The Test

Rapid Experiment

Design a quick test to learn before you build.

Fastest, cheapest test that could disprove the riskiest assumption. Build-Measure-Learn loop: identify the hypothesis that, if wrong, kills the idea — then design the minimum test. Goal is not proof but a decision: keep going, pivot, or stop.

Quick Facts
Duration~20 minutes
CategoryThe Test
OriginLean Startup, 2011
Created byEric Ries
Used in pathways
The Test
Try Rapid Experiment → Download canvas PDF ↓

When to use it

Use when you have a promising idea and want to validate before investing, when about to build on an untested assumption, or when the team is debating something that could simply be tested in days.

Core tool in The Test when you need evidence, not opinions.

How it works in The Studio

Here's how a session works with WAiDE:

1
State the idea or assumption
What do you believe is true?
2
Identify the riskiest assumption
What would kill this if you're wrong?
3
Design the minimum test
What's the cheapest way to learn?
4
Define the success metric
What number would convince you?
5
Set pass/fail criteria
What's good enough? What means stop?
6
Plan the timeline
Can you run this by Friday?
7
Coaching summary
WAiDE highlights what you're avoiding testing.

Sample output

Here's what a Rapid Experiment design looks like in practice:

Example: Monthly Insight Briefing
Hypothesis
"Mid-level managers at enterprise clients would pay $200/month for a curated industry insight briefing."
Test Method
Email 50 target contacts offering a free 4-week trial, then ask for credit card at week 4.
Success Metric
>15% convert from free trial to paid ($200/month).
Timeline
5 weeks total (1 week setup + 4 week trial).
Pass
8+ paying subscribers from 50 contacts (16%). Build it properly.
Fail
<4 paying subscribers (<8%). The problem isn't urgent enough to pay for — pivot or stop.
Decision Criteria
If 8+ people pay, build it properly. If 4–7 pay, investigate what's missing. If fewer than 4, the problem isn't urgent enough to pay for — pivot or stop.

What you get

Clear experiment design: hypothesis, minimum viable test method, metric, and decision criteria (what counts as pass or fail). Designed to run in days, not months.

Foundation

Developed by Eric Ries in "The Lean Startup" (2011), building on Steve Blank's Customer Development methodology. Used by Y Combinator, Dropbox, Zappos, Airbnb, GE, P&G, and Microsoft innovation labs.

Y CombinatorDropboxAirbnbGEP&GMicrosoft

Why it works

Rapid Experiment works because the biggest enemy of innovation isn't failure — it's expensive, slow, late failure. The traditional product development model collects assumptions, bundles them into a full build, and discovers which ones were wrong after significant resources have been committed. The Lean Startup's core contribution was resequencing this: surface your riskiest assumption, design the cheapest possible test, learn, then decide whether to continue.

The scientific method has always known this. A hypothesis is only useful if it's falsifiable — if you can design a test whose result could, in principle, prove you wrong. Most business planning fails this test. Business plans are typically arguments, not hypotheses. They're structured to persuade rather than to learn. Rapid Experiment applies the discipline of scientific falsifiability to commercial questions: what would have to be true for this to work, and what's the cheapest way to find out if it is?

Dropbox's famous explainer video tested demand before a line of code was written. Zappos tested willingness to buy shoes online by manually fulfilling orders from a local shop before building any infrastructure. These aren't lucky hacks — they're applications of a principle: the unit of progress is validated learning, not features shipped. Every assumption that survives a cheap test becomes a foundation you can build on confidently.

The mechanism: The experiment discipline forces clarity before action. Most teams are fuzzy on what they actually believe and what they're actually testing. Articulating a hypothesis, a method, a metric, and a pass/fail criterion — before running anything — is itself valuable. It reveals disagreements, surfaces hidden assumptions, and creates shared language for evaluating the result.

Frequently asked questions

How small does the experiment need to be?

Small enough to run in days, not weeks. The test should be designed to answer one question — not validate the entire concept, just the single riskiest assumption underneath it. If your experiment takes more than a week to run, it's almost certainly trying to answer too many questions at once. WAiDE will help you find the version you can run by Friday.

What if the experiment fails?

A failed experiment is only a failure if you haven't defined in advance what "fail" means and what you'll do when it happens. WAiDE asks you to set decision criteria before running the test — if the result is X, we pivot; if it's Y, we continue; if it's Z, we stop. A result that triggers a pivot or a stop is exactly what the experiment was designed to produce. That's learning, not failure.

How do I know which assumption to test first?

Test the assumption whose failure would kill the entire idea. Not the assumption you're most curious about, or the one that's easiest to test — the one that, if wrong, makes everything else irrelevant. This is usually the demand assumption (will anyone actually want this?) rather than the technical or operational assumptions that teams tend to reach for first because they feel more controllable.

Isn't a small experiment too artificial to produce valid results?

The question isn't whether the experiment is perfect — it's whether it's more informative than not running it. A concierge test, a landing page, a wizard-of-oz prototype, or five customer conversations will always surface something you didn't know. The alternative is building more and learning later, which is almost always more expensive and more misleading, because by then you're too committed to change course.

Try Rapid Experiment?

WAiDE will guide you. About 20 minutes.

← Back to toolboxStart Rapid Experiment →