This book has now become a classical for any person trying to start a company that isn’t going to replicate an established business. It may be a product (software or hardware) or a new service. So long as what you’re going to start isn’t just replicating an established and well understood business, you should read this book. The fact that it is associated to the start up world of silicon valley is an “unfortunate" association that excludes many businesses that could benefit from multiple ideas in the book.

Eric starts by debunking the myth that start ups are about getting brilliant energetic young people to work in their garage having the right idea at the right moment and magically obtaining success. He points out that to make a start up work, one needs management. Not the traditional management that one usually finds in common large corporations. He provides his own example at IMVU, a company he founded and grew with three other co-founders. IMVU will be used largely throughout the book as a reference and example for many of Eric’s arguments. His main short point is that one has to try things ago and get feedback from customers to build something meaningful. Planning and hiding your product only increases the risk it doesn’t solve anybody’s problems.

He points out five principles for the Lean Startup method:

The rest of the book is divided in three main Parts: Vision, Steer, Accelerate.

In Vision, Eric goes into explaining the origins of the Lean Startup method in Lean Manufacturing, which originated in Japan with Taiichi Ohno and Shigeo Shingo from the Toyota Production System. Eric makes a metaphor that building a startup is a lot like driving a car in that we have to constantly adjust our direction and speed according to the feedback we get from the road. In his view, most startup plans, however, are laid out like we were going to launch a rocket out of the atmosphere: any slight mistake can destroy the whole thing. This is not the case.

He moves to talk about issues around traditional management and why it doesn’t work in a startup but that doesn’t mean large organizations cannot benefit from it. So long as one has a bit of “a human institution designed to create new products and services under conditions of extreme uncertainty", then there is a startup and there is (or should be) an entrepreneur.

Vision finishes by getting ready for the next part: identifying what it is that needs to be validated and how to effectively validate it. The answer is that we validate learning and, unfortunately, previous attempts to prove that learning have failed and left a bad reputation. Therefore we need to apply a scientific method: experimentation. Establish hypothesis, define a test, execute it, measure the results and validate or invalidate the hypothesis based on those results.

Two basic hypothesis are important:

In Steer, we move on to understanding how to go about testing assumptions and iterating. The goal is to minimize the time to go through a full loop of the Build (creates product) - Measure (obtains data) - Learn (generates ideas). To do so, we must identify what Eric calls the leap-of-faith assumptions. Those are the core assumptions that the start up makes and need to be quickly identified and validated if possible. This is the goal of the famous MVP (Minimum Viable Product). The product with the least features that can help you go through a full Build-Measure-Learn loop as quickly as possible and decide whether to persevere or pivot.

Eric moves on to talk about how to define strategies and ideas about analogs and antilogs that both provide us with previous successful examples and previous opposite behaviors. Those can help identify good arguments in favor or against your plan. He also advocates for genchi gambutsu, or “go and see for yourself". A principle from Toyota that calls for everyone on the business to understand their customer deeply.

As we continue Eric reinforces the need to try small simpler versions of your product to get feedback even if it means faking those features with humans behind it. He describes a number of techniques to build MVPs that are unexpected:

He moves on to explain that quality is a variable thing and that “if we do not know who the customer is, we do not know what quality is".

As we continue, we get to one the most interesting parts of the book: innovation accounting. Eric goes in length into what vanity metrics are and how to switch to actionable metrics. The former being metrics that don’t represent the acceleration of growth but rather just the speed of it (which may increase even though it’s really slow). The latter focusing on what is growing and how.

Eric finished the Steer part by addressing the question of whether to persevere on an idea/hypothesis or to pivot and change the approach. He defines the lifespan of a startup and the number of pivots that startup can still make. It can shrink or expand according to investments, spendings and cost of each experiment and should be measured as such. However those decisions should be made based on facts collected from tests and not emotions or feelings from team members. He goes on to describe some types of pivot that can happen:

In the last part, we look at how to accelerate. Once startups are working, they will eventually need to grow and Eric argues strongly that size is not something that would get in the way of being Lean. He starts off explaining some of the advantages that small batches present by using some manufacturing examples. Doing the same thing many many times seems intuitively more productive as we assume that we will get better at that one thing quickly. However, keeping small batches allows us to identify problems faster as we go through the whole process. It also allows us to deliver partial results faster. Imagine two tasks that need to be done 10 times each. Each task takes 6 seconds to be completed. If we complete all of task 1 and then all of task 2, we have all finished products in 2 minutes. However, if we are stopped after 1 minute, we cannot deliver anything. If we, on the other hand, complete task 1 then 2 and go back to task 1 and 2 for the next product, even if we consider that we need say 3 seconds to go back, by the end of the first minute, we have completed 4 products and are ready to produce more. Not only that, but it also takes us 6 seconds (and only one unfinished product) to find out in this case if the outcome of task 1 is compatible with task 2. While in the other strategy it takes us 1 minute and 10 unfinished (and potentially wasted) products.

In the case of a Lean Startup, the product is learning (as previously stressed). So the smaller the learning, the faster we will be able to grasp it and react accordingly. In software, this means less changes and more frequent releases. Those gated by a technical andon cord which allows the system to “stop the line" in case of identified problems.

Large batches incur more planning, more effort for each batch and longer times between batches in exchange for an economy of scale benefit. There are many scenarios in which those benefits come at even more prices. Whenever we talk about anything that perishes over time, there is an obvious cost at leaving things waiting. But even non perishables create larger stocking cost as well as larger initial investments.

With smaller batches, you can also move to a pull system which reacts to events instead of trying to predict them. Smaller stocks reduce the chance (or the quantity at least) that those stocks will become waste. As soon as an item is grabbed from the small stock, it can trigger a pull to the producing side to replenish that one single item.

Eric moves then to the engine of growth which is what he calls the strategy for the startup to grow. He lists 4 common growth powers:

From those, he defined 3 engines of growth:

Although most products and services will likely have a bit of all 3 engines of growth, focusing on a single one allows for a startup to make more conscious decisions about what to validate and which experiments to create. It will also determine the product/market fit since it’ll hint which communities to reach to and how.

Eric follows by explaining that engines of growth, for as good as they are, will eventually run out of power. There are only so many early adopter customers we can reach in a network, only so long that we can hold a customer around and only so much revenue we can make out of a lifetime customer. At some point a company needs to transition to the mainstream customers. And those are different to please than early adopters. Startups should recognize that and be prepared to shift gears and priorities accordingly.

Adapting is the next focus going into why it is important to adapt to the situation the company faces. And, in order to adapt, a company must have a certain level of quality built into the product. Quality is something that changes over time depending on the users of the product/service but not being careful about it can hinder a startups’ ability to react to new findings. One powerful tool to allow a deeper quality and adaptation is known as The Five Whys. It’s a simple process in which for every identified problem, we ask ‘why did this happen?’, write the answer down and then reply to the answer with another ‘why?’ question. Doing so 5 times usually takes us closer to the real root cause of the issue and provides a path to invest in quality spending most efforts at the root of the issue and less at the surfacing one in order to get the best return on investment. However, Eric cautions the readers about the common miss-use of this tool to find “guilty people" for the observed issues. If blame gets into the exercise, the company rapidly develops at culture of mistrust and hiding issues which directly and immediately hinders the company’s ability to react.

Finally, the books closes by address the point of maintaining innovation in mature or large organizations. His suggestions is mainly to recreate a third party startup environment within a company. Secure funds, be granted autonomy and ability to reach out to their needs and demand results. He also cautions people about the problems that this solution causes upon re-absortion of the startup in the larger organization. He suggests than an Innovation Sandbox that allows for any team to run experiments and obtain actionable metrics to drive their development. He also mentions that some people will stick with the product and some people will stick with experimenting and this is normal and should be allowed.

In the epilogue, Eric talks about evolving our companies and products around the scientific method as a way to reduce waste in general and allow us to find a sustainable industry development that better uses our cognitive abilities and natural resources.

Overall, the book is more valuable than the summaries most people give of it. The one word that the book made widely common is ‘pivot’ which I don’t believe should be nearly as important as other good ideas in the book. The Build-Measure-Learn loop is a nice adaptation of Deming’s PDCA. The idea that a startup lies on two fundamental hypothesis (value and growth) is an awesome basis to get a company started as well. I would definitively recommend the book to anyone thinking about starting a business that wants to suggest a new way of doing things or creating a new product.