Lean Startup Methodology: Test Before You Build

Lean Startup is a way to test a business idea before committing to a full product. You identify a risky assumption, run the smallest useful experiment, observe what customers actually do, and decide what to change. Eric Ries calls this the Build–Measure–Learn loop.

For a founder, the useful question is specific: What do I need to learn before I spend another month building? Below, we’ll turn an invoice-reminder app idea into one testable assumption. You can use the same process with the free experiment card.

How Build–Measure–Learn works

Ries’s methodology connects an experiment to a business decision. “Build” might mean a working feature, but it can also mean delivering a service manually. The right test depends on what you need to learn.

This is validated learning: evidence that addresses a business risk. Shipping ten features tells you how much you built. Seeing eligible customers return for a second use tells you something about the usefulness of that experience. Neither, by itself, proves a viable business.

A Lean Startup example: test the workflow before the app

Suppose you want to build an app that helps freelancers chase overdue invoices. An AI-powered dashboard would take time to build. First, you need to know whether the reminder workflow is useful enough for people to request it again.

Before recruiting, set a time budget. Record how many participants had a genuine opportunity to repeat, how many requested it, and how much support each use took. A missed repeat request means little if the second invoice was already paid.

Don’t turn an ambiguous result into “the market rejected the idea.” This test concerns one workflow, one group, and one repeat-use assumption. Payment, acquisition cost and retention need their own evidence.

If you want to organize these notes in Notion, the two-database setup shows how to link sources to a planned test.

What counts as a minimum viable product?

An MVP is a version of a product or service built to test a specific hypothesis with as little effort as useful learning requires. In Ries’s explanation, the hypothesis determines the work. “Minimum” does not mean careless design or a broken experience.

Not every early question requires an MVP. Choose a test that fits the question:

• Unsure whether the problem matters? Investigate recent real examples and the customer’s existing workaround before making a product.
• Unsure whether people can use your proposed flow? Observe them trying a focused prototype. That tests usability; it does not establish purchase demand.
• Unsure whether the service is worth repeating? Deliver it manually, as in the example above.
• Unsure whether an offer attracts the right people? Try a clear landing page with an honest next step. Waitlist signups measure interest, not payment.

Use a production product when the question requires one—for example, whether customers keep using a feature in their real workflow. Keep the scope narrow enough to interpret the result.

The five Lean Startup principles

Ries names five principles. Here is what each changes in a founder’s day-to-day work:

1. Entrepreneurs are everywhere

Work under uncertainty can happen inside an established company or a new venture.

2. Entrepreneurship is management

Give experiments an owner, a budget and a date to review the evidence.

3. Validated learning

Measure evidence about business assumptions, rather than work completed.

4. Innovation accounting

Choose actionable measures suited to the experiment and business stage.

5. Build–Measure–Learn

Use customer response to decide whether to continue or change course.

When the method helps—and where it falls short

Lean Startup is useful when a mistaken assumption could consume months of work and you can test it responsibly at a smaller scale. It can make the next decision clearer. It cannot remove uncertainty or guarantee success.

Early volunteers may differ from future paying customers. Small samples can reveal a problem without estimating its prevalence. A short experiment may miss long sales cycles or long-term retention. In safety-critical or regulated products, a “small test” still needs the appropriate safeguards and approvals.

Problem–solution fit, product–market fit and growth can be useful labels for different questions. They are not three compulsory steps that every company completes once, in order. A growing business can still discover that a new segment has a different problem.

Use AI to sharpen the test, not supply the result

An AI assistant can help turn a vague idea into assumptions or organize research notes you are allowed to use. It cannot tell you what actual customers did unless you provide that evidence. Simulated interviews are not customer research.

Try this prompt with a non-sensitive description of your idea. Replace the bracketed fields before sending it to your preferred assistant:

My proposed customer is [specific group].
Their current workaround is [what they do today].
My idea is [one sentence].
My time and budget limits are [limits].

List three assumptions this idea depends on. For each, suggest one observable customer behavior that would challenge it. Then propose the smallest honest test for the riskiest assumption.

Separate facts I supplied from your guesses. Do not invent interviews, results, benchmarks or customer quotes. Ask me to choose a decision rule before the test.

Check the suggestions yourself, choose one assumption, and fill in the card. Keep real client details and sensitive research out of an assistant unless you have a suitable basis and setup to use them there.

Lean Canvas maps the business. An experiment tests one part.

Ash Maurya created Lean Canvas as a startup-focused adaptation of Alexander Osterwalder’s Business Model Canvas. Its nine areas map your problem, customers, value proposition, solution, channels, revenue, costs, key metrics and unfair advantage.

Use the Lean Canvas PDF to put the business assumptions on one page. Then select the riskiest assumption and use the experiment card below to plan a test. A completed canvas is a set of hypotheses, not proof of demand.

Download the free Lean Startup experiment card

Your next experiment

One assumption. One useful test.

A two-page PDF: a blank card with seven fillable fields, plus the hypothetical invoice-reminder plan.

Free, no signup. Open in a PDF reader that supports forms, or print and write. Record actual observations after your test.

Download the experiment card (PDF)

For more of Ries’s reasoning and examples, consider The Lean Startup. Our business book guide explains when it is a useful read and how it differs from other founder books.

Start with the assumption you would least like to be wrong about. Write the customer, the observable behavior and your decision rule on the card before you build anything new.