Lean startup methodology

Lean startup methodology is a way of building a company by testing its riskiest assumptions on real customers before you spend much money. It borrows the word lean from a different idea than Toyota: spend little until you know what sells.

Eric Ries published the approach in 2011, building on Steve Blank’s customer development work. The lean here is not the factory sense of waste and flow. It is about learning fast on a small budget — the opposite of a long build in private.

Do not confuse the method with the page. A lean canvas is the one page sheet that captures your model; lean startup methodology is the loop that tests what the sheet claims and tells you when to change it.

Key takeaways

• Build, measure, learn is the core loop: ship something small, watch real behaviour, then decide.
• A minimum viable product (MVP) is the smallest build that produces an honest answer.
• Pivot or persevere is a scheduled decision made on evidence, not a mood.
• Validated learning counts only when it comes from what customers do, not what they say.

How it works

Lean startup methodology runs one short loop: build the smallest testable thing, measure how real customers behave, then learn enough to decide. Every turn ends in a single call, pivot or persevere, made on evidence rather than on the plan.

The MVP is the smallest build that produces an honest answer. It can be a landing page, a spreadsheet run by hand, or a service delivered manually behind a form — the point is the answer, not the artefact.

Write the model down before you test it. Most teams sketch it on a business model canvas and treat every box as an assumption carrying a price if it turns out to be wrong.

Then rank those assumptions by what being wrong would cost. A startup has limited runway, so the riskiest belief gets tested first rather than the easiest one.

Measurement is where most teams slip. Totals always rise, so an entrepreneur watching signups alone will see progress that is not there. Cohort retention and conversion show whether the behaviour repeats.

Outside the technology scene the same discipline has a plainer name. The Small Business Administration (SBA) advises that market research cuts risk before a business exists, and lists demand, market size and pricing as the first questions.

The pivot is the part people get wrong. A pivot changes one element of the model, such as the customer or the revenue line, while keeping what the evidence already supports. Scrapping everything is not a pivot, it is a restart.

Set the decision date when you start the test, not when the data arrives. Teams that leave it open keep running the test — which is usually a way of avoiding the answer rather than improving it.

Examples

Lean startup methodology shows up well beyond software. The pattern is the same everywhere: a cheap test, a real customer, and a date by which somebody decides. Three settings show how different the surface can look while the loop stays identical.

Public research funding uses it as policy. The United States National Science Foundation (NSF) has taught customer discovery to university teams through the National I-Corps Teams program, developed in 2012, and reports 52% of those teams have been linked to startups.

In fiscal years 2023 and 2024 the same program trained 685 teams made up of 2,307 people, across ten regional hubs representing 128 universities and colleges.

Programs exist to force the pace. A startup incubator gives early teams space and mentoring for months, and the good ones structure that time as a run of customer interviews rather than a build schedule.

Service firms test the same way. A provider can sell a new analytics offer by hand to three clients before any platform exists, which is the usual shape of innovation outsourcing work that begins as a paid pilot.

Consumer brands use the cheapest version of all. A skincare line can run a page with real prices and a waitlist, then treat paid preorders as the only evidence that anyone wants the product.

None of these teams skipped the work of building. They refused to build the expensive version first — the test comes before the platform, every time.

Related terms

• Startup: the young company the method was written for; this page is the method, not the company type.
• Entrepreneur: the person who runs the loop and owns the pivot call.
• Business model canvas: the planning sheet; this method is what tests the boxes drawn on it.
• Accelerator: a fixed term cohort program that compresses the loop into a few months.
• Venture capitalist: the investor who funds the next loop once evidence exists.

FAQ

What are the three stages of the build measure learn loop?

Build the smallest testable version, measure how customers actually behave, then learn whether the assumption held. The loop is meant to run in weeks, not quarters.

What counts as a minimum viable product?

The smallest thing that produces a trustworthy answer to one question. A manual service, a spreadsheet or a single page can all qualify if a fuller build would not change the result.

When should a startup pivot?

When the evidence says the current model will not work and no amount of execution fixes it. Set the decision date in advance so the call is made on data rather than on fatigue.

How is lean startup methodology different from a lean canvas?

The canvas is a one page record of your current model. The methodology is the testing loop that proves parts of that model wrong and tells you which box to rewrite.

Is lean startup methodology the same as lean manufacturing?

No. Lean manufacturing attacks waste and flow in a process that already runs, while this method attacks doubt about whether the business should exist.

Browse provider research at Outsource Accelerator before you fund the next build.