Jump to content
Pickup Asia Forums
French Ben

Eric Ries - The Lean Startup NOTES

Recommended Posts

#Lean Startup

##The Lean Startup Method
1. Entrepreneurs are everywhere: startups, small companies, big companies
2. Entrepreneurship is management
3. valudated learning  - frequentky learn through experiments
4. Build-Measure-learn - turn ideas into products, measure how customers respond, and learn to pivot or persevere
5. Innovation accounting: you need to focus on the boring stuff: measure proregress, set up milestones, prioritize work.

##Why do startups fails?
They do not know who they customer really is and what their product should really be.

##How this book is organized
1. Vision (how is an entrepreneur)
2. Steer (how to Build-Measure-learn)
3. Accelerate (how do it fast)

#1. PART 1: VISION
A startup should laways be managed and not "just do it"
Lean startup comes from Toyota: rely on knowledge and creativity of individual workers, shrinking of batch-sizes, just-in-time production and inventory control and acceleration of cycle times. Mianly, lean startup should allow entrepenurs to make testable predictions.

The goal of a startup is to figure out the right thing to build as fast as possible.
Many startups plan like then launch a rocket: one tiny mistake can mean a huge issue thousands of miles later. The Build-Measure-learn way allows you to drive a car: you see the road and you can always steer left or right. And it alows yu to stay focused on the destination. 
But even the smallest startups have to face the challenge of servicing existing customers while trying to innovate. 

#2. Define
##Who is an entrepreneur?
Some entrepreneurs are intrapreneurs - they work in big companies. Some have already gathered all the structure, personnel, vision for the future and appetite for risk-taking but the fire still isn't there despite them having the wood and the matches. They have issues like:
Once the team is set up, what do they do? What process should they use? How should it be held accountable?
The Lean Startup management covers this.

##If I'm an entrepreneur, what is a startup?
A startup is a human institution designed to create a new product or service under conditions of extreme uncertainty
Doesn't matter how big the company is, whether it is non-profit, or government-backed.
Innovation could mean new technology but also, repurposing technology, a new business model or bringing a product/service to a new place or a new category of customers.

##The Snaptax story
SnapTxa allowed customers to fill in their tax returns by taking a picture of them. But they were actually a big company from the beginning (7000 people) and they were able to create sustaining innovation but not disruptive innovation. They could do so by having lots of freedom from their management.
The problem is tax season is only once a year, so the company had a very conservative culture: they would only have one initiative that would be rolled out for the season. 
But after the Lean Startup, they now run 70 tests a week. When you have only one test, you have to sell only one idea - you then have a company of politicians, because you have to sell. When you have 500 tests, anyone's idea can run. 
Usually, the biggest obstacle is not the team or the entrepreneurs but the leaders and the middle managers. They are good because they are analysts, and all they want to do is plan and analyze. Management needs to put in place systems that allow lots of experiments. 

#3. Learn
What if you build something nobody wants? Many startups fail bu successfully executing a plan that leads nowhere.
Run experiments on your customers. Look at what they do, not what they say.
Validated learning is not after-the-fact rationalization or a good story designed to hide failure. It is a rigorous method for demonstrating progress.

##Validated learning at IMVU
We realized it was to hard to ask customers to switch from AOL to another kind of IM so we built an add-on on top of it instead. We planned the first product launch in 6 months, and despite lots of bugs, we still shipped it. And nobody downloaded it. After interviews, customers couldn't even say they wanted a 3D avatar, as they didn't know what it was. And they didn't know what an IM add-on was. And they don't want to invite friends to try a product that might suck. 
So we built a single-player mode. And they still wouldn't invite their friends (because they didn't know if the multi-player version would be cool). So we created an online game with strangers, that you could add on your IM buddy list. And people didn't want to invite strangers into their buddy list.
We thought customers didn't want to learn a new software and moving their friends over a buddy list, but that wasn't true. The real obstacle was that they didn't want an IM dd-on; they wanted a standalone IM network. They already had several. They also were ok inviting their existing friends to a new network. And they also wanted to make new friends with their 3D avatars, not stay with their old friends.

I had to give up all the interoperability software I had coded. It was depressing to have built something nobody wanted to use. And to have taken 6 months to learn all of that. We could have learned faster without trying to make a product "better" by fixing bugs and adding features.

###Value vs waste
Lean thinking defines value as providing value to the customer. All the rest is waste.
In manufacturing, customers don't care how the product is built as long as it works. But in startups, who the customer is and what they want is unknown. IMVU allowed us to discover what creates value for customers. That was the only value we got from these 6 months. All the rest was waste. 
Waste could have been avoided by shipping sooner, offering interoperability with only 1 network and not 12, or not even offering any interoperability. What if we had offered customers to download before building anything?
The effort that is not absolutely necessary to understand what customers want can be eliminated. It's easy to kid yourself about what customers want and learn things that are irrelevant. Validated learning is backed by collecting data from real customers.

##Where do you find validation?
In the next days, despite our improvements, nobody downloaded the product.
We did A/B testing on the website. After months of dicthing the IM add-on strategy, we finally understood: customers wanted to use IMVU to make NEW friends online. We were the only "anonymous" network. We found out through hundreds of experiments. Each bit of knowledge suggested new experiments to run.

###The audacity of zero
It is better to say you have zero customers than very few. Investors can still imagine overnight success.

#4. Experiment
What can be changed safely, what will anger customers? Which customer opinions should we listen to, if any?

Zappos started with a small experiment: They took pictures of shoes in real stores and promised to come back and buy them if a customer bought them online. This answered the question: do people want to buy shoes online?
By bulding a product instead of asking questions, they got to observe real customer behavior, learned about their needs (eg. for discounts and the effect it has on their perception) and saw interesting unepexted behaviors (eg. customers returning shoes)

##For long term change, experiment immediately
HP employees didn't volunteer despite programs existing, and volunteered without offering their skills (eg. manual labour).

