In 30 seconds
- An MVP tests a key assumption with the least effort.
- Its main output is learning, not revenue or polish.
- It can be a landing page or a manual service, not only software.
What an MVP is for
Every new product rests on a risky assumption, such as "people will pay for this". An MVP is the cheapest experiment that tests that assumption with real users. If the assumption fails, you've saved months of building the wrong thing.
Forms an MVP can take
- Landing page: describe the product and measure sign-ups.
- Concierge: deliver the service manually to a few customers.
- Wizard of Oz: a real-looking front end with humans doing the work behind it.
- Single-feature product: only the one capability that matters most.
Deciding what goes in
Write down the riskiest assumption, then the evidence that would prove it wrong. Build only what is needed to collect that evidence. Everything else goes in the Product Backlog for later.
Where the term comes from
The phrase was coined around 2001 and popularized by Eric Ries in The Lean Startup (2011), where it sits inside a build-measure-learn loop.
Myth vs reality
The AI angle
AI makes MVPs much cheaper to build. The constraint shifts to learning: a cheap MVP is only valuable if real users see it and you measure what they do.
FAQ
How is an MVP different from a prototype?
A prototype tests how something works or looks; an MVP tests whether real users want it.
How long should an MVP take?
Weeks, not months. If it takes longer, the scope is too big.
What comes after an MVP?
Either you pivot based on what you learned, or you build the next increment toward a real product.