“The decision that hurt us most was overbuilding the backend before users showed us what they actually needed. For MVPs, I now ship the simplest thing that can validate the core loop, then let real feedback drive the stack.”
“I shipped Git City in a day on Next.js, Supabase, Vercel, and Stripe, and I have not regretted those choices because they are easy to swap out once you actually have users. The painful reversals are almost never the framework: they are the data model you locked yourself into, the auth provider that owns your users, and the payment rails you cannot migrate without asking customers to re-enter cards.”
“The hardest call to unwind is the one you make because it sounds impressive. At Hiatus I built the MVP in Swift and Node on AWS because I already knew them, not because they were perfect, and that let us ship fast enough to learn what actually mattered.”
“The stack itself is usually reversible. The commitments that harden around it are not: the team you hired because they knew that language, the integrations you promised customers, the dashboards and reporting cadences that make switching feel like a turnaround.”
“The hard ones to reverse are rarely the framework or language. I’ve swapped those without drama. What hurts is the shape underneath: a schema you baked too early, a multi-service split that forces coordination before you have product clarity, or a deployment path that only works on localhost. With my lyrics translator, I kept it as one Next.js app until Genius started blocking server-side fetches in production.”
“The stack itself is rarely the thing that kills an MVP; it’s the shape underneath it. At Scalio, I watched teams ship fast on whatever was familiar, then spend six months untangling a data model or a single overloaded API contract that had calcified into everything.”
“The hardest thing to unwind is rarely the language or framework; it is the data model and the boundaries you draw around integrations. At SiteOwl we picked tools that were easy to hire for, but the real cost came later when a vendor's pricing and roadmap locked us into shapes that did not fit the business.”
“I have built MVPs in health tech and seen which bets lock you in. The painful ones are almost never the frontend framework or the database. They are the integrations you chose, the data model you baked in, and the compliance posture you assumed you could add later.”