Test these 2 things:
* Value hypothesis: 
    Find an indicator that employees find donating their time valuable
    For example, you could send some of them volunteering and look at the retention rate (how many do it again)
    
* Growth hypothesis:
    Find how new customers will find a product or service
    Once a program is up and running, you could see how fast it spreads among employees, how often the word is spread.
    In this case, take a small number of employees, like 12 and offer them a great opportunity to volunteer.
    The first customers should be the ones who really need the product: they will be more forgiving to mistakes and will be more eager to give feedback.
    In this case, it was best to take employees who were numotivated and disconnected from the company's values.
    
    Once they participated in the great volunteering program, yu could measure what they did: how many completed their volunteers assignment? How many re-volunteered. How many referred a colleague?
    The whole experiment could take place in only a few weeks. And it can happen in parallel with the plan still being formulated.
    
##An experiment is a product
Traditionally, the product manager says "build that" and the engineer does it. Instead, have your team answer these questions:
1. Do consumers recognize that they have the problem we are trying to solve?
2. If there was a solution, would they buy it?
3. Would they buy it from us?
4. Can we build a solution for that problem?

Many people just skip and jump to question 4 and build something nobody wants.

###The village laundry service
To see if Indians were ready to give their clothes to someone else to have them cleaned, P&G had trucks park around many areas and offer the service. This way they could test several questions in real life: did speed matter? Was cleanliness a concern? What were people asking when the truck left with their clothes? It turned out they were happy but were suspicious of the washing maching installed on the truck, which was therefore replaced by a kiosk.
They parked the trucks at different places to identify target customers.
They discovered customers were ready to pay more to have the clothes ironed and returned in less than 4 hours.

###A lean startup in government
Obama launched CFPB to help Americns with their tax problems. despite a budget of 500 million USD, it was managed as a startup. Therefore, they created an MVP to test their assumptions: they thought too many people would call for help. But this is an assumption. What if victims of fraud don't seek help? What if they have different notions of what problems are important? What if they call for help not related with the agency's purpose?

They created an MVP with a phone platform using Twilio. It just took a few hours. Then they advertised in just a few blocks of one city, where they could afford to pay for billboards, newspaper ads and targeted online ads. It just took a few weeks and a few thousand dollars.

Even well-seasoned managers and executives struggle to launch innovative new products. Their challenge is to overcome the prevailing management thinking that puts its faith in well-researched plans.

#PART 2: Steer
##How vision leads to steering
1. Build a product
2. Measure data
3. Learn ideas
4. Repeat

You need to minimize the TOTAL time of this feedback loop.
The first things you need to test are your *value hypothesis* and your *growth hypothesis*. You will need to test your *leap of faith assumptions*, the ones one which everything depends.
Create an MVP as fast as possible, even if many features are lacking.

#5. Leap
Facebook could sell high because of several metrics. More than half the users came back to the website every single day - which validates the value hypothesis (customers find the product valuable). The growth hypothesis was also validated as 75% of Harvard students used it after less than a year, with no advertising.
Being free and not psending money on advertising would it make it easy for you to build the new Facebook? For startups, the strategy is to figure out the right question to ask.

##Strategy is based on assumptions
Every company makes assumptions. A startup should test its assumptions as fast as possible.
Leap-of-faith assumptions might include: customers have a significant desire to use a product like ours, supermarkets will varry our product. If they are true, great opportunity awaits. If they are false, the company risks total failure.

###Analogs and Antilogs
If you are selling iPods, the Walkman is an analog: it answers the question whether people would be ready to listen to music with earphones in public.
Napster was the antilog - it suggested that people would want to download music but not pay for it. Steve Jobs took the leap of faith that they would still be ready to pay for it and he was right.

###Beyond "the right place at the right time"
Many people were at the right place at the right time. Ford had 500 competitors. Facebook had lots too. But very few had the ability to discover which parts of their plan were working and which were misguided.

###Value and growth
Some companies use the appearance of growth despite destroying value - Ponzi scheme or Lehman Brothers. Innovation accounting helps differentiate false startups from true innovators.
A startup must have a financial model- the instincts must be transformed into data.

###Genchi gembutsu
It is the Toyota Way of "going out and see for yourself". It means you cannot rely on the reports of others. Whether you are in sales, distribution or public affairs, you have to go see for yourself. 
That is how the manager went on a trip with a Sienna and spoke to owners and realized that the family who sat in the back of the truck was also responsible in the purchase decision. he then spent his budget on internal comfort features, which are critical to long-distance family trip (which are more common in the US than Japan, too)

###Get out of the building
Go and meet your customers. Cook took 2 phonebooks, called random people and asked them if it was frustrating for them to pay bills by hand. They didn't know much about PCs so more specific questions were useless as these people were mainstream customers and not early adopters.
This allows to create a *customer archetype*, a document that humanizes the proposed target customer. To create this document, look at interaction design and design thinking. But the problem is you never know who your customer really is. Which is why it must be tested with user Experience - called **Lean UX**

###Analysis Paralysis
Some entrepreneurs build too fast. Others build too slow, endlessly refining ther plans, talking to customers, reading research reports, whiteboard strategizing. They miss all the subtle ineractions with customers that allow to test their assumptions. How do you know when to stop analyzing and to start building? The answer is the MVP.

##6. Test
Groupon started as a Wordpress blog with PDF coupons sent through email.
The MVP should be the fastest way to get through the Build-Measure-learn loop. The MVP begins the process of learning, it doesn't end it.

