The Lean Startup
by Eric Ries
American entrepreneur and author who co-founded IMVU and developed the Lean Startup methodology, which has become the dominant framework for building new ventures in Silicon Valley and beyond.
In a nutshell
Eric Ries introduces a scientific approach to building startups -- test assumptions fast, measure what matters, and pivot before you run out of money.
The Lean Startup by Eric Ries: A Complete Summary
"The only way to win is to learn faster than anyone else."
Overview
The Lean Startup (2011) introduced a methodology that changed how the world builds new businesses. Eric Ries argues that the traditional approach to startups -- write a detailed business plan, pitch it to investors, build the product, then launch -- is fundamentally broken. It assumes you know what customers want before you have tested anything. Most startups fail not because they cannot build what they planned, but because they build something nobody wants.
Ries proposes a different approach: treat the startup as a scientific experiment. Every business plan is a set of untested hypotheses. Instead of spending months or years building a product based on assumptions, build a minimum viable product (MVP), get it in front of real customers as fast as possible, measure their behavior, and learn whether your assumptions are correct. Then either continue on your current path or pivot -- make a fundamental change to your strategy based on what you have learned.
The Lean Startup methodology draws on lean manufacturing (Toyota Production System), agile software development, and Steve Blank's customer development process. It has been adopted not only by startups but by large corporations, nonprofits, and even government agencies seeking to innovate under conditions of extreme uncertainty.
About the Author
Eric Ries co-founded IMVU, a social entertainment company, in 2004. His experience at IMVU -- where the team spent months building features that customers did not want -- became the crucible for the Lean Startup methodology. Before IMVU, Ries worked at another startup called There, Inc., which raised $40 million and built a sophisticated product that nobody used. These painful experiences with wasted effort and wrong assumptions drove him to develop a more systematic, evidence-based approach to entrepreneurship. Ries studied at Yale University and now advises startups and corporations worldwide.
Validated Learning
The fundamental unit of progress for a startup is validated learning -- demonstrating empirically that a team has discovered a valuable truth about a present or future business opportunity. This is not learning in the academic sense. It is learning that is validated by real customer behavior.
Ries argues that the question "Did we build what we set out to build?" is the wrong question. The right question is: "Should we have built it at all?" Many startups execute their plans perfectly and still fail because the plan was wrong from the start.
Validated learning replaces vanity metrics (total users, page views, time on site) with actionable metrics -- data that demonstrates a clear cause-and-effect relationship and can inform decisions. If you cannot use a metric to decide what to do next, it is a vanity metric.
Build-Measure-Learn
The core of the Lean Startup methodology is the Build-Measure-Learn feedback loop:
- Build -- Create a minimum viable product that tests your most important assumption
- Measure -- Collect data on how customers actually behave (not what they say they would do)
- Learn -- Analyze the data to determine whether to persevere or pivot
The goal is to get through this loop as quickly as possible. The speed of iteration is the competitive advantage. A startup that can test, measure, and learn in one week will outperform one that takes one month, regardless of funding, talent, or market.
Importantly, the loop is described in Build-Measure-Learn order, but planned in reverse: start by deciding what you need to learn, then figure out what to measure, then determine what to build to get that measurement.
The Minimum Viable Product (MVP)
The minimum viable product is the version of a new product that allows a team to collect the maximum amount of validated learning with the least effort. It is not a prototype, a beta version, or a stripped-down product. It is the smallest experiment that tests a specific hypothesis.
Ries provides several examples:
- Dropbox used a simple video demonstrating the product concept before building it. The video drove sign-ups from 5,000 to 75,000 overnight, validating demand without writing a single line of file-syncing code.
- Zappos founder Nick Swinmurn tested whether people would buy shoes online by taking photos of shoes at local stores, posting them on a website, and buying the shoes at retail price when orders came in. The "product" was a simple website, but it validated the core hypothesis.
- Food on the Table started with a concierge MVP -- the founder personally went to customers' homes, asked about their preferences, checked local grocery sales, and created meal plans by hand. The product was the founder doing the work manually.
