

Your CTO walks into your one-on-one and says: "I think we should throw it all away and rebuild from scratch."

The principle: When you're 18 months in with no PMF and the product feels wrong, the courageous move is to nuke what you have and rebuild on first principles — but only if you can name what was wrong in one sentence. Pivots that can't be articulated are just thrash.