###Why first products aren't meant to be perfect
First products are designed for early adopters. They only need a product that works at 80%.
These people lined up for iPhones that couldn't copy and paste, and the first people using Google could only answer queries about Stanfard University.
What they care about is being the first ones to adpot a new products or technology - a shoe, a player, a phone. They are suspicious of products that look too polished- it makes them feel not special for being the first one. 
Entrepreneurs dream of a pretty, polished product and it is hard for them to accept that the MVP should be buggy or incomplete. Do not overestimate how many features an MVP requires. When in doubt, simplify.
For example, you have a one-month free trial and customers will sing up for it once your product is ready. You should write in your assumptions: we think 10 percent of customers will sign up for the trial. And the number of features should be really low.

###The video MVP
Dropbox assumed that file synchronization was a problem most people didn't know they had. Once you try it, you can't imagine how you lived without it. You couldn't expect to hear this from a focus group. Dropbox's strength was that it worked seamlessly, like magic. But you couldn't demonstrate that in an MVP, so they made a demonstration video with lots of hidden jokes for early adopters. Their beta waiting list went from 5K to 75.

###The concierge MVP
Food on the Table started with one customer - they did everything offline, including delivering her grocery list for just 10 bucks a month. It was extremely inefficient, and all customers got the same VIP treatment until they started automatizing more and more processes. This way, they could scale something that worked instead of inventing something that might work in the future.

###Pay no attention to the 8 people behind the curtain
Aardvark was a Google for subjective questions: (eg. what is the best place to eat around here?) - while people thought they were talking to an AI-machine, the answers were actually coming from other humans. Doing this every week to real people they invited to their offices, they pivoted lots of time before being bought over by Google.

###The role and quality of design in an MVP
If we do not know who the customer is, we do no know what quality is.
In other terms, you do not know that your customer wants a beautiful website, or that is perceives it as valuable.
Do not be afraid of shipping something that feels incomplete. Criagslist and Groupon were loved by customers despite being low-quality.
And sometimes, a low-quality MVP can help you build a better product. Customers trying IMVU didn't like static avatars. It was bad news, because it meant we had to code all the animation and movement; so we tried something else: a "teleportation" by just clicking anywhere they wanted, with no animation. And they loved it.
MVPs require the courage to put your assumptions to the test. If you create something cheap and the customers cannot use it, that means you must invest more into design. But you should also ask yourself: what if they don't care about design the way I do?
When creating an MVP, remove any feature, process or effort that doe not ontribut directly to the learning you seek.

###Speed bumps in building a MVP
Most common speed bumps: legal issues, competitors, branding risks and impact on morale.
You will not get your idea stolen, even if you try. Time spent away from customers (in stealth mode) is unlikely to provide a head start. The only way to win is to learn faster than anybody else. If you fail, almost nobody knows you anyway so you do not have any reputation to lose.
Finally, MVPs provide a needed dose of reality and often bring bad news. Visionaries are often afraid of releasing a flawed MVP, that is too small or too limited. But you need to be able to make a go/kill decision on a regular basis.

###From the MVP to innovation accounting
You have to commit to never give up hope, no matter what comes from the MVP. The MVP is the first step to a journey of learning. Outside stakeholders and investors need to know that when we fail, it is not because of a failure to plan or failure to execute.

##7. Measure
Entrepreneurs are naturally optimistic, this is why the myth of perseverance is so dangerous.

###Why something as seemingly dull as accounting will change your life

Accounting allowed GM to set lar milestones for each of its divisions.
Many startups think the same way: when asked "are you making any progress?" they answer "we made a number of changes, our customers seem to like them and our overall numbers are high erthis month. We must be on the right track"
But this is a wrong indicator. 

### An accountability framework that works across industries
Innovation accounting begins by turning leap-of-faith assumtions into a quantitative financial model.
For a manufacturing company, rate of growth depends on the profitability of each customer, the cost of acquiring new customers and the repeat purchase rate.
For eBay, sellers and buyers are both coming to use a big network. The metrics are the retention rates or sellers and buyers.
These businesses have different drivers of growth but can still use a common framework.

###How innovation accounting works- three learning milestones
1. Establishing the baseline
When choosing the many assumptions in a business plan, it makes sense to test the riskiest assumptions first. A media business selling advertising must test 2 basic assumptions: can it capture the attention of a defined customer on an ongoing basis? Can it sell that attention to advertisers?
Here, the advertising rates for a particular customer segment are known, so the riskier assumption is the ability to capture attention. So the first experiments should involve content production rather than advertising nsales. The company can produce a pilot episode or issue to see how customers engage.

2. Tuning the engine
Every product development, marketing or other intiative should be improving one of the drivers of its growth model.
If you improve the design of your product, you presuppose that the activation rate of new customers is a driver of growth. If the new design doesn't change this activation rate, then it should be considered as a failure.

3. Pivot or persevere
If you are not moving the drivers of your business model, then you are not making progress.

###Innovation accounting at IMVU
We used to have no sales, improve the product and measure conversion rates, customer counts and revenue to show we had done a good job. But customer behavior didn't change. Our product changes were having no effect.

###Improving a product on 5 dollars a day
We tracked the funnel metrics critical to growth: customer registration, downloads, trials, repeat usage, purchase.
We bought 5 usd of Google AdWords a day. and got a hundred clicks a day. Every day we could test our product's performance with a brand new set of customers.
One day we would target first-time customers. The next day we changed the way new customers were intiaited into the product. On other days we would add features, fix bugs, roll out a new design, try a new layout for the website.
Each days' customers were independent of those the day before. It became clear our funnel metrics were not changing. All numbers stayed the same after 7 months.

###Cohort Analysis
Instead of looking at total revenue or total number of customers, look at cohorts, i.e the performance of each group of customers that comes into contact with the product independently.
Our numbers changed very little, so we invited customers again. 7 months prior, we dismissed uninterested customers assuming they were not our target market. But now I needed to understand why customers didn't respond to our product improvements. We made it easier for customers to engage with their friends on IMVU but they still didn't use this feature. making it easier to use was a waste of time.

