The Lean Startup

Introduction

• Lean Startup is the application of lean thinking to the process of innovation.
• The five principles of the Lean Startup:
  • Entrepreneurs are everywhere: The Lean Startup method works in companies of all sizes
  • Entrepreneurship is management: Startups require a new kind of management geared to handle extreme uncertainty
  • Validated learning: Startups exist to learn how to build a sustainable business
  • Build-Measure-Learn: The fundamental activity of a startup is to turn ideas into products, measure customer response and pivot/persevere
  • Innovation accounting: Focus on the boring stuff: How to measure progress, how to set up milestones, how to prioritize work
• Startups fail because entrepreneurs use an all or nothing approach to management. They either heavily depend on traditional planning and forecasting methods. Startups neither have a stable history, nor a relatively static environment. Or they adopt a "Just do it" approach and have no plan in place.

Start

• Lean Startup's unit of progress is Validated Learning.
• An entrepreneurship theory should focus on providing a method of measuring progress in the context of extreme uncertainty. It should allow entrepreneurs to make testable predictions.
• Cross-functional teams, with learning milestones, perform better in a startup, than departments with specialists who may measure efficiency as working in large batches and passing work between departments.
• It's easy to build things that nobody wants, so it doesn't matter if you ship on time and on budget. The goal of a startup is to figure out the right thing to build.
• A startup is driven by a feedback loop that allows it change tack and adjust according to changing circumstances easily. Failure to track, or adjust, against incoming feedback results in "achieving failure" successfully.
• Lean startup advocates a Build-Measure-Learn feedback loop, to know when to pivot and when to persevere.
• While using the feedback loop, you move towards your true north - the outcomes you are aiming for. The destination is a vision, strategy is employed to achieve it, and the product is the output.

Define

• A startup is not just about a product, a technological breakthrough, or even a brilliant idea. It is a human enterprise.
• Startups deal with extreme uncertainty. A business that is replicating an existing business model is not a startup.
• Even an established corporate can instill a culture of experimentation. Instead of rallying behind one bet, teams would test hundreds of bets together and choose whichever delivers on its promise.
• A platform to conduct experiments is the missing link in getting products to a faster Product-Market fit.

Learn

• Learning is the oldest excuse in the book for a failed execution. It's what managers fall back on when they fail to achieve the results we promised. Learning does nothing for the different stakeholders involved in a startup: early employees, investors, organizations that depend on innovation.
• Yet, Learning is the most important task of a startup, to find what parts of the product vision are working and which ones are to be discarded. Validated Learning is the process of empirically establishing that the startup's prospects are real. It is more concrete, more accurate, and faster than market forecasting or classical business planning.
• Quantitative targets (revenue targets) created the motivation to engage in qualitative inquiry (understanding value proposition).
• Which activities are value-creating and which ones are wasteful? is the typical mandate that a lean manufacturing program would adapt. It doesn't work in the world of a startup though, because what provides value to the customer is not clear.
• The effort that is not absolutely necessary for learning what customers want can be eliminated. It's easy to assume what your customers want, and it is also easy to learn something completely irrelevant. Validated Learning is backed up by empirical data collected from real customers.
• Progress in a startup should be in terms of how much validated learning is happening, instead of in terms of how much stuff was built. True startup productivity is systematically figuring out the right things to build.
• Products with zero customers, zero revenue, and zero traction tend to be valued better than those with small numbers. "Zero" numbers are an invitation to the imagination, while small numbers are real data. The Audacity of Zero is real.
• Faced with the dilemma of small numbers, startups tend to go back to vanity metrics or waste dollars on marketing gimmicks, just to juice up numbers.
• Every startup should be seen as a grand experiment. The question is not "Can this product be built?". A better question is "Should this product be built?".