The MVP is psychologically difficult because it requires launching something imperfect. Ries emphasizes that if you are not embarrassed by the first version of your product, you launched too late. The purpose of the MVP is not to impress -- it is to learn.
Common MVP Pitfalls
- Overbuilding -- Including features beyond what is needed to test the hypothesis
- Confusing MVP with minimum product -- The MVP must still provide value; it is the minimum product that enables learning, not the minimum possible product
- Fear of competitors stealing the idea -- Most startups fail from a lack of customers, not from competition. Speed of learning matters more than secrecy
- Legal and brand concerns -- Large companies worry that a rough MVP will damage their brand. Ries suggests using a separate brand or limited pilot
Innovation Accounting
Traditional accounting cannot measure a startup's progress because startups operate under conditions of extreme uncertainty. Ries proposes innovation accounting -- a framework for measuring whether a startup is making real progress toward a sustainable business.
Innovation accounting works in three steps:
- Establish the baseline -- Use an MVP to measure where the startup currently stands on key metrics
- Tune the engine -- Run experiments to improve the metrics from the baseline toward the ideal
- Pivot or persevere -- If you are making progress, persevere. If not, it is time to pivot
The key distinction is between vanity metrics and actionable metrics:
- Vanity metrics: total users, total revenue, total page views. These always go up and always look good but tell you nothing about whether your business model works.
- Actionable metrics: conversion rate, revenue per customer, retention rate, referral rate. These tell you whether your product is actually solving a problem people will pay to solve.
Ries recommends using cohort analysis -- comparing the behavior of different groups of customers over time -- rather than looking at aggregate totals. A cohort approach reveals whether changes you make actually improve customer behavior.
The Pivot
A pivot is a structured course correction designed to test a new fundamental hypothesis about the product, business model, or engine of growth. It is not giving up. It is recognizing that your current strategy is not working and making a data-informed change.
Ries catalogs several types of pivots:
- Zoom-in pivot -- A single feature becomes the whole product
- Zoom-out pivot -- The whole product becomes a single feature of a larger product
- Customer segment pivot -- The product solves a real problem, but for a different customer than originally intended
- Customer need pivot -- The target customer has a problem, but not the one you thought
- Platform pivot -- Change from an application to a platform, or vice versa
- Business architecture pivot -- Switch from high-margin/low-volume to low-margin/high-volume, or vice versa
- Value capture pivot -- Change how you monetize
- Engine of growth pivot -- Switch between viral, sticky, or paid growth engines
- Channel pivot -- Change how you reach customers
- Technology pivot -- Achieve the same solution with different technology
The decision to pivot is one of the hardest in entrepreneurship. Ries recommends scheduling regular pivot-or-persevere meetings where the team reviews innovation accounting data and makes an explicit decision about direction.
Engines of Growth
Ries identifies three engines of growth -- mechanisms by which startups achieve sustainable growth:
-
The Sticky Engine -- Focuses on retention. Growth comes from keeping existing customers rather than acquiring new ones. Key metric: churn rate. If the rate of new customer acquisition exceeds the churn rate, the business grows.
-
The Viral Engine -- Growth comes from person-to-person transmission as a natural byproduct of product use. Key metric: viral coefficient. If each user brings in more than one new user, growth is exponential. Facebook and Hotmail grew through viral engines.
-
The Paid Engine -- Growth comes from paid acquisition. Key metric: customer lifetime value (LTV) must exceed customer acquisition cost (CAC). If LTV > CAC, the business can invest in growth profitably.
Most startups should focus on one engine at a time. Trying to run multiple engines simultaneously creates confusion and makes it harder to measure what is working.
Key Quotes
On learning:
"The only way to win is to learn faster than anyone else."
On MVPs:
"If you're not embarrassed by the first version of your product, you've launched too late." (Often attributed to Reid Hoffman, but central to Ries's philosophy)
On waste:
"The lesson of the MVP is that any additional work beyond what was required to start learning is waste."