##Optimization vs learning
For startups, predictions, product and strategy decisions are difficult because they might be building the wrong thing.
Bigger companies often spend lots on making a better product and pressing engineers, and the results come not from better products but come from decisions taken in the past- the engineers' work doesn't change the metrics. And managers use cheap tricks to pretend it works: last-minute ad buys, channel stuffing, whiz=bang demos. These are vanity metrics you shouldn't be using.

###Vanity metrics:  a word of caution
Vanity tactics allow you to add new customers but not to improve the yield on each new group. The efforts to tune up the engine are not bearing much fruit.
You need actionable metrics

##Actionable metrics vs vanity metrics
Grockit is a startup bringing a style of education to students where the teacher isn't there and the students are autonomous. The MVP was just Farb simply teaching test prep via WebEx (like Zoom). Within a few months, he was making $15,000 a month and could raise money from investors.
Their growth followed a very rigorous evrsion of the agile development methodology known as Extreme Programming.
The problem: there was not enough growth in the use of the product by customers.
They worked in *sprints*, one-month iteration cycles. They would prioritize the work of each month by writing *user stories*, from the point of view of the customer.
At the end of each sprint, teams delivered new product features, collected feedback from customers through interviews and saw many liked the features.
There were always positive metrics, but the teams didn't know whether the prioritization decisions made sense; they just listened to the boss, hoping he knew which features to prioritize and how to get more customers to sign up and pay, and how to get the word out about Grockit.
Farb saw the company was making progress but couldn't say if his vision was right. He agreed to put his vision to the test of customers.

The problem was that Grockit's progress was measure by vanity metrics: total number of customers, total number of questions answered. It seemed they were making progress but it was making little progress.

###Cohorts and split-tests
They switched to cohrot-based metrics. Instead of looking at casue-and-effect relationships after the fact, they launched each new feature as a split-test-experiment. They did A/B testing and realized many new features didn't change customer behavior. They also realized customers didn't want more social communication with other students as they believed.
So Grocki tdeveloped a complete solo-studying mode, where students could learn alone or with others. Over time, it became clear what students wanted was a combination of both.

###Kanban
Grockit changed the product prioritization process. user stories are always either:
1. In the backlog
2. In progress
3. Built
4. Validated
The kanban rules says that none of these 4 bucks can accept more than 3 stories at the same time. If an idea in progress has been built, it cannot go in the Built bucket before one of the 3 ideas in it have moed to the Validated bucket.
Engineers can also learn to include validation exercises from the beginning. Do not start working on a feature that hasn't been validated by a split-test experiment.
Productivity needs to be measure not from production of new features but through validated learning.

###Hypothesis testing at Grockit
Even though lazy registration was considered as a best practice in the industry (users are asked to register only after using the product for a while), A/B test showed that the lazy registration group has the same rates of registration, activation and retention as others. The extra effort of lazy registration was therefore a waste.
It also showed that customers based their decision on Grockit on something other than the use of the product. Theyc ame on the website with very little information and still registered., suggesting the company's marketing and positioning had a more significant impact on attracting customers than adding new features.

##The value of the 3 A's
Metrics should be Actionable, Accessible and Auditable

###Actionable
It should show clear cause and effect. The results should make it very clear what actions need to be taken to replicate the results.
Or else it's just vanity metrics, like number of clicks or number of users, which could be the result of a PR push or one guy clicking a lot of time from his browser.
With no split-testing, you assum enumbers go up thanks to your actions and they went down because of someone.

###Accessible
Make reports as simple as possible so everyone understands them.
Cohort-based reports are the gold standard of learning metrics: they turn complex actions into people-based reports. They say out of 100 people, X downloaded the product/logged in/chatted/purchased. This is more helpful than the simple number of people who purchased
Grockit sent the data about each split-test experiment to all employees every week explained in plain English.

###Auditable
Many people will challenge the data, if they don't like it. Managers need to be able to spot check the dat with real customers.
Hollywood movies like to show an entrepreneur who has an epiphany and becomes a millionaire, but in real life, 95% of entrepreneurship has to be measure by innovation accounting: product prioritization decisions, deciding which customers to target or listen to and having the courage to subject a vision to constant testing.
One decision is the most difficult one, the most time-consuming and the biggest source of waste: pivoting.

##8. Pivot or Persevere
When is it time to pivot? Ask yourself this question: are we making sufficient progress to believe that our original hypothesis is correct?
The biggest destroyer of human creativity is the misguided decision to persevere.
Creativity is not adding more gadgets or features, it is creating value and driving growth, which both can be measured.

###Innovation accounting leads to faster pivots
Many companies stay in the land of the living dead - that have achieved just enough to stay alive and employees and founders don't want to give up, thinking success is around the corner.
A good startup pivots early, like 8 months down the line. The more money and time have sunk into an idea, the harder it is to pivot.
It is also necessary to identify your leap-of-faith questions explitictly and make quantitative predictions about each of them.
Sometimes, you want to declare success because some metrics do well, like total usage (vanity metric).
It is easier and smarter not to spend on premature PR as it allows to pivot without public embarrassment. 
Get direct feedback from your customers.

A pivot can be a *customer segment pivot*, where you keep the product and adapt it to different customers. Like @2Gov which first targeted activists who wanted to reach to senators and pivoted to target large organizations wanting to lobby.
Despite many letters of intent, these companies stalled and didn't buy the product once it was available.
David then did a *platform pivot*: instead of investing the investors' money in a value-destroying engine of growth, he fired his salespeople and created a platform where anyone could become a customer, like a Google AdWords platform. For these 3 pivots, the MVPs accelerated every time. The first MVP took 8 months, the second took 4 and the third one took 3.

###A startup's runway is the number of pivots it can still make
How long can your money last? Your runway is 10 months if you have 1M and spend 100K every month. But the real runway is the number of Build-Measure-learn loops you can still do.
Count your runway in pivots, not months. You have to pivot fast.

