Build-Measure-Learn

The loop of shipping something small, measuring what happens and deciding what's next.

Also known as: build measure learn, BML loop, lean loop

Build-Measure-Learn is the lean startup loop: build the smallest thing that tests a belief, measure what real users do with it, learn whether the belief held — then loop. Each cycle should change what you believe or what you build next; a loop that spins without changing decisions is motion mistaken for progress.

believe → build (minimum) → measure (behavior) → learn (persevere / pivot / adjust)

Speed matters, but the loop’s quality sets the learning rate. Fast cycles testing trivial beliefs teach trivia; slow cycles testing the riskiest assumption teach survival. One honest loop per week beats ten cosmetic ones.

The classic mistakes:

Building too much per loop. A month-long “minimum” product tests a month of compounded assumptions at once — when it fails, you learn nothing about which belief broke. Shrink the build until one loop tests one belief.
Measuring vanity metrics. Signups and pageviews that rise regardless of product quality. Measure behavior that costs the user something: payment, retention, referrals.
Skipping the learn step. Teams ship, glance at a dashboard, and build the next thing. Schedule the decision explicitly: what did we believe, what happened, what changes?
Looping forever on a dead idea. Fifty loops with no pull is data, not bad luck. Set kill criteria in advance or the loop becomes a way to avoid deciding (see pivot).

Run it weekly: one belief, one minimum test, one measured behavior, one written conclusion. Paired with fast iteration speed, this loop is the whole pre-fit game.