On planning:
"Planning is a tool that only works in the presence of a long and stable operating history. And yet, do any of us feel that the world around us is getting more and more stable?"
On pivoting:
"A pivot is not just an exhortation to change. It is a special kind of structured change designed to test a new fundamental hypothesis."
Criticisms and Limitations
- Not all businesses can MVP -- Deep technology, pharmaceuticals, and capital-intensive industries often cannot test hypotheses with a minimum product
- Lean can become an excuse for mediocrity -- Some teams use the MVP concept to ship low-quality work indefinitely
- Overemphasis on speed -- The urgency to iterate can undermine deep thinking and long-term planning
- Survivorship bias in examples -- Dropbox and Zappos are extreme successes; most MVPs do not lead to billion-dollar companies
- Metrics obsession -- Overreliance on data can crowd out vision, intuition, and breakthrough thinking
- Pivot fatigue -- Teams that pivot too frequently lose direction and morale
Context: The Lean Startup is most applicable to businesses operating under conditions of extreme uncertainty -- where you genuinely do not know what customers want. For businesses with clear, well-understood demand, traditional planning may be more appropriate.
Summary: Key Takeaways
- Startups are experiments, not plans -- Treat every assumption as a hypothesis to be tested
- Build-Measure-Learn is the core loop -- Speed through this cycle is the primary competitive advantage
- The MVP is a learning tool, not a product -- Build the minimum needed to test your most important assumption
- Validated learning is the unit of progress -- Real data from real customers, not opinions or projections
- Vanity metrics lie -- Use actionable metrics that demonstrate cause and effect
- Pivot when the data tells you to -- Changing direction is not failure; it is intelligence
- Innovation accounting measures what matters -- Baseline, tune, pivot-or-persevere
- Focus on one engine of growth -- Sticky, viral, or paid -- pick one and master it
- Small batches beat big batches -- Ship frequently, learn continuously, waste less
- Entrepreneurship is management -- It requires discipline, measurement, and accountability, not just creativity
Discuss This Book with AI
Here are some questions to explore with Chapterly's AI tutor:
- Ries argues that if you are not embarrassed by your MVP, you launched too late. But in an era of social media backlash, could a premature launch permanently damage your brand?
- How do you decide when to pivot versus when to persevere? What frameworks or signals do you use to distinguish a temporary setback from a fundamental flaw in your strategy?
- The Lean Startup methodology was designed for conditions of extreme uncertainty. How does it apply (or not) to your current work, whether that is a startup, a corporate innovation team, or a personal project?
Read deeper with Chapterly. Discuss each chapter with an AI tutor, save the highlights that matter, and actually remember what you read — powered by spaced repetition. Start your free trial →
Topics covered:
How readers use Chapterly with this book
The Lean Startup is a methodology book whose vocabulary (MVP, pivot, validated learning, build-measure-learn) has been so completely absorbed into startup culture that most people now use the terms incorrectly. Inside Chapterly you can save the precise definitions, run them on a spaced schedule so you stop using "MVP" to mean "first version of the product" when Ries means "the minimum thing required to learn the answer to your most important question," and pull the AI tutor into the cases where the methodology genuinely does not apply (capital-intensive industries, deep tech, regulated markets) instead of treating it as a universal playbook.
Spaced-Repetition Flashcards
Tap a card to flip it. Chapterly will resurface these on the optimal day so you actually remember them.
Flashcard 1 for The Lean Startup: What is Ries's precise definition of a startup, and why does the definition matter for which advice applies? — Answer: "A startup is a human institution designed to create a new product or service under conditions of extreme uncertainty." Every word matters: institution (so the methodology applies to teams inside large companies, not just garages), new product or service (which means improving existing well-understood offerings is not the same problem), and extreme uncertainty (which is what makes traditional business planning fail). The implication is that Lean Startup advice applies wherever extreme uncertainty exists and stops applying when uncertainty is moderate or low. A McDonald's franchise is not a startup by Ries's definition; an AI startup probably is; many corporate "innovation" projects are somewhere in between.