###Pivots require courage
Why do entrepreneurs pivot too late?
1. They look at vanity metrics and form false conclusions.
2. They have unclear hypothesis. They launch and "see what happens". But early results are always ambiguous and they don't know whether to pivot or not.
3. They are afraid. So they only test once the vision is fully represented and the funding is running out. If there are people watching, it makes it even scarier.
When launching Path, Dave Morin faced lots of criticism from the press but had the courage to persevere as customers liked his product.

###The pivot or persevere meeting
I recommend every company has a regular "pivot or persevere" meeting set in advance every few weeks or months. Bring reports of product optimization efforts, an da comparison of how these results stack up against expectations. Bring conversations with current and potential customers.
Pretty often, you can repurpose what you have bought and bring it to a new business model or customers.

###Failure to pivot
When business goes well, you often to not want to pivot. It is a classic startup trap. You forget to create more MVPs to test new ideas.
However, it is also important to realize that MVPs are designed for early adopters. Once you have acquired them, the mainstream customers have different needs and are much more demanding.
Very fast, your startup will need to to a *customer segment pivot*. You know you are there when the vanity metrics are good but the effots of tuning the engine reach diminshing returns.
At IMVU, we did everything to improve the product's activation rate (the rate at which new customers become active consumers): lots of A/B testing with usability improvements, new persuasion techniques, incentive program, customer quests and game-like features, but the activation rate stayed pretty low, despite the fact the company was still growing.
The truth was that we already had the early adopters, and addressing mainstream customers was a different game.
It became clear that our testing was only to chase growth and revenue, and we had forgotten that testing is designed to to test a clear hypothesis in the service of the company's vision.
We developed a clear customer archetype based on observation and conversations, made the product way easier to use. Tests showed the new design worked better.

##A catalogue of pivots
###Zoom-in pivot
When one simple feature becomes the whole product. Eg. Votizen pivoted from a full social network to a simple voter contact product

###Zoom-out pivot
The product becomes way larger

###Customer Segment pivot
You change your target customer

###Customer Need pivot
When you know your customer very well, you can realize your customer's problem is not the one you wanted to solve. It's like Potbelly Sandwich Shop which started as an antique shop.

###Platform pivot
When an application becomes a platform for third parties and vice-versa.

###Business architecture pivot
Geoffrey Moore observed companies follow either business architectures:
1. high margin, low volume or
2. low margin, high volume
You can pivot from one to another

###Value capture pivot
Changes to the revenue model, to the way that the company captures value or monetizes its services

###Engine of growth pivot
There are 3 primary engines of growth: sticky, paid, and viral. You can pivot to seek faster or more profitable growth.
Changing the engine of growth often requires changing the way value is captured (the revenue model).

###Channel pivot
This is a change in the distribution channel. You can deliver your service or product through a different channel.
eg. pivoting from a complex sales process to "sell direct".
This is the one that the Internet has forced.

###Technology pivot
When the customer segment, customer's problem, value-capture model and channel partners don't change and you can provide superior performance with different technology.

##A pivot is a strategic hypothesis
Your business will have to pivot many times. Don't just copy the pivots of companies because they are famous. They had to pivot many times too.
A pivot is a new hypothesis that you can test with an MVP.
Companies pivot all their life. There are also many later-stage pivots like: the Chasm, the Tornado, the Bowling Alley.

#Part 3: Accelerate
##Start your engines
How often should you release a product? Daily, weekly, quarterly, yearly?
What should employees be doing? Hw do we hold people accountable for learning?
The first critical question is which activities create value and which are waste- value in a startup is not the creation of stuff but validated learning in creating a sustainable business: what products do customers really want? How will business grow? Who is our customer? Which customers should we listen to and which should we ignore? These questions must be answered as quickly as possible.
Take advantage of the power of small batches. Conduct experiemtns without massive upfront investments in planning and design.

##9. Batch
The small-batch approach creates a new finished product every few seconds (a stuffed and sealed envelope).
The big-batch approach creates lots of finished products at once at the end (100 envelopes get stuffed, then sealed- takes 15 minutes)
What if the customers don't want the product? Small-batch allows you to find out sooner with lower investment.
Big batch is ok for mass production but for startups, you should focus on small batches and do things one at a time. It allows to identify quality problems faster, even if the whole production line has to stop to fix them.

###Small batches in entrepreneurship
The goal of the learn startup is not to build more stuff efficiently, it is to build a sustainable business.

####Small batches at IMVU
Apple uses mass production methods to release 1,500 new features in one batch: product managers figure out which features will please customers, designers design them, engineers build them, and it gets tested.
But at IMVU, these people work together on ONE feature at a time, release it immediately for a small portion of the website visitors.
The website gets changed almost 50 times a day. This allows to spot problems very fast and prevent bigger problems later.
We have an *andon cord* equivalent to find a defect- like an engineer turning the Checkout button to white on a white background:
the "immune system" detects the change and removes it automatically, everyone in the team is notified and the team is blocked from inducing further changes (to avoid a snowball effect) until the root of the problem is found and fixed.
This is called *continuous deployment*.

##Small batches in action
SGW created a military X-ray machine prototype in just 3 days, had it tested, re-designed it after 5 days and started mass production after only 15 days.

###Small batches in education
In education, the curriculum can change only once a year. School of One allows to focus on one student and adapt to their level and speed for just a few days.

##The large batch death spiral
If you are a designer, you might be tempted to draw your 30 designs and send them to the engineering team when you are finished.
But the engineers come back with problems/questions which prevent you from focusing on the next batch (and when you do, they have to wait for you).
You might be tempted to work on even bigger batches to solve this problem but it is a death spiral. It takes longer and longer to develop the product and nobody wants it to have bugs after all this hard work.
Working on one single drawing at a time solves this problem.

