Notes on The Lean Startup

I had the pleasure of recently finishing The Lean Startup by Eric Ries. I got the book as part of an AppSumo Lean Startup bundle last year, and so it came in the mail as a surprise. I thought it would simply be a nice summary of all the things I had already seen from Lessons Learned and various LeanLA meetups I've attended over the years. Oh, how wrong I was.

This book literally turned my world upside-down. I dropped all my other reading and worked hard to finish my first pass through the book as quickly as I could. Yes, the book did highlight all the major parts of the lean movement, including customer development and agile development practices, but it did so in a way that was much more effective and novel than I ever expected. The examples used are immediately understandable, the language concise and ultra-specific, and the book is full of advice I could put into action immediately. I'm in the process of planning the launch of a start-up, and this book has caused me to question and reinvent much of the launch strategy (take a wild guess -- we're planning to launch ASAP, like yesterday if possible).

I don't often recommend books so highly to close friends and family, but I've already hyped this book to over 10 people, and I know for a fact that at least 5 have purchased and enjoyed their copy at my recommendation. If you're an entrepreneur working on a start-up, buy a copy today, and read it now. I'm serious.

Below are my main notes on the book. Eric, thanks for your support to entrepreneurs everywhere.


I. A startup is any institution trying to innovate/create new product or service in conditions of extreme uncertainty.

a. Entrepreneurship is management of an institution; needs discipline and process.


II. Sustaining innovation vs. disruptive innovation


III. Goal of a startup is to get validated learning as quickly as possible

a. Use innovation accounting (cohort analysis and split tests to measure what you're learning)

b. Continuous and frequent experimentation with real users


IV. Launch now


V. Test 2 assumptions

a. Value hypothesis: that people will get value from your product

b. Growth hypothesis: that your product will spread and increase in adoption

c. Must test both as quickly as possible

d. Learning milestones


VI. Validated learning: empirical

a. Qualitative learning, then quantitative, then back and repeat


VII. Vision triangle

a. Product at top; can be optimized; tuning the engine

b. Strategy in middle; pivot

c. Vision at bottom (foundation); changes less


VIII. Goal is to get through Build Measure Learn cycle as quickly as possible

a. Learn -> ideas -> build -> product -> measure -> data -> learn...

b. Minimize TOTAL time through the loop


IX. Minimize batch size


X. Build only what need to directly experiment with in order to know if need to pivot or persevere

a. Remove every other feature or task


XI. Genchi gembatsu (get out of the building and interview; see for yourself instead of relying on others' reports)


XII. Concierge MVP: do all the work manually behind the scenes and fake automation

a. Actually build real automated product later


XIII. Video MVP: show a video of how product works and fake it behind the scenes

a. See how many people sign up


XIV. Quality not a concern with most MVPs; learning much more important

a. Can learn where customers care about quality and where they don't


XV. Secrecy not worth it; learning more important

a. Startup assumes it can go through Build Measure Learn cycle faster than competitor

XVI. KISSInsights

XVII. No vanity metrics

a. Gross sales, gross customers signed up, etc.

b. Will go up and right but won’t know if it's because of your recent work or rising tide


XVIII. Actionable metrics

a. Only meaningful metrics are cohort analyses to see if what you're building new is actually making any difference

b. Three learning milestones: establish baseline, tune the engine/optimize, pivot or persevere

c. Spend $5/day on Google Adwords to get 100 clicks; use those clicks to experiment and learn
i. Analyze customer adoption funnel


XIX. Kanban cycle: backlog, in progress, built, validated

a. Limited # in each bucket

b. Cannot move forward in process unless get through validation

c. Design everything in mind with validation and experimentation


XX. Analogs and antilogs: disruptive changes can’t just be compared to other different successes in past


XXI. Pivot: structured course correction designed to test a new fundamental hypothesis about product, strategy, or engine of growth

a. Zoom-in pivot: focus on one feature of a larger whole

b. Zoom out pivot: whole product needs more features

c. Customer segment pivot: focus on one customer segment
i. Sign LOIs

d. Customer need pivot: change in understanding of real need

e. Platform pivot: change from app to platform or back
i. Like creating self-serve model to create similar apps

f. Business architecture pivot: high margin/low vol to low margin/high vol (or back)

g. Value capture pivot: different monetization

h. Engine of growth pivot: viral, sticky, or paid growth models

i. Channel pivot: change sales channel

j. Technology pivot: change underlying technology

k. A pivot is a new strategic hypothesis to test

l. A startup’s runway is the # of pivots it can still make

i. Get to each pivot faster to extend runway


XXII. Hypotheses to test: Registration, retention, activation, referral


XXIII. Batch size

a. Stuffing one envelope at a time more efficient than large batch step by step
i. No WIP to move around
ii. Find out about quality or customer demand problems earlier
iii. Little rework

b. Continuous deployment including automated defect tests and reversions for problems and automatic rollout and experimentation to small subset of clients
i. ~50 releases per day


XXIV. Metrics

a. Actionable: specifically related to things you control

b. Accessible: understandable and available to all

c. Auditable: should be able to check data against real customers


XXV. Big batch death spiral: incentive to keep increasing and never ship


XXVI. Pull, not push


XXVII. Just in time


XXVIII. Engines of growth

a. Sticky: want high retention rate; metric: new customer acquisition rate minus churn rate

b. Viral: natural spread inherent in product; viral coefficient (# of new customers one customer refers)

c. Paid: customer acquisition cost and revenue per user


XXIX. Product-market fit


XXX. Building an adaptive organization


XXXI. Ask Five Why's to get to root problem


XXXII. Net promoter score