Flashcard 2 for The Lean Startup: What is a Minimum Viable Product (MVP), and what is the most common misunderstanding of it? — Answer: An MVP is "the version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort." The common misunderstanding is treating it as "the first version of the product." Ries's MVP is a learning tool, not a product launch. The Dropbox MVP was a video explaining the concept (validated whether people wanted it before any product existed). The Zappos MVP was photographing shoes at local stores and selling them online (validated whether people would buy shoes online before any inventory or logistics). The point of an MVP is to test the riskiest assumption with the cheapest possible experiment, not to ship a small version of your real product.
Flashcard 3 for The Lean Startup: What is the Build-Measure-Learn loop, and why does Ries argue you should optimize for loop speed? — Answer: The three-step cycle: build the minimum thing needed to test a hypothesis, measure the result, learn whether to persevere or pivot. Ries argues that the speed at which you complete loops is the primary competitive advantage of a startup, because every loop produces validated learning and reduces the uncertainty that traditional competitors cannot reduce by planning alone. A startup that runs ten loops in the time a competitor runs one will discover the right product, market, or model before the competitor can finish their original plan. The mistake most startups make is optimizing for the quality of each individual step (especially Build) rather than for the speed and number of complete loops.
Flashcard 4 for The Lean Startup: What is "validated learning," and how does it differ from "we ran some tests"? — Answer: Validated learning is empirical evidence — typically behavioral data from real customers — that confirms or refutes a specific hypothesis about the business. The contrast Ries draws is with traditional product progress measures: lines of code, features shipped, opinions gathered, projected revenue. Those measures feel like progress but produce no information about whether the business model actually works. Validated learning is the unit of startup progress because it is the only kind of progress that reduces the extreme uncertainty the startup is operating under. The implication is uncomfortable: a feature you shipped that nobody uses is negative progress, not positive, because it cost time and did not produce learning.
Flashcard 5 for The Lean Startup: What is the difference between a vanity metric and an actionable metric? — Answer: Vanity metrics always go up: total users, total pageviews, total signups, total revenue. They look good in board decks but tell you nothing about whether your business model works. Actionable metrics measure cause and effect: conversion rate, retention rate, revenue per customer, viral coefficient. Ries's rule is that any metric you cannot use to make a decision is a vanity metric. The recommended technique is cohort analysis — comparing the behavior of groups of customers acquired in different periods — because cohort analysis reveals whether changes you make actually improve the customer behavior you care about, while aggregate totals hide it under the growth of the user base.
Flashcard 6 for The Lean Startup: What is a pivot in Ries's precise sense, and what is the difference between pivoting and giving up? — Answer: A pivot is "a structured course correction designed to test a new fundamental hypothesis about the product, business model, or engine of growth." It is not abandoning the company; it is keeping the team, mission, and accumulated learning while changing the specific hypothesis being tested. Ries catalogs ten types — zoom-in, zoom-out, customer segment, customer need, platform, business architecture, value capture, engine of growth, channel, and technology pivots — each defined by exactly what changes and what stays. The discipline that distinguishes pivoting from flailing is that the decision is driven by validated learning that the current hypothesis is wrong, not by a hunch that something else might be more interesting.
Flashcard 7 for The Lean Startup: What are the three engines of growth, and why does Ries argue you should focus on one? — Answer: The sticky engine (growth from retention — keep more existing customers than you lose; key metric is churn). The viral engine (growth from each user bringing in more users; key metric is viral coefficient). The paid engine (growth from spending acquisition dollars whose lifetime value exceeds their cost; key metric is LTV/CAC). Ries argues that startups should focus on one engine at a time because each engine has different metrics, different optimization moves, and different failure modes — running all three simultaneously creates confusion about which experiments to run and makes it impossible to tell which engine is actually working.
