ENTREPRENEURSHIPWeeks to result

Hypothesis-Driven Testing with Test and Learning Cards

Structure experiments to systematically reduce business risk

Problem it solves

business risk

Best for

Teams launching new products or ventures who need a disciplined process to test critical assumptions before committing significant resources

Not ideal for

Mature products with well-understood markets where data analytics and optimization are more appropriate than hypothesis testing

Overview

Why this framework exists

This framework provides a structured, repeatable process for converting business ideas into testable hypotheses, designing experiments to validate or invalidate them, and capturing insights that drive better decisions. It centers on two practical tools: the Test Card and the Learning Card, which together create a disciplined experimentation cycle.

The Test Card forces you to articulate what you believe (hypothesis), why it matters (criticality), how you will test it (experiment design), what you will measure (metrics), and what threshold constitutes success or failure (criteria). The Learning Card then captures what actually happened, what you concluded, and what actions you will take next. Together they create an audit trail of evidence that replaces opinion-based decision making.

The framework emphasizes starting with the most critical hypotheses first, the ones that could kill your entire idea if proven wrong. It also stresses starting cheap and fast when uncertainty is highest, then increasing investment in experiments as certainty grows. This prevents the common trap of spending months perfecting a business plan for an idea that a few days of testing could have invalidated.

Core principles

5 total
  1. Evidence from experiments always trumps the opinions of founders, bosses, investors, or any other stakeholder.
  2. Start testing the hypotheses that could kill your idea first, because all other assumptions become irrelevant if the critical ones fail.
  3. What customers say and what they do are two different things; design experiments that observe actions, not just collect opinions.
  4. Start with cheap, fast experiments when uncertainty is high and increase spending on more reliable experiments as certainty grows.
  5. Failing cheaply and quickly leads to more learning, which ultimately reduces risk more than polished business plans.

Steps

6 steps
  1. Extract Hypotheses from Your Canvases
    Use the Value Proposition and Business Model Canvases to identify everything that must be true for your idea to work. Convert each assumption into an explicit, testable hypothesis. Ask yourself what needs to be true about your customer, your value proposition, and your business model.
    Pro tipFrame hypotheses as specific, falsifiable statements. Instead of 'customers will like our product,' write 'at least 30 percent of target customers will sign up for a free trial when shown our landing page.'
  2. Prioritize by Criticality
    Rank hypotheses by how critical they are to your idea's survival. Business killers go to the top of the list. These are the assumptions that, if wrong, would make the entire venture unviable regardless of how well everything else works.
    Pro tipSeparate hypotheses about the customer profile from hypotheses about your value proposition. Test that the problem exists before testing that your solution works.
    WarningDo not get seduced into testing easy or comfortable hypotheses first. The most critical assumptions are often the most uncomfortable to confront.
  3. Design Experiments with Test Cards
    For each critical hypothesis, complete a Test Card specifying the hypothesis, the experiment you will run, the data you will measure, and the success criteria that would validate or invalidate it. Include cost, time, and data reliability estimates.
    Pro tipConsider testing the same hypothesis with multiple experiments of increasing reliability. Start with a quick cheap test, then follow up with more rigorous ones if initial signals are promising.
  4. Run Experiments and Measure
    Execute experiments starting from the top of your prioritized list. Collect data rigorously and compare actual results against your predefined success criteria. Do not move goalposts after seeing results.
    Pro tipIf early experiments invalidate critical hypotheses, be prepared to go back to the drawing board entirely. The remaining test cards in your list may become irrelevant.
    WarningBeware of the five data traps: false positives, false negatives, local maximum, exhausted maximum, and wrong data. No single experiment proves anything definitively.
  5. Capture Insights with Learning Cards
    After each experiment, complete a Learning Card documenting the hypothesis tested, the actual outcomes and data, your conclusions and insights, and the specific actions you will take next. Determine whether to pivot, persevere, or seek more evidence.
    Pro tipA Learning Card may aggregate insights from several Test Cards targeting the same hypothesis. The key outputs are the decision about what to do next and clear reasoning for that decision.
  6. Track Progress on the Progress Board
    Use the Progress Board to visualize which hypotheses have been tested, validated, or invalidated across your Value Proposition and Business Model Canvases. Monitor your advancement from initial idea toward problem-solution fit, product-market fit, and business model fit.
    Pro tipKeep every Test Card and Learning Card as an audit trail. You may need to revisit earlier assumptions as new evidence emerges or as your idea evolves.

Checklist

Saved in your browser

Examples

2 cases
Value Proposition Design Book Title Split Test

The authors practiced what they preached by split-testing three different titles for their own book. They redirected traffic from businessmodelgeneration.com to test the titles with over 120,000 people across five weeks. CTAs progressed from clicking a 'learn more' button to email sign-up to completing a survey about their