The Lean Startup

Summary

Eric Ries breaks down the "lean startup methodology", a process in which to come up with, execute and grow an idea from just that to a successful business.

Review

Clear, actionable and effective ways of starting a new venture.

Takeaways

don't assume, measure
be explicit about value and growth assumptions and then build-measure-learn
progress should be measured with Validated Learning

Key Points

what is startup
- a Startup tries to build something in extreme uncertainty
- a startup tends to be based on a Leap of Faith Assumption
  - these assumptions have a Growth Hypothesis and Value Hypothesis
how to startup
- progress should be measured by Validated Learning
- to achieve progress, adopt Build-Measure-Learn Loop loop
  - hypothesize, predict, measure
- kanban with limit on works in progress at any stage
prioritize
- systematically testing assumptions
- connect tests with long term vision
- everything that doesn't help with learning is wasted work
- making progress means understanding value that product is actually providing users and figuring out the Engine of Growth
etc
- make Proportional Investment when growing and scaling
startup should
- know where they are now
- figure out how to get closer to ideal
metrics
- actionable
- accessible
- auditable

Concepts

See Concepts

Notes

vision

1 start

goal of startup is to figure out what to build (as quickly as possible)
startup is like driving a car, you want to continuously steer vs planning for a rocket launch where a single error could be a disaster
this process is the Build-Measure-Learn Loop
true north, strategy, product

2 define

a Startup is an organization that is building something in environments of extreme uncertainty

3 learn

Validated Learning is discovering truth about current and future prospectsl
real progress in a startup is learning what work created value for users (anything else is waste)
goal of learning is to find the point of intersectino btw startup vision and what customers want
EVERYTHING that a startup is doing should be in the service of validated learning
eg: IMVU had faulty assumptions about what users wanted (people did not want to learn new social network, people wanted to talk to their existing friends)

steer

4 experiment

if you can't fail then you can't learn
- eg. Zappos beginning, take picture of other people's inventory and try to sell online to see if there is demand
  - source: The Lean Startup (Private)
to experiment, hypothesize, predict, and then measure
two most important hypothesis
- Value Hypothesis
- Growth Hypothesis

5 leap

Leap of Faith Assumption is riskiest part of plan
once assumptions are clear, build and validate
to measure learning, use Innovation Accounting and Learning Milestones
after Build-Measure-Learn Loop, startup needs to decide to Pivot or Preserve
Analogue and Antilogue can be a useful framework in Leap of Faith Assumption
priorities of a startup
- be able to systematically test assumptions
- connect present tests with long term vision
as founder
- figure out what parts of original plan were good and what parts need to change
- need to understand what is responsible for startup growth
nothing as effective as Genchi Gembutsu
- "see for yourself", talk to the customer

6 test

create a Minimal Viable Product
early adopters will be suspicious if product too polished, time on polish can be waste of time
types of MVPs
- Video MVP
- Concierge MVP
when creating a MVP, don't waste time trying to fully solve problem
- until you know your customer, you don't know what quality even is
- #mw: doing things that don't scale
remove everything that doesn't contribute to learning

7 measure

startup needs to do 2 things
- measure where they are now
- figure out how to get closer to the ideal
most startups, when asked if they're making the product better, will say yes but have no data to back it up
- eg. we launched a few features and our numbers went up (but then subsequently went down again)
- see The Lean Startup (Private)
begin process of Innovation Accounting
- 1: build mvp to gather baseline data
- 2: tune Engine of Growth to get closer to ideal
- 3: make decision to Pivot or Preserve
when testing, test riskiest assumption first
- #MW: deploy in IAD
beware of Vanity Metrics
focus on Cohort Based Metrics and Split Testing
kanban
- backlog, in progress, built, validated
story not complete until we have validated learning from it
have upper limit on items in any given stage #star
metric focuses
- actionable: establish cause and effect
- accessible: concrete units
- auditable: everyone has access (ideally, team that owns product should also own this)
  - you should be able to spot check data (eg. talk to customers and see if user behavior confirms the data)

accelerate

8 pivot

startup run rate is amount of pivots it has left
productivity is measured not in features released by the alignment of effort with value and growth
need clear hypothesis or can't decide about pivot
- eg. potbelly sandwich shop began as antique shop in 1977
things to measures registration, activation, rentention, referral
types of pivots
- zoom: focus on one from a whole
- zoom out: expand feature
- customer segment: b2b to b2c
- platform pivot: change from app to platform or ice versa
- technology: switch tech for same wndat
- channel: direct vs cto
- engine of growth: viral, sticky and paid
- value capture: charge at diff point in chain
- business architecture: high margin low volume vs opp

9 batch

small batch sizes help validate ideas quicker
- do continuous deployment
same person during multiple function, take advantage of breadth of skills
- corporate doesn't do this because of accountability hard
can't have too much work in progress, in manufacturing line, fills up all the machines but no visible signs of this with knowledge work #star
use hypothesis for what customer wants to work on batch and move forward

10 grow

sustainable growth: new customer come from actions of old
growth types:
- word of mouth
- side effect of using product (eg fashion)
- funded advertising
- repeat purchase (eg lightbulb)
startups don't starve, they drown #star
engines of growths
- sticky: rate of customer acquisition should exceed churn rate, can increase retention
- viral: customers spread product, aim for coefficient of 1 (every customer will result in 1 more customer)
- paid: cpa (cost per acquisition) lt ltv (life time value of customer)
focus on only one engine of growth at a time
when making customer segment pivot, consider changing engine of growth
product market fit, Marc Horowitz, customers pull product into mainstream #star

11 adapt

adjust processes and performance according to problem
- #MW: cross the bridge when we get there
small problems should have small response, large should have larger response
root of every tech problem tends to be human #star
use 5 whys
- have everyone involved in incident present
- avoid 5 blames
- don't send baggage through system, tackle new prob
- appoint 5 why master
mistakes
- be tolerant of 1st mistake
- no 2nd mistake
adopt Proportional Investment
failure is working on things that don't matter #star
small batch sizes, quickbooks waterfall anti-example

12 innovate

qualities of startup
- scarce but secure resource
- ind dev auth
- personal stake
in org, how to protect parent org from startup? manager will feel threatened

epilogue

frederick winsolw taylor's principles of scientific management #todo (Private)
see forest vanishing but can't see minds being wasted #analogy (Private)
nothing quite so useless as doing with great efficiency what should not be done at all, peter drucker #quote
100 years, what techniques and believes will seem outlandish and stupid

Children