The Lean Startup Method Explained: Build, Measure, Learn

The Lean Startup is a way of running a new venture that treats the early company as an experiment. Instead of writing a long plan and then building to it, you state what you believe, test it cheaply with real customers, and let the results decide what happens next. The method was popularised by Eric Ries, and its official site lists five principles that hold the whole thing together.

The core claim is simple. A new company's biggest risk usually isn't building something badly. It's building something nobody wants. So the method organises work around learning that answer as fast and as cheaply as possible.

This article walks through the loop, the metrics, the pivot decision, where the ideas came from, and where the method stops being a good fit. If you're still choosing a problem to work on, start with startup idea sources and the idea stage first.

The five principles

The Lean Startup principles page names five ideas:

1. Entrepreneurs are everywhere. The site's line is that you don't have to work in a garage to be in a startup. A team inside a large company can be one too.
2. Entrepreneurship is management. A startup is an institution, not just a product, so it needs a new kind of management geared to its context.
3. Validated learning. Startups exist, in the site's words, not to make stuff, make money, or serve customers, but to learn how to build a sustainable business.
4. Build-Measure-Learn. The fundamental activity of a startup is to turn ideas into products, measure how customers respond, and then learn whether to pivot or persevere.
5. Innovation accounting. A way to measure progress, set milestones, and prioritise work when you have no history to compare against.

"Validated learning" is the term people trip over. It means a lesson backed by evidence from real customer behavior, not by opinion or by what the team hoped was true. A founder who says "I think people want this" has an opinion. A founder who says "12 of 15 buyers we called reordered without prompting" has a validated lesson.

The build-measure-learn loop

The loop is the engine of the method. You don't run it once. You run it again and again, and the goal is to shrink the time it takes to go around.

Notice that the loop starts with the idea, but it's planned backwards. You decide what you need to learn, then work out which measurement would show it, then build only enough to get that measurement. That's why the principles page describes the MVP as the thing that begins the process of learning as quickly as possible, not as a small version of the finished product.

The minimum viable product can be a working prototype, but it can also be a landing page, a video, or a service you deliver by hand while the software doesn't exist yet. If you want to pick what to test first, the riskiest assumption test helps you aim at the belief that would sink the business if it were wrong.

Pivot or persevere

After each pass through the loop, the team makes one of two calls. The principles page says that when measurement shows the business model isn't advancing, it's a sign that it's time to pivot, a structural course correction. The alternative is to persevere: the numbers are moving the right way, so you keep going.

A pivot changes one major part of the plan while keeping what you've learned. You might keep the customer but change the product, or keep the technology but aim it at a different group. It is not a cosmetic rebrand, and it isn't the same as quitting. Our guide to what a pivot is covers the main types.

The hard part isn't the vocabulary. It's honesty. Teams tend to read ambiguous data as good news, and they keep going because stopping feels like failure. Writing down in advance what result would count as "working" makes the call less emotional.

Innovation accounting and actionable metrics

Ordinary accounting measures a business that already exists. A startup has no stable revenue base, so Ries's fourth principle asks for a different scorekeeping system, one focused on whether the learning is actually happening.

The idea that carries most of the weight is the difference between vanity metrics and actionable ones. In a post on vanity metrics, Ries criticises numbers like raw hits because they count a technical process rather than a number of human beings. He also argues that when such numbers go up, teams credit their own work, and when they fall, they blame someone else, which lets each person live in a private reality.

An actionable metric is one that tells you what caused a change and how to get more of it. The comparison below is our own framing of that idea, not a quote.

The pattern is cohorts and rates. Look at groups of customers who started at the same time, and compare them against each other. If the new onboarding flow really helped, the cohort that got it should behave differently from the one before.

Key Facts: The Lean Startup Method

• The Lean Startup site names five principles: entrepreneurs are everywhere, entrepreneurship is management, validated learning, build-measure-learn, and innovation accounting (source).
• It describes the startup's fundamental activity as turning ideas into products, measuring how customers respond, and learning whether to pivot or persevere (source).
• It defines an MVP as the thing that begins the process of learning as quickly as possible (source).
• Steve Blank defines a startup as an organization formed to search for a repeatable and scalable business model.
• Blank describes a Lean Startup as the intersection of Customer Development, Agile Development and, if available, open platforms and open source (source).
• Ries argues that vanity metrics count a technical process rather than people, and that actionable metrics show what caused a change (source).