Experiment

• Thing big, start small. What is the basic minimum work one can do to discover if a product is valuable? What is the avenue to test this assumption? What other assumptions need to be parked on the side while this one assumption is tested?
• Zappos implemented a manual workflow for servicing shoe orders, but it was testing the assumption whether customers would even buy shoes online, and if they do, what do they search for, what their preferences are, and if they are willing to pay. #FFW
• Experiment immediately. No amount of planning can replace real customer feedback. What is the shortest possible route to get a product to customers? What assumptions does this initial product test?
• The Value Hypothesis tests whether a product or service really delivers value to customers once they are using it. Surveys would not be accurate because people have a hard time assessing their feelings objectively. Actual Customer Behavior is the only certain feedback to assess value.
• The Growth Hypothesis tests how new customers will discover a product or service. This involves testing the assumptions around how to recruit early adopters and how they will share the word with their cohorts.
• An Experiment is a Product. Even when experimenting, startups enlist early adopters, add employees to each further experiment or iteration and eventually start building the product. By the time the product is ready, it has a base of established customers.
• Kodak Gallery used this line of thought process to release a basic version of the online product, lacking all bells and whistles. Though it resulted in dissatisfied customers and a ton of feedback, the team was able to validate that the value was real and they had prioritized the right features on the roadmap. #FFW
• Success is not delivering a feature; success is learning how to solve the customer's problem.

Steer

• Startups build experiments as products, and learning how to build a sustainable business is the outcome of the experiment.
• No individual part of Build-Measure-Learn is valuable by itself. What matters is minimising the total time taken for one feedback loop. Learning where and when to invest effort saves money.
• MVP is that version with a minimum amount of effort and the lease amount of development time.
• Our planning runs in reverse to Build-Measure-Learn. First, we figure out what to learn, then we identify what we need to measure to be able to learn, and then we build the feature to run the experiment and get that measurement.

Leap

• There is no one rule on what startups should do in the beginning. They need to conduct experiments and figure out what works for their unique circumstances. For a startup, the role of strategy is to help figure out the right questions to ask.
• Leap of faith assumptions are the riskiest assumptions that can break the startup easily. The entrepreneurs first goal is to build a company that can test these assumptions. Second is to perform that rigorous testing without losing sight of the company's vision.
• Leap of faith assumptions usually come in two flavors: Value hypothesis and growth hypothesis. Both need to be tested to ensure success.
• Use analogs and antilogs to plot strategy. Analogs show similarities while antilogs show opposites, from which a set of unique, unanswered questions arise.
• Successes were made possible by entrepreneurs willing to test out what parts of their plan work, what don't and adapt their strategies accordingly.
• Unlike established companies that know their customer base and can discover what customers want, entrepreneurs and startups deal with a higher degree of uncertainty. They can only understand what assumptions need to be tested early.
• Hypotheses can only be validated with real customers, in both B2B as well as B2C businesses. It's important for startups to get out of the building and go meet customers.
• Conversations with customers don't reveal specific features, but help the startup understand the potential customer and their problems, helping create a Customer Archetype. This customer profile should be considered provisional until it is proven by validated learning that the customer can be served in a sustainable manner.

Test

• An MVP helps a startup start the process of learning as early as possible, but it is not necessarily the smallest product imaginable. It is simply the fastest way to get through the Build-Measure-Learn loop with a minimum amount of effort. The goal of MVP is to begin the process of learning, not end it. It is not designed to answer product design or technical questions. Its goal is to test fundamental business hypotheses.
• A polished product actually turns away early adopters. When in doubt about the feature set, simplify. Any additional work beyond the minimum required to test assumptions is a waste of work.
• There are many ways to reduce the amount of work to be done upfront, like the video MVP, concierge MVP, and the Wizard of Oz MVP.
• MVPs can appear to be of low quality because they are meant to only test hypotheses. It is better to take the blame for low quality than to assume what is good quality for customers. Sometimes we get confirmation that our beliefs around quality were correct, and othe times, we are pleasantly surprised.
• A startup can only survive if it learns faster than anyone else. It's greatest strength is the obscurity and small size that doesn't attract any attention in the beginning.
• MVPs often result in bad news. It's good to expect this, persist through the testing and pivot, rather than discover and fail completely at the end. The MVP is just the first step in the journey, and one should commit to continuing the process irrespective of what the MVP brings up.