Flashcard 8 for The Lean Startup: What is the strongest critique of the Lean Startup methodology? — Answer: That it produces a generation of teams optimizing for cheap experiments and fast loops at the expense of the kind of patient, expensive, deep work that some of the most important businesses require. Pharmaceuticals cannot MVP a drug. SpaceX could not MVP a rocket. Some categories of deep technology require years of investment before any meaningful customer feedback is possible, and applying lean methodology to them produces small, incremental improvements where breakthrough work was needed. Ries himself acknowledges the methodology fits "extreme uncertainty" rather than every situation; the critique is that the cultural absorption of the framework has been less discriminating than the original argument was. A careful reader applies the methodology where uncertainty is genuinely extreme and recognizes the categories where its assumptions do not hold.
Test Your Recall
Self-quiz before you keep reading. Retrieval practice beats re-reading every time.
Q1.How does Ries argue that the build-measure-learn loop should be planned in reverse, and why does this counterintuitive approach matter?▾
Ries argues that despite the name, you should plan the loop in reverse: start with what you need to learn, then design the metrics that would measure it, then build the minimum thing required to produce those metrics. The conventional flow is Build first (what feature should we add?), Measure second (let us look at the data), Learn last (what does the data tell us?). The reversed flow forces clarity about the hypothesis before any code is written, which dramatically changes what gets built. If the learning goal is "do customers actually want this feature at all?" the minimum thing to build is a video, a landing page, or a manual concierge service — not the feature itself. If the learning goal is "can we acquire customers at a sustainable cost?" the minimum thing to build is an ad campaign and a sign-up form. By starting with the learning goal, the build phase shrinks dramatically because most of what teams normally build is not required to answer the actual question. The matter is operationally important because most failed startups built the wrong thing — not because they could not execute but because they spent months building something that had not been validated as worth building. Reverse planning is the technique that prevents this failure mode by forcing the riskiest assumption to be tested before significant build effort is committed.
Q2.What is innovation accounting, and how does it solve a problem that traditional financial accounting cannot?▾
Innovation accounting is Ries's system for measuring progress in a startup whose product, market, or business model is still being discovered, when traditional revenue and profit metrics are too small or volatile to indicate whether the business is actually working. The system has three steps. First, establish a baseline: use a minimum viable product to measure where the startup currently stands on key actionable metrics — conversion rate, retention, revenue per customer, viral coefficient. The baseline is honest about how bad current performance is, which most founders resist because it feels like failure. Second, tune the engine: run a series of experiments whose hypothesis is "this change will move metric X by Y amount," and measure whether the change actually moves the metric. Each tuning cycle should produce a measurable improvement; if it does not, the underlying business model may be wrong. Third, decide whether to pivot or persevere: if cumulative tuning is producing meaningful progress toward an ideal target, persevere; if tuning has stalled or the gap between baseline and ideal is not closing, the hypothesis is wrong and a pivot is warranted. The problem this solves is that traditional accounting is silent during the period when a startup most needs to know whether it is on track. Revenue is small, often anomalous, and dominated by founder effort rather than systematic business mechanics. Innovation accounting provides a framework for honest internal measurement during the discovery phase, when the question "is this working?" cannot be answered by the income statement. The honest version is uncomfortable because it tends to surface that the model is not yet working long before the founders are ready to confront that, which is precisely why most startups skip it.
Q3.What is the difference between a zoom-in pivot and a zoom-out pivot, and what do the other pivot types add to the catalog?▾
A zoom-in pivot is what happens when one feature inside the original product turns out to be more valuable than the whole product, and the team makes that feature the entire offering — Instagram famously zoomed in from a complex location-check-in app (Burbn) to the single photo-filter feature that users actually loved. A zoom-out pivot is the reverse: the original product turns out to be a feature inside something larger, and the team rebuilds around the larger offering. Ries's catalog of ten pivots is useful because each names a specific change to a specific element of the business model, which forces precision about what is actually being changed. A customer segment pivot keeps the product but switches which customer it serves; a customer need pivot keeps the customer but redefines the problem being solved; a platform pivot moves from being an application to being the platform other applications run on; a business architecture pivot switches between high-margin/low-volume and low-margin/high-volume; a value capture pivot changes how the business monetizes; an engine of growth pivot switches between sticky, viral, and paid growth mechanics; a channel pivot changes how the product reaches customers; a technology pivot keeps everything else but achieves the same outcome with different underlying technology. The catalog is operationally useful because it lets a team look at its current situation and ask which specific kind of pivot the validated learning is pointing toward — rather than the vague "we should change something" that most pivot discussions actually consist of. Naming the pivot type sharpens what specifically needs to change and what stays, which dramatically increases the odds of the pivot working rather than producing more drift.
