Eric Ries on build, measure, learn and the lean startup method

Executive overview

Most startups fail not because they build badly, but because they build the wrong thing for too long. The fix is treating every business as a scientific experiment: form a hypothesis, ship the minimum viable version, measure real user response, and iterate.

Eric Ries codified this as build, measure, learn — a feedback loop adapted from lean manufacturing that shortens the distance between assumption and evidence. The pivot and innovation accounting complete the system, giving teams language and metrics to know when to stay the course and when to change strategy.

The core insight: perfection through iteration, not perfection before iteration.

Why careful planning fails startups

• Without customer feedback, you can't define quality — you're optimising for the wrong target
• InView spent six months building software nobody downloaded; the failure exposed the value of early data
• A business plan that assumes customers will behave as predicted is making an untested hypothesis
• Aristotle never dropped anything from the Lyceum; he just thought about it — and was wrong

The build, measure, learn loop

• Minimum viable product (MVP): the most bare-bones version that tests a specific hypothesis
• Ship early not for speed's sake, but to get empirical data from real users as fast as possible
• Tighten the feedback loop: InView shipped code multiple times a day and studied user response each time
• If you aren't embarrassed by your first release, you've released too late
• Perfection is achieved through iteration, not by avoiding imperfection

The pivot: changing strategy without changing vision

• A pivot is not abandoning the mission; it is finding a different route to the same destination
• Classic pivot types: consumer to enterprise, B2B to B2C, subscription to freemium
• Everyone wishes they had pivoted sooner — the hard part is establishing the facts of what is actually happening now
• Use a GPS analogy: the destination (vision) is fixed; the route (strategy) adjusts to road closures (data)
• If each experiment yields smaller and smaller gains, you are on an asymptote — time to pivot

Innovation accounting

• A discipline for evaluating the net present value of experiments during the flat part of a hockey-stick curve
• Tracks whether incremental effort still produces incremental result, or whether returns are diminishing
• When every idea (change button colour, rewrite copy) has already been tried, the path is a dead end
• Establishing objective ground truth about current progress is what enables a clean decision to pivot

From lean startup to the Long-Term Stock Exchange

• Public markets reward quarterly results; this conflicts with the long-term thinking that produces great companies
• The Long-Term Stock Exchange (LTSE) separates investors into citizens (long-term, full voting rights) and tourists (short-term, limited rights)
• Johnson & Johnson's 1982 Tylenol recall looked costly in the short term but was correct for long-term trust — current markets make that call harder
• LTSE itself was built lean: software tools for equity management, headcount planning, and financial modelling were shipped first to gather data from the target audience before the exchange launched
• Same mission as lean startup, evolution of strategy — a pivot, not a departure