Measure

• A startup has two jobs:
  • Measure where it is right now, taking full cognizance of reality
  • Conduct experiments to discover how to move the real numbers closer to the ideal
• Startups need to track their progress quantitatively to ensure that they are making positive progress:
  • Use an MVP to establish a baseline
  • Tune the engine from the baseline toward the ideal
  • Pivot or persevere, based on the results
• To establish a baseline, a startup can build one wholesome MVP to most of the its assumptions. Or it may conduct different MVPs to test each of its assumptions. Either way, an MVP should test for the riskiest assumptions.
• To tune the engine, each experiment should be measured to check if it moves the startup towards the ideal. If it doesn't, the experiment should be deemed as a failure.
• Cohort Analysis helps breakdown the gross numbers and quantitatively measure each step of the conversion process. Poor quantitative results force startups to face the failure and create the motivation, context, space for more quantitative research.
• Optimizing the wrong thing won't lead to improved results in a startup. Quantitatively measuring impact of each change will help hold the mirror to stakeholders, instead of laying simplistic blame on a specific department.
• Vanity Metrics mislead the startup, for example, by focusing on gross numbers and statistics. We need to switch to Actionable Metrics to be able to gather quantitatively useful data.
• Vanity Metrics destroy the value created by following an Agile methodology.
• A team following Agile methodology is not agile, if it doesn't measure what it builds, and uses quantitative data to measure/learn/pivot/persevere.
• Conducting split-tests on Cohorts would clearly indicate any positive improvement caused by a change.
• Unless confirmed through Validated Learning, user stories should not be marked as complete. Validated is defined as "knowing whether the story was a good idea to have been done in the first place." So each story can be present in one of four states: Backlog, Actively being built, Done, and Validation in progress.
• Kanban is a good fit for such a process because each stage can be restricted to specific number of stories. If the validation fails, the feature is removed from the system.
• Including a Validation exercise from the start as part of a user story makes it easier to get through the validation phase. Teams automatically start measuring their productivity according to Validated Learning instead of in terms of production of new features.
• To be considered a good metric, it should be Actionable, Accessible, and Auditable.
• Actionable: For a report to be considered Actionable, it must demonstrate clear cause and effort. But finding out what is going on is extremely costly, so people tend to skip the validation part and move on, attributing improvements to their own judgement.
• Accessible: Validation reports should not only be accessible to everybody on the team, but also easily understandable. Features are only considered successful based on metrics, and all features should be associated with expected numbers without fail.
• Auditable: All data gathered as part of validation should be auditable, and people should be able to trust them. So reporting systems should be better organized to be able to dig into data reports. Metrics are people, too. One should be able to talk to real customers to verify data.

Pivot (or Persevere)

• The Lean Startup does not offer a rigid pivot or persevere formula, even though it is based on scientific methodology. Such decisions are to be made combining our intuition along with insightful data.
• Innovation Accounting leads to faster Pivots. The quantitative numbers are confirmation that a pivot is needed, and also provide insight into whether the pivot worked, paving the way for future pivots.
• A startup's runway is the number of pivots it can make. In other words, the shorter your pivot, the more the number of pivots you can make. Work on making your pivots shorter.
• Pivots require courage because:
  • Vanity metrics make it difficult for people to truly want to pivot.
  • Without a clear hypothesis to test, it is difficult to know whether the startup is succeeding, or failing.
  • Many entrepreneurs are afraid of failure.
• Every startup should have a Pivot or Persevere meeting at regular intervals to get together and discuss whether the startup should undergo a pivot.
• A Pivot is not just change - it is a special kind of change designed to test a new fundamental hypothesis about the product, business model, and engine of growth.
• There are different kinds of Pivots:
  • Zoom-in Pivot: What was considered a single feature in a product becomes the whole product.
  • Zoom-out Pivot: The whole products becomes a single feature of a much larger product.
  • Customer Segment Pivot: The product is changed to serve a new segment of customers.
  • Customer Need Pivot: The product is changed to address other aspects that provide more value to the customer.
  • Platform Pivot: The product starts as an application, but becomes a platform.
  • Business Architecture Pivot: The product pivots from a high volume, low margin architecture to a low volume, high margin architecture, or vice versa.
  • Value Capture Pivot: The product changes how the value is captured.
  • Engine of Growth Pivot: The product switches between the engines of growth: viral, sticky, or paid.
  • Channel Pivot: The product changes the channel of delivery for greater effectiveness.
  • Technology Pivot: The product switches to a new technology for cost savings or better value.
• A Pivot is a Strategic Hypothesis. It leads to a test of the fundamental hypothesis related to the product. Startups should be careful about taking advice on pivots, but they have a guard rail in the form of Lean Startup methodology to test the hypothesis.

