Minimum viable product and validated learning

In one paragraph

Rather than building the finished product and hoping, build the least you can that tests the riskiest assumption, and learn from what actually happens.

An MVP is not a cheap version of the product. It is the smallest thing that produces validated learning — evidence about whether a specific assumption holds. Sometimes it is not a product at all: a landing page, a manual service, a spreadsheet behind a form.

The loop is build, measure, learn, and it is meant to be run as fast as possible. The output of each cycle is a decision: persevere with the current approach, or pivot — change one fundamental element while keeping what has been learned.

The common misuse

'MVP' has come to mean 'the first version we shipped', which drains the idea of its meaning. If you did not know in advance which assumption you were testing, and what result would change your mind, you built a small product rather than an experiment.

Write down the assumption and the number that would falsify it *before* you build. If you cannot, you are not testing anything.

How to actually use it

• Take the idea you are most confident about and ask what evidence would prove it wrong. Build only enough to get that evidence.
• Do the manual version first. Serving ten customers by hand teaches you more than automating for a hundred who have not arrived.

Where it breaks down

• Some things cannot be shipped minimally — regulated products, safety-critical work, anything where a poor first impression is unrecoverable.
• Constant pivoting can be indistinguishable from having no strategy. The method needs a stopping rule as much as a starting one.

If you are being examined on it

Name the assumption being tested and how the result would be judged. An MVP with no hypothesis is not lean, it is just small.

More on enterprise & ownership