##Pull, don't push
At Toyota, when your local dealership changes your bumper, he pulls one from the regional dealership (to replace the one he just gave you), which in turn pulls one from the central dealership, which pulls one from the factory. Every node on the chain bulls one bumper from the one above. It's Toyota's **Just in time** method. This way the amount of inventory is always low and warehouses shrink.
There are less WIP parts cluttering the floor this way. In a startup, the WIP waste is your incomplete designs, business plans and not-validated-yet assumptions.
They need to be reduced by :
1. reducing batch size and 
2. converting push methods to pull

In manufacturing, pull means producing not more or less but exactly what the customers need.
In lean startups, pull means running experiments to know how much they want.
As soon as you formulate a hypothesis that you want to test, design and run this experiment ASAP using the smallest batch size possible.
Remember that before building, you need to make sure the product will allow you to test what you want to learn.

###Hypothesis pull in clean tech
Alphabet could build a physical product (of recycling waste heat into energy) in just 6 weeks by using low-cost semiconductor materials that massive existing infrastructures already use to make CPUs.
They thought attaching semiconductors to turbines would be imple and cheap. They tested this in small batches by building small-scale solutions for its customers as a way of learning and discovered it wasn't. Power companies having low tolerance for risk, that made them unlikely early adopters, so Alphabet was able to pivot after just 3 months.
They also experimented with manufacturing firms, who can experiment with new technologies in separate part of their factory.
They also realized that customers were not willing to pay the most, and wanted a low cost per watt.
They were able to realize this faster and pivot faster thanks to working in small batches.

Toyota created the best system of management in the world, but they also have unleashed the creativity of its employees. Startups are made of people and must face human challenges.

##10. Grow
2 companies weere different: one was en eBay for collectibles, the other sold database software. Both had the same problem: they had early customers, early revenue, had validated and invalidated hypotheses, and had raised money from investors. But neither company was growing. Their early growth had flatlined. They didn't know whether to advertise more, focus on product quality or new features.
Both didn't know how to grow. Both were using the same *engine of growth*.

##Where does growth come from?
Sustainable growth is when new customers come from the actions of past customers. This is done by:
1. Word of mouth
2. As a side effect of product usage (seeing someone's car or Paypal payment)
3. Through funded advertising paid by revenue. The lower the cost of acquiring a new customer than the revenue they generata, the faster the growth
4. Through repeat purchase (lightbulbs, cable subscription, not weddings)

Each source is a loop that should turn fast to bring growth.

##Th e 3 engines of growth
Most startups create waste by arguing how to prioritize new development once they have a product on the market. At any time, the startup should be investing its energy in finding new customers, servicing existing customers better, improving overall quality or driving down costs.
There are always plenty of ways to make a product better but most of these ideas only make a change at the margins, they are mere optimizations.

###The Sticky Engine of Growth
Here, both businesses focus on having long customer retention: people chasing collectibles will log in every day and companies choosing a database will keep it for years. Both businesses expect you will continue to use their product once you start.
Therefore they should look at their **churn rate** very carefully, the fraction of customers who fail to stay engaged.
For a sticky engine of growth, if the customer acquisition rate exceed the churn rate, the business will grow. 
Companies there need to look at the **rate of compounding**, i.e natural growth rate minus churn rate. A high rate of compounding means rapid growth even without advertising or viral growth.
But both these startups were looking at vanity metrics: total number of customers. They were also looking at actionable metrics such as activation rate and revenue per customer, but in a sticky engine of growth, they have little impact on growth.
This would allow to see that many customers are coming through the door but they don't stay long- which means they should get reasons to stay, like better listings, direct messages on special offers. 
In this case, growth doesn't come from sales or marketing but from improving customer retention. 

###The Viral Engine of Growth
Hotmail added a "sent with Hotmail" signature and attracted lots of new customers.
Tupperware salespeople try to convince their friends to also become reps.
The **viral coefficient** calculates how many new customers will use a product for each new customer signin up, i.e how many friends the customer will bring with him.
For a product with a viral coefficient of 0.1, each customer brings 0.1 friends.
Even tiny changes in this number will cause huge changes in growth. So customers should be able to invite friends easily. And it can be a good idea not to charge customers directly but generate revenu through indirect sources like advertising.

In the viral engine of growth, the customer's money does not drive growth, it only shows the customer likes you. If Facebook and Hotmail had started by charging customers, they would probably have failed. Customers didn't give money but they gave their attention and time, which are very valuable to advertisers.

###The Paid Engine of Growth
Company A pays 80 cents in Adwords to make 1$ from a new customer. Company B makes 100K from each new customer and spends 80K acquiring it. Both have the same growth rate: 20%.
IMVU customers wanted to meet new people on the app. They didn't want to invite their friends, they thought it was the company's job. So IMVU couldn't use viral growth and had to use paid growth instead, through advertising.
When doing so, look at the lifetime value of a customer. The LTV should be higher than the cost per acquisition (CPA).
When you pivot, like to a new customer segment, you might also need to change your engine of growth.
IMVU knew that their customers were not very wealthy and allowed them to send cheques and pay cash.

###A technical caveat
A product can be both viral and low churn rate. Or be high margin and high retention. But usually it is recommended to focus on one engine of growth.
If you do not know which one to choose, time out of the building with customers will tell you.
Only after pursuing one engine thoroughly should you consider pivoting to another one.

##Engines ot growth determine product and market fit
When you have achieved product/market fit, the market is trying to pull out the product out of your startup faster than you can produce it- that's Facebook taking over the world overnight. If you are asking whether you are there yet, then you are not.
How do you get closer to that point?
A startup with a viral coefficient of 0.9 or more is on the merge of success.
If you focus on the viral ngine of growth, then you should focus on affecting customer behavior and ignore things that don't. You do not need to specialize in marketing, advertising or sales.
If you are focusing on paid engine of growth, you need to develop those marketing and sales functions urgently.
You can evaluate how closer you are getting to the product/market fit with each Build-Measure-Learn test iteration.