Q4.In what categories does the Lean Startup methodology not apply well, and how should a reader recognize when they are in such a category?▾
The methodology assumes that the riskiest assumptions can be tested with cheap, fast experiments — that you can build a minimum thing in days, get real-customer feedback in weeks, and iterate to product-market fit within months. This assumption holds in software, consumer apps, and many internet-enabled services. It fails badly in several categories. Pharmaceuticals require years of regulated clinical trials before any meaningful customer feedback is possible; the minimum viable cancer drug does not exist. Deep technology — rockets, fusion reactors, novel chip architectures — requires sustained investment over years before the technology itself works well enough to test customer demand. Capital-intensive infrastructure (toll bridges, fiber networks, utility power plants) requires the full asset to exist before it produces any value. Heavily regulated industries (banking, defense, healthcare) often require the full compliance apparatus to exist before any product can be sold at all. The signal that you are in a category where the methodology does not fit cleanly is that the cheapest experiment to test your riskiest assumption is still very expensive or takes a very long time. In those categories, planning, capital, and specialist expertise tend to outperform iteration, and applying lean methodology produces small wrong moves at the expense of the patient deep work that the category actually rewards. Ries himself is clear that the methodology applies under "extreme uncertainty," not universally; the cultural absorption of the framework has been less careful, and a reader doing serious deep technology should treat the methodology as one tool among several rather than as a complete operating system.
Want a personalized quiz for every chapter of The Lean Startup? Try Chapterly free.
Discuss with the AI Tutor
Five passages worth thinking about, each paired with a prompt your Chapterly tutor can pick up.
"The only way to win is to learn faster than anyone else."
Prompt: Ries treats learning speed as the primary competitive advantage of a startup. Apply this to a project you are currently running. What is the actual cycle time between hypothesis and evidence in your work, and what are the bottlenecks slowing it down? If you cut your loop time in half, what would change about your odds of finding the right answer before you run out of money or attention?
Discuss this with your AI tutor"If we do not know who the customer is, we do not know what quality is."
Prompt: Ries is making a specific claim: "quality" is meaningless without a target customer because quality is defined by what that customer values. Apply this to a product or piece of work where you have been investing in "quality" without a clear picture of who you are building it for. What does your work look like if you stop optimizing for "great" in the abstract and start optimizing for great-for-this-specific-customer?
Discuss this with your AI tutor"Vanity metrics are dangerous because they appeal to our human desire to find a pattern, even when none exists."
Prompt: Audit the metrics you currently look at most often. How many of them only ever go up regardless of whether anything is actually improving? What is the actionable metric — defined by cohort, by causality, by something you could change next week — that should be replacing each vanity metric in your dashboard?
Discuss this with your AI tutor"A pivot is not just an exhortation to change. It is a special kind of structured change designed to test a new fundamental hypothesis."
Prompt: The cultural use of "pivot" has drifted from Ries's precise definition. Pick a recent organizational change you have witnessed or been part of. By Ries's standard, was it a pivot (specific hypothesis change driven by validated learning, with most of the team, mission, and learning preserved) or was it something else (panic, flailing, founder boredom, board pressure)? What does the framing reveal about whether the change is likely to produce better outcomes?
Discuss this with your AI tutor"The lesson of the MVP is that any additional work beyond what was required to start learning is waste."
Prompt: This is Ries's strongest claim and his most frequently violated one. Pick a feature, document, or polish task you are currently working on. What is the minimum version of it that would have let you learn what you need to learn? What is the difference between the minimum and what you are actually doing, and what is that difference buying you besides comfort?
Discuss this with your AI tutor