Concept
Minimum viable product (MVP)
A minimum viable product is the smallest experiment that lets you complete one full Build-Measure-Learn loop with real customers against your riskiest assumption; it is defined by what it teaches, not by how much product it contains.
The term is the most misread idea in The Lean Startup because founders hear 'minimum' and build a cheaper version of the full product, with fewer features and worse design, which tests nothing in particular. Ries's MVP is chosen backwards from the learning: which belief must be true for the company to exist, and what is the least work that would put that belief in front of a customer? That is why the famous examples look nothing like products. Dropbox tested demand for seamless syncing with a three-minute video and a waiting list. Zappos tested whether people would buy shoes online by photographing shoes in local shops and buying them at retail when an order came in. A concierge MVP delivers the service by hand to a handful of customers, and every manual delivery doubles as a customer conversation. The embarrassment of shipping something so small is, Ries says, the price of learning early rather than late.
A real-life example
A team in Eskişehir believed university students would pay for laundry pick-up. Their MVP was a Google Form, a WhatsApp number and a friend's car. Forty orders in two weeks proved the demand; the complaints about timing told them what the app, built two months later, had to get right. The app they had originally planned had a map, ratings and a loyalty scheme, none of which any customer had asked for.
How to use it
- 1Name the riskiest assumption first, then ask what the smallest thing is that puts it in front of a real customer: a page, a pre-order, a video, a service done by hand.
- 2Deliver the service manually to your first ten customers before you automate anything; the manual version teaches faster than code.
- 3Build half a product properly rather than a whole product badly, and leave the rest out rather than half-done.