###When engines run out
Every engine of growth eventually runs out of gas. It is tied to a given set of customers with their habits, preferences, advertising channels and interconnections. At some point, they will be exhausted.
Transitioning from the early adopters who used the MVP to the mainstream customers take lots of energy. Once you have many early adopters, this is usually when you should theoretically stop work in product development.

Many companies look at vanity metrics, think they are making their product better when they aren't. The growth is all coming from an engine of growth that is working and bringing new customers. When growth slows down, it provokes a crisis.
All companies, even big ones will suffer this, and should always be tuning their engines of growth and developing new sources of growth for when that engine inevitably runs its course.

##11.Adapt 
Your company might be a new company for the past 6 months even though it kept the same name and same people, and you didn't realize.

###Building an adaptive organization
Even a small startup should have a training program for new hires. Which means your processes should be standardized and each new employee should have a mentor.
Adaptive organization is when your processes adapt to current conditions.

###Can you go too fast?
Startups should go fast before they run out of resources and die. However, focusing on speed alone is destructive. You need speed regulators.
One in the **andon** cord that evey Toyota employee can pull to stop the production chain. Defects cause a lot of rework, low morale and customer complaints.
Yes, the MVP can be buggy, but it is only an issue if you spend time fixing everything to build a product nobody wants.

###The wisdom of the 5 why's
To accelerate, lean startups need a process thatprovides a natural feedback loop. Adaptive processes force you to slow down and prevent the problems that are wasting time, so you can speed up again. So should you create a complete training program? First, answer the 5 whys.
At the root of every technical problem is a human problem.
1. Why did the machine stop? (there was an overload)
2. Why was there an overload? (it wasn't lubricated enough)
3. Why not? (the lubrication pump was not pumping enough)
4. Why? (the shaft of the pump was worn)
5. Why? (there was no strainer attached and some metal scrap got in)

Ask why 5 times to get to the root of the problem. Often you will see it is a human problem (nobody attached a strainer).

Imagine IMVU received complaints about a new version of the product: eg. a new release disabled a feature for customers
1. Why? Because a server failed
2. Why? Because an subsystem used it the wrong way
3. Why was it used the wrong way? The engineer didn't know how to
4. Why? he was never trained
5. Why? Because his manager is "too busy" and won't train him.

Solution: train people.
####Make a proportional investment
Do not overinvest if the problem is small
Do not overengineer

###Automatic speed regulator
Many startups work too fast and trade quality for time, which causes sloppy mistakes. Five whys prevent that and allow the teams to find their optimal pace.
Some engineers even say the 5 whys coupled with working in small batches allow you to derive all the other principles of the Lean Startup and allow you to respond to problems quickly as they appear without overinvesting or overengineering.

###The curse of the 5 blames
When things go bad, we want to point fingers. Ways of avoiding that include:
Making sure everyone is in the room when analyzing the cause of a failure. Whoever gets left out often ends up being the scapegoat, whether it's ajunionr or the CEO. If a mistake happens, shame on the seniors for making it so easy to make that mistake.
At IMVU, new engineers were asked to change something to the production environment on their first day. If it broke, they would be in charge of making sure it doesn't happen again. At first it was intimidating for them but then they realized the development system had to be robust and they could be more creative and less fearful.

####Getting Started
Be tolerant of all mistakes, and make sure they only happen once.
Mistakes happens because of bad processes, not bad people.

####Facing unpleasant truths
You will have to invest in prevention.
Start small - apply the 5 why's to something not involving the customer. You can choose to call an automatic 5 why's meeting for every credit card transaction problem.
Mave a meeting moderator who will make sure steps are followed and decisions implemented.

##The Five Whys in Action
It brings team closer together, especially if you do it well by having everyone involved attend.
eg. Why did the user get an error 500? A gem was incompatible -> Remove it
Why? We added a new version of the gem -> Use Bundler to manage gems
Why did we add a version of the gem in production without testing? We thought it was useless -> Write a functional test that will catch that in the future
Why do we add gems we don't plan to use right away? To prepare for code push in production -> Automate gem management into Continuous Ingration and Deployment
Why are we doing production on Friday? Because we can -> Deploy on Monday now


###Adapting to smaller batches
1. Small teams who are fully engaged and with members working on different roles
2. Short cycle times
3. Fast customer feedback
4. Empower teams to make fast and courageous decisions

Quickbooks now has lots of small teams of 5 people that iterate with customers as fast as possible and run experiments. Each team works on a feature for 6 weeks end to end, testing it with real customers all throughout the process.

##12. Innovate
###How to nurture disruptive innovation
You can keep innovating while growing. You need the 3 following keys:

####1.Scare but secure resources
Startups require less capital but they must be sure this capital won't be reduced because of budget or anything. 

####2. Independent development authority
Startups need to be able to run experiments without approvals.
Small teams should have all the people required to launch a real product, not just a prototype.

####3. A personal stake in the outcome
Entrepreneurs should have stock options or equity. Innovators should be named so the organization can give them credit by putting their name on the door or some other kind of reward. At Toyota, the **shusa**, deisgn leader, has final, absolute authority over every aspect of vehicle development.
If the boss is the one who hands out the awards with no objective criteria, teams will not be motivated to take risks.

###Creating a platform for innovation
You want to empower autonomous tartup teams, but at the same time, you should protect the parent organization, hold entrepreneurial managers accountable and reintegrate innovation back into the organization.

####Protecting the parent organization
In company X, nobody could really explain the numbers of an experiment on pricing: number of gross sales of the product, broken down by auqrter and customer segment. Lots of data to comprehend. Nobody knew which customers had been exposed to the experiment. Different teams had implemented it, different parts of the product had been updated, everything had taken 6 months, and many people had changed jobs.
You can see many problems there:
- the use of vanity metrics instead of actionable metrics
- cycle time too long
- batch size too big
- unclear growth hypothesis
- a weak experimental design
- lack of team ownership
- and therefore, little learning

Then every team interpreted the numbers as they wanted and the meeting turned into a battle of opinions.

