6. Lean Startup

§ 6.1What it gave us#

§Eric Ries's 2011 book did for startups what accounting did for firms: it defined what progress is. Before The Lean Startup, a startup that shipped was making progress and one that did not was stuck. Ries proposed a different unit: validated learning, the amount you now know to be true about your customers and your business that you did not know before, demonstrated by evidence rather than argued from a plan. A team that ships a feature nobody uses has not progressed. A team that learns nobody wants the feature, before building it, has.

§Around that unit he built the loop everyone can recite. Build the smallest thing that can test a belief. Measure what real people do with it. Learn, and either persevere or pivot. The insight that matters most, and the one the original guide emphasized, is about the loop's speed: it should be as short as it can be while still teaching something. Not short for its own sake. Short because a startup's supply of loops is its runway divided by the length of one.

§Steve Blank, whose customer development work Ries built on, called it the thing that changes everything for a reason. It turned the question a startup asks itself from "what should we build" to "what do we need to learn, and what is the cheapest way to learn it."

§ 6.2What it assumes#

§Read the book again with 2026 eyes and one assumption runs through every page: building is expensive. The minimum viable product is minimal because a full product would cost too much to risk. The loop is short because each turn consumes engineers. The famous examples, Zappos photographing shoes in a store before stocking any, Dropbox's video, the concierge services that faked automation with people, are all ways of avoiding the cost of building. The method is, at its core, a discipline for spending scarce build capacity wisely.

§That was the right discipline for its decade.

§ 6.3Where it breaks at one to five, now#

§It does not break the way the others do. Its core, validated learning, is more true at a startup than anywhere, and more true now. What changes is the balance.

§When building was expensive, the minimum viable product protected a team from over-building. Now that a working prototype takes an afternoon, the risk has flipped: teams under-learn, because the loop's build step is so cheap that the measure and learn steps get skipped. A team can turn the loop ten times a week and learn nothing, because nothing was decided in advance about what each turn was meant to teach.

§The method's second quiet assumption was that the pivot is expensive and therefore rare. Pivots are now cheap too, which sounds like good news and produces drift: a team that changes direction every fortnight because it can, and never stays anywhere long enough to accumulate the validated learning the method was named for.

§ 6.4What to keep#

§All of it, reweighted. Validated learning as the only progress. The loop, with the emphasis moved from build to the question that precedes it: what belief does this turn test, and what would falsify it? Product Experimentation, the sixth whitepaper, is that discipline applied after launch. Product Sprint is it applied before. Ries would recognize both.