Eric Ries

the-lean-startup

Best Thing: Reviewers often praise "The Lean Startup" for its practical approach to entrepreneurship. The concept of validated learning and the emphasis on iterating through the build-measure-learn feedback loop are highlighted as valuable tools for startups to efficiently test their assumptions and adapt their strategies based on real customer feedback. Worst Thing: On the downside, some reviewers criticize the book for being repetitive and lacking in depth regarding certain topics. They feel that while the concepts are useful, the execution could be more thorough, and some readers are left wanting more detailed examples and case studies to illustrate the principles discussed.

Key Insights

  • Validated learning as the product. A startup’s primary output is not revenue or code — it is learning about what customers actually want. Every action should be designed to test an assumption and generate data. Speed of learning is the competitive variable, not speed of shipping.
  • Build-Measure-Learn as the operating loop. Ideas become products; products generate data; data produces the decision to pivot or persevere. The design question is always: what is the fastest way to get through this loop? MVP is the answer — not because it’s good enough, but because learning requires contact with reality.
  • Innovation accounting — knowing if you’re making real progress. Traditional accounting doesn’t work for startups because revenue and growth metrics are lagging and context-free. Innovation accounting requires identifying the specific assumptions your business model rests on (value hypothesis, growth hypothesis), testing them with leading metrics, and tracking movement from baseline to ideal.
  • The pivot as a structured hypothesis change. A pivot is not failure — it is the decision that the current set of assumptions is not working and a specific alternative should be tested instead. Ries catalogs types: zoom-in pivot (a feature becomes the product), customer segment pivot, platform pivot, engine-of-growth pivot. The key is that a pivot preserves accumulated learning while changing direction.
  • Small batches beat large batches. Counter-intuitive but empirically supported: shipping smaller increments more frequently is more efficient than large-batch production, because defects are caught sooner, feedback is faster, and the cost of being wrong is contained. The startup context is where this principle is most clearly visible.
  • Three engines of growth — viral, sticky, paid. Each engine has its own metrics and optimization logic. Viral: each user brings in >1 new user (viral coefficient). Sticky: churn rate vs. acquisition rate. Paid: customer acquisition cost vs. lifetime value. Choose one engine and optimize it before switching; mixing them is a sign of not knowing what’s working.
  • Five Whys as the root cause discipline. When something goes wrong, asking “why” five times in sequence typically leads from surface symptoms to systemic causes. The proportional response principle: invest in fixes proportional to the severity of the root cause, not the severity of the symptom.

— Drafted from external sources; review and edit to make your own.

From earlier notes:

  • Validated learning - point of a startup is in part to learn about a new idea
  • Build, measure, learn - turn ideas into product, learn what customers like, pivot or not
  • Innovation accounting
  • Build, measure, turn feedback loop (design in reverse)
    1. ID your leap of faith assumptions: value and growth
      • How to validate you are adding value and can grow?
        • Fb had high engagement and super fast vital growth. That was always going to be valuable, just a question of how much
      • Need to design your org to constantly test the most critical assumptions
    2. Build mvp
      • All about learning fastest way possible
      • Head start, stealth mode, etc aren’t much of an advantage
      • Only way to win long term is to iterate through build measure turn process faster and better than anyone else
    3. Innovation accounting:
      • How to tell if you’re screwing around or making real progress / learning
      • Value driver, business model, growth driver - know and testthese
      • Traditional accounting doesn’t work for startups, nor do most milestone metrics
    • If you are professing from baseline metrics to ideal as you learn, persevere. If not, pivot
    • Test directly all the time - eg use Google click purchases to get new customers and see which changes move the needle
    • Need to use cohort analysis not absolute metrics, which hide truths
    • Split testing (a/b) is almost always useful in product development
    • Kanban - must validate stories and investigate the learning before you can start new stories
      • Understanding and validation and split tests are built in, rather than tacked on after the fact
    • Learning milestones:
    1. Pivot or not
      • Determine your metrics up front.
      • Zoom in pivot (feature), platform pivot, customer segment pivot , business volume pivot, engine of growthpivot, value capture pivot, channel pivot
  • Small batches are actually more efficient
  • choose an engine of growth - viral, paid, or sticky
  • Building an adaptive organization
    • Constantly improving
    • Training new employees
  • 5 whys
    • About getting around blame to root causes
    • Engineers expected to make changes on day 1 - production system shouldn’t be so fragile that this is risky
    • Proportional responses
  • Engine of growth - mustbe via previous customers through
    • Word of mouth
    • Side effect of usage
    • Repeat purchase
    • Sustainable ads (
  • Viral coefficient