Where the ideas came from

The Lean Startup didn't appear from nowhere. Two older streams feed it.

The first is customer development, from Steve Blank. His core claim is that a startup is a search for a business model, not a smaller copy of a large company. In his first-principles essay, Blank says founders begin with hypotheses about the whole model: who the customers are, how the product reaches them, and how it's priced. Customer development is how they test those guesses quickly, by getting out and talking to people. If you want to practise that part, the article on customer discovery covers how to run the conversations.

The same essay says a Lean Startup sits at the intersection of customer development and agile development. Agile is the engineering half: build in small increments so you can change direction as you learn.

The second stream is lean manufacturing. Blank compares the method to the way Toyota's lean production system changed how factories work, and the Lean Startup book's own page notes that lean manufacturing ideas such as small batches apply to startups too. For the factory side of that lineage, including waste and flow, read lean methodology. The idea carried over is that big batches hide problems. Shipping a small experiment exposes them sooner.

How it relates to agile and design thinking

People often blur these three, so it helps to separate the questions each one answers. This is our own summary, not a claim from the sources above.

They work together. A team might use design thinking to understand users, the Lean Startup loop to test whether the idea has demand, and agile to build the parts that survive. The Lean Startup is the one most focused on the business model itself, and on whether the whole thing can become a company.

The method in practice

A small walk-through, clearly hypothetical, shows how the loop feels.

Imagine a two-person team that believes independent bakeries lose sales because they can't take pre-orders online. Their riskiest belief is that bakery owners will pay to fix this. Instead of building an app, they set up a simple order form and a shared spreadsheet, and offer it free to three bakeries for two weeks.

They measure how many customers each bakery actually gets to use the form, and whether the owners ask to keep it. If owners shrug, the team has learned that cheaply. If they ask what it'll cost, the next experiment is a price test. Each round removes one uncertainty before more money goes in.

Teams that measure in this way also tend to find the problem-solution fit question coming up early: is the problem real, and does this particular fix address it? Later comes the harder question of whether enough customers want it at a price that works, which is where product-market fit enters.

Criticisms and limits

The method has real strengths, but it isn't a fit everywhere. Some of the strongest objections come from regulated industries.

Writing in VentureBeat, consultant Iliya Rybchin argued that in healthcare and financial services the Lean Startup approach can not only be wrong but dangerous. His case is that there is no "M" in a healthcare MVP, since a bad outcome for a patient costs far more than a lost customer, and that a financial product can't skip risk management, security, compliance and licensing. That's one practitioner's opinion, not a measured finding, but it names a real constraint: if a failed experiment can hurt people or break the law, "ship it and see" isn't available.

Related limits worth knowing, drawn from the same logic rather than from a study:

• Long build times. Hardware, biotech, and other deep-tech ventures can't produce a cheap prototype in a week. The loop still applies in spirit, with simulations, pilots and customer interviews standing in for a full build.
• Misread signals. Early adopters may love a product that the mainstream never will. A small test shows interest, not market size.
• Metric theatre. Teams can pick convenient numbers and call the result learning. The vanity-metric warning applies to experiments too.
• Minimum can mean too little. An MVP that's too thin can fail for reasons unrelated to the idea, such as poor quality, which looks like a demand problem and leads to the wrong pivot.

The takeaway isn't to drop the method. It's to adapt it: keep the habit of naming assumptions and testing them, and replace the "launch something rough" step with whatever form of evidence your industry allows.

Where to start

If you're applying this on your own venture, a short sequence works:

1. Write your three biggest assumptions about customer, problem, and willingness to pay.
2. Pick the one that would hurt most if wrong and design the cheapest test for it.
3. Decide in advance what result means "keep going" and what means "change something."
4. Run the test, record the result, and make the call.
5. Repeat, and track how long each loop takes.

Steve Blank's Harvard Business Review article, "Why the Lean Start-Up Changes Everything" (May 2013), presents the method as a faster alternative to the traditional route of writing a business plan, pitching investors and then launching. Much of the article sits behind a paywall, so this page relies on the open sources above for details.

To see where this fits in a company's life, the overview of the stages of a startup is a good map.

Related reading

• Minimum viable product
• Customer discovery
• Riskiest assumption test
• Validated learning
• Problem-solution fit
• Idea stage of a startup
• Lean methodology