####Rational fears
There was another issue at hand: seom guys sold B2B and others B2C, with B2B bigger but slowing. The B2C wanted to run experiments with cheaper products.
But the company was afraid of angering current customers.

####The dangers of hiding innovation inside the black box
If the company didn't innovate, they would face the **Innovator's Dilemma**: very high margins and profits until it suddenly collapsed.
Many startups want to protect themselves from the parent organization but it should be the opposite: IBM should protect itself from IBM PC in Boca Raton, Florida. Managers of IBM can feel threatened by the innovation that can suddenly jump at them from the startup.

####Creating an innovation sandbox
1. Any team can create a split-test experiment for one customer segment or territory or sandboxed part of a product
2. The team must see the whole experiment end-to-end
3. The experiment can only run a few weeks
4. The experiment can only affect X% of the company's total customer base
5. Every experiment must be evaluted on a standard report using 5 to 10 actionable metrics
6. Every team working inside the sandbox must use the same metrics to measure success
7. The team must stay ready to abort the experiment is something catastrophic happens

Example of a sandbox: the pricing page for certain stores or geographic areas. Whenever possible, there should be a clear team leader, like the **shusa**, who can build, market, deploy without approval.

##Cultivating the management portfolio
As the startup grows, new markets are conquered, the product becomes the face of the company, with implications for PR, marketing, sales. Copycats, fast followers and imitators arrive. 
The market for the new product becomes established and procedures become more routine. It becomes important to roll up line extensions, incremental upgrades and new forms of marketing.
The problem with all kinds of companies is that employees follow the product they develop as they move from phase to phase. The inventor starts managing the people who commercialize his invention. Strong creative managers end up getting stuck working on growth and product optimization instead of creating new products. And companies struggle finding creative people.

###Entrepreneur is a job title
When products move from phase to phase, they should be handed off to a new team. Employees can choose to follow the product and change teams, or work on something new. After a product has been incubated in the innovation sandbox, it should be reintegrated in the parent organization so a larger team can commercialize and scale it.
At first, the earliest experiments won't get lots of learning and scalable success, but over time they are guaranteed to improve as they get constant feedback of small-batch development and actionable metrics and are held accountable to learning milestones.
The sandbox will grow, run bigger experiments, until one day the sandbox can contain the whole product. At this point, the former innovators will become guardians of the status quo and a new sandbox will be needed.

###Becoming the status quo
It's disturbing, but innovators transform from radical outsiders to the embodiment of the status quo.
Many individuals are reticent to change: coders want to code and don't want to be interrupted by meetings, cross-functional handoffs and explanations on numbers.
Lean Startups indeed cause individuals to suboptimize for their individual functions. It does not matter how fast we can build or how fast we can measure. What matters is how fast we can get through the entire loop.
Switching to validated learning feels worse before it feels better.

##13. Waste not
The big question of our time is not "Can it be built?" but "Should it be built?"
Today we produce a lot of waste by working on the wrong things. And we work on the wrong things very efficiently.
We lose sight of the real goal of innovation: to learn what is currently unknown. What matters is not setting quantitative goals but fixing the method by which these goals are attained. The Lean Startup method aims to answer the question "How can we build a sustainable organization around a new set of products or services?"

##Organizational superpowers
###Product development pseudoscience
If Taylor were alive today, he would laugh: we ahve so much scientific knowledge but we green-light new projects with more intuition than facts.
But only by building a model of customer behavior can we establish real facts about the validity of our vision.

###A new research program
To stimulate productivity in uncertain environments, we could set up a lab where small cross-functional teams come and look at problems with clear right answers taken from international programming competitions that have developed databases of problems with clear solutions. Then they could vary the conditions of the experiments and measure results from real tests and real MVPs.
And we need to develop clear methods to hold teams accountable for validated learning. The one I proposed in this book uses a well-defined financial model and engine of growth.

##The long-term stock exchange 
Instead of pushing managers to focus on short-term profit, we could set up a stock exhange using innovation accounting, reporting on the revenue generated from products that didn't exist a few years earlier. There would be transaction costs and fees to minimize day trading and massive price swings.

##14. Join the movement
* Lean startup Meetups; Shanghai: Techyizu, Barcamp

* Lean Startup wiki: leanstartup.pbworks.com

* Lean Startup circle; leanstartupcircle.com

* Startup Lessons Learned; sllconf.com

* blog.500startups.com; see "Startup Metrics for Pirates"

* startup-marketing.com

* andrewchenblog.com

* venturehacks.com

* runningleanhq.com; helping startups become lean

* skmurphy.com/blog; early-stage software startups

* market-by-numbers.com, Brant Cooper

* vlaskovits.com, on technology, customer development and pricing

* blog.kissmetrics.com

* hitenism.com


*Required Reading*

* The Four Steps to Epiphany, Steve Blank (it's indispensable)

* The Entrepreneur's Guide to Customer Development, Brant Cooper

* The Innovator's Dilemma, The Innovator's Solution, Clayton Christensen

* Crossing the Chasm, Inside the Tornado, Dealing with Darwin: How Great 
Companies Innovate at Every Phase of ther Evolution, Geoggry Moore
* Reinertsen, The Principle of Product Development Flow

* Jeffrey Liker, The Toyota Way

* Steven Watts, The People's Tycoon

* Womack, Lean Thinking

* Robert Kanigel, The One Best Way

* Principle of Scientific Management, Taylor

* Extreme Programming Explained, Kent Beck

* Certain to Win: the Strategy of John Boyd, Chet Richards

* Out of the crisis, Edrwards Deming

* My years with General Motors, Alfred Sloan

* Bill, Alfred and General Motors, William Pelfrey

* The Practice of Management, Drucker

* Getting to Plan B, Randy Komisar

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...