Batch

• Work in small batches, though counter-intuitive, can be more efficient than large batches. Also, mistakes are caught early in the process and are fixed permanently for the rest of the process.
• Continuous Deployment can be leveraged to catch mistakes quickly, be remedy and push them to customers, and also to produce and deploy changes in small batches.
• By reducing the batch size, one can shrink the Build-Measure-Learn feedback loop, which leads to learning faster.
• There is an inherent dichotomy in companies that follow lean manufacturing practices - their design practices are stuck in the past churning out their output in large batch sizes.
• It is possible to implement lean manufacturing principles even in the hardware space - by moving hardware into software, by allowing faster production changes, and with the help of 3D and rapid prototyping tools.
• People work in large batches assuming that the process gives the highest efficiency possible. But there are always questions, changes, and feedback that needs to be incorporated back into the work. So the interruptions tend to make the large batch work inefficient and also cause disruptions in future batches.
• Large batches tend to grow over time. Because moving the batch forward means handling all interruptions, rework, and delays, everyone has an incentive to do work in ever-larger batches, trying to minimize this overhead. The batch size continues to grow in the absence of any physical limits, leading to a Large Batch Death Spiral. Eventually, one batch will become the highest-priority project, on which the future of the entire company depends on.
• Lean production solves the problem of excess inventory by using a Pull mechanism instead of a Push mechanism. Necessary Parts are manufactured JIT as and when the inventory is used up by customers. The ideal goal is to achieve single-piece flow along the entire supply chain.
• Applying this JIT concept to products does not mean that customer demands have to be met JIT. Customers rarely know what they want. It means that our Hypothesis about the customer acts as the pull factor for new product development. Only those features that test the hypothesis make it to development, and everything else is ignored.
• Lean Startup techniques provide the foundation to build a startup, but it is also important to build a learning organization to encourage learning, creativity, and innovation.

Grow

• The eventual goal of Startups is to get to Sustainable Growth. Sustainable translates to a mechanism that will power the engine of growth for the startup, outside one-time activities like advertising, discounts, or a publicity stunt.
• Sustainable growth is characterized by one simple rule: New customers come from the actions of past customers:
  • Word of mouth
  • As a side effect of product usage
  • Through funded advertising
  • Through repeat purchase or use
• A Startup should invest heavily in experiments to improve metrics surrounding the Engine of Growth. Other growth aspects can be safely ignored as long as these metrics are showing positive improvements.
• The Sticky Engine of Growth is one where customers come back repeatedly to use the product once they start. The most important metric to measure in this engine of growth is the Churn Rate or Attrition Rate. If the rate of new customer acquisition exceeds the churn rate, the product will grow.
• The Viral Engine of Growth is when growth happens as a side-effect of customers using the product. It's when customer-to-customer transmission happens as a necessary consequence of normal product use. A startup depending on viral growth should focus on the Viral Coefficient more than anything else, to power its growth. If the Viral Coefficient is greater than 1, then the startup will grow exponentially.
• The Paid Engine of Growth is powered by Customer Lifetime Value, the revenue earned from each customer after deducting variable costs.
• A startup can use more than one Engine of Growth as its revenue model, but it is suggested that it concentrate on one model at a time. Only after pursuing one engine thoroughly should a startup consider a pivot to another.
• The Engine of Growth determines how and what to do to achieve a Product/Market Fit. Depending on the engine, a startup can come up with concrete metrics to understand if they have a product/market fit. For the Paid Engine of Growth, it could be when the Lifetime Value is higher than the acquisition cost of a new customer. For the Viral Engine of Growth, it could be when the Viral Coefficient > 1.
• Every Engine of Growth of growth eventually runs out. A startup should not only work hard to monetize and grow its Engine of Growth but it should constantly explore new avenues of growth.

Adapt

• Adaptive Organization: An organization that automatically adjusts its process and performance to current conditions.
• While most of startup advice is to run fast and get products as quickly as possible into customer's hands, there is also a need to slow down and build quality into the product. If the product that reaches mainstream customers suffers from low quality or bugs, it prevents learning and experimentation.
• The Five Whys lays down a framework whose core idea is to link investments to the prevention of the most problematic symptoms. At the root of every seemingly technical problem is a human problem.
• Asking the Five Whys helps us get to the root of the problem, often hidden by symptoms.
• The problems uncovered by the Five Whys inquiry process should receive proportional investment. One should never try to solve all problems at the same time. If the problem is a minor glitch, spending an hour on it may be enough instead of constructing an 8-week preventive work schedule. Over time, such small changes compound and produce exponential benefits, without having had to invest dedicated time and energy for extended periods of time.
• The Five Whys acts as an automatic speed regulator because startup teams invest in solutions for problems as they occur. When there are many problems, the team slows down to answer the Five Whys and go about fixing the core issue. They speed up as the investments compound and produce ever-improving quality in the product.
• The Five Whys process can sometimes be used by teams to place blame, instead of fix issues. The best antidote to this is to ensure everybody involved with the problem or the symptom is present in the room when the discussion takes place. The mantra is if a mistake happens, shape on us for making it so easy to make that mistake.
• To get started on the Five Whys, a team can follow these simple rules:
  • Be tolerant of all mistakes the first time.
  • Never allow the same mistake to happen again.
• Get buy-in from executive management to truly imbibe the true spirit of Five Whys into the team. The process can bring up unpleasant truths and the team should be aligned in order to go about solving the issues.
• To start with, choose specific problems on which to experiment the process of Five Whys. Over time, as the team gains experience, expand it into all parts of the system. Never start by attacking a large or complex problem, before the team has imbibed the learnings.
• Appoint a Five Whys master who is senior enough to take complete accountability for the process and who can decide whether the time and effort investments are paying off.
• The proportional investment (the fixes) that come out of a Five Whys sessions are valuable, but the learnings can be more formative for the team to grow individually and collectively.
• Achieving Failure - Successfully executing a flawed plan.
• Lean Startup methodologies allow startups to grow over time, without sacrificing the core principles of the Build-Measure-Learn feedback loop. The most important processes, like Five Whys, Build-Measure-Learn loop, and small batches, scale well as the startup grows bigger.

Innovate

• Startup teams require three structural attributes:
  • Scarce but secure resources
  • Independent authority to develop the business
  • Personal stake in the outcome
• Large budgets are as harmful to startups as too small budgets. Also, startup teams have to be protected from budget tampering.
• Startup teams should have complete authority on the development and experimentation process and should not need any approvals from the patent organization
• When monetary or recognition-based incentives are made available (and can be tracked accurately), startup teams tend to be more innovative.
• An Innovation Sandbox:
  • Any team can create a true split-test experiment that affects only the sandboxed parts of the product or service or only certain customer segments or territories.
  • One team must see through the whole experiment end to end.
  • No experiment can run longer than the specified amount of time.
  • No experiment can affect more than a specific number of customers.
  • Every experiment has to be evaluated on the basis of a single standard report of five to ten actionable metrics.
  • Every team that works inside the sandbox and every product that is built must use the same metrics to evaluate success.
  • Any team that creates an experiment must monitor the metrics and customer reactions while the experiment is in progress and abort if something catastrophic happens.
• Entrepreneurship is a viable career path. People can choose to move out of a product once it has been established and properly transitioned.
• The Lean Startup is a framework and not a set of rules. It has to be evaluated in every situation and adapted as necessary to the problem.
• Switching to validated learning produces more pain in the interim before the team starts feeling better and gets aligned with the thought process.

Epilogue: Waste Not

• The Lean Startup advocates that most form of waste can be eliminated from startups once their cause has been understood.
• In New Product Development, Entrepreneurship and Innovation still use outdated frameworks. New Projects are sanctioned purely based on intuition than facts.
• Teams engage in pseudoscience when they use:
  • Gross metrics to depict growth
  • Use learning as an excuse to failure
  • Create a data-free zone for unlimited "experimentation" that is devoid of customer feedback
• A scientific approach is necessary to investigate, fact-find and analyze the work that is done in typical modern workforces. A knowledge worker adopting the same scientific mentality of Frederick Taylor and others can discover hidden gems in just analyzing his or her own work environment.