Concept
Validated learning
Validated learning is the unit of progress for a startup: a demonstrated, empirically grounded change in what you know about your customers, proven by an experiment rather than asserted in a meeting.
Eric Ries introduced the term in The Lean Startup to replace the wrong scoreboard. An established company measures progress in revenue and features because it already knows its customers. A startup exists precisely because it does not, so shipping a feature nobody uses is motion without progress. Validated learning counts only what an experiment with real customers proved: that they have the problem, that they will change their behaviour for the solution, that they will pay. The discipline is to state each belief as a hypothesis, design the cheapest test that could show it wrong, and let the result, not the opinion of the loudest person in the room, decide what is built next. It is the reason the road you are on runs on loops rather than plans.
A real-life example
Merve and Can spent four months building a meal-planning app in İzmir and launched to silence. In their second attempt they wrote their beliefs on one page, circled the riskiest one, that busy parents would pay for weekly plans, and tested it with a WhatsApp group and a manual plan for twenty families at 150 lira a month. Eleven paid. That number was the first real progress the company had ever made.
How to use it
- 1Rewrite your plan as a numbered list of beliefs about customers, and circle the one that kills the company if it is wrong.
- 2For that belief, design the smallest experiment with real customers whose result could embarrass you, and run it this week.
- 3At the end of each week write one sentence beginning with what you learned from a customer that changed what you will build next. If the sentence is empty, the week was not progress.