From Idea to MVP: How to Launch a Software Product in 90 Days

Weeks 1–2: discovery that draws hard lines

Good discovery produces three artifacts: the one workflow the product must nail, the riskiest assumption to test, and a written not-doing list longer than the doing list. Every feature idea gets a question attached — what would we learn if users had this? If the answer is nothing, it waits.

Weeks 3–10: a shipping rhythm, not a build phase

Working software every single week, demoed to someone outside the team. The first weeks establish the walking skeleton — authentication, deployment pipeline, the core data model — so every later week compounds on production-grade rails rather than a prototype that must be rewritten.

• Week 3–4: walking skeleton live on real infrastructure
• Week 5–8: the core workflow, end to end, rough but real
• Week 9–10: the edges users hit first — onboarding, empty states, billing
• Continuous: automated tests on the paths that would embarrass you if broken

Technology that keeps options open

Choose boring, proven foundations — a mainstream framework, a managed database, one cloud provider — and spend your innovation budget on the product itself. The stack should let a two-person team deploy on Friday afternoon without fear. Clever infrastructure is a tax the MVP cannot afford.

Weeks 11–13: launch as measurement, not ceremony

Define activation before launch day: the action that proves a user got the value. Instrument it, watch real sessions, and interview the users who churned within minutes. The MVP succeeds when it returns a verdict — build more, pivot, or stop — backed by behavior instead of applause.