Minimum viable product
A minimum viable product is the simplest version of an offering that can test a core hypothesis with real users.
The temptation is to build the whole vision before showing anyone. The minimum viable product is the discipline of resisting it.
A minimum viable product is the simplest version of an offering that lets a team test a core hypothesis with real users while spending as little time and money as possible. Its purpose is not to be a small, polished product but to generate valid learning about whether an assumption holds, before committing to building the full thing.
Learning, not launching
The word that does the work is viable, and it is widely misread. The minimum viable product is not the first release of the real product; it is an experiment. The question it answers is not does this work but is our belief about customers true. That reframing matters, because it changes what to build: the smallest thing that produces a trustworthy answer to the riskiest question, which is often far less than a working product, and sometimes not a product at all.
The two failure modes
These experiments fail in two opposite ways. Some teams build too much, polishing a full product before testing whether anyone wants it, and so learn the expensive lesson late. Others build too little, shipping something so threadbare that it tests nothing, and draw false conclusions from a fair rejection of an unfair sample. The right minimum is the smallest build that still yields a valid answer, which takes judgement, not just minimalism.
What it protects against
The discipline guards against the most common startup failure: spending months or years building something elaborate, in secret, only to discover at launch that the central assumption was wrong. By forcing an early, cheap test of that assumption, it converts a catastrophic late failure into a survivable early one.
A minimum viable product is best understood as a question asked in the cheapest possible form. Teams that treat it as a tiny product miss the point; teams that treat it as an experiment, designed to learn rather than to launch, get the value it was meant to deliver.