Qu'est-ce que MVP (Minimum Viable Product) ?

The MVP concept, popularized by Eric Ries in The Lean Startup, is built on the principle that the biggest risk in product development is building something nobody wants. An MVP is not a half-finished product — it is a deliberately scoped product that tests the riskiest assumption at the heart of the business: will people use this, and will it solve their problem in a way they value? The scope of the MVP is determined by working backward from that core hypothesis, stripping away every feature that doesn't directly test it. This focus is counterintuitive to most founders, who want to ship the full vision.

MVPs exist on a spectrum of fidelity. At the low end: landing pages (testing whether people will sign up before you build anything), concierge MVPs (manually delivering the service that software would eventually automate), and Wizard-of-Oz MVPs (simulating automated functionality with human work behind the curtain). At the higher end: functional software that delivers core value but is missing secondary features, integrations, and polish. The right MVP fidelity depends on what you're testing and what evidence would convince you to proceed or pivot.

The output of an MVP phase is not just a product — it is validated learning. Key metrics to observe include activation rate (do users do the core action?), retention (do they come back?), and qualitative feedback (what is confusing, what is missing, what do they love?). A successful MVP produces enough evidence to make an informed decision about the next stage of development, rather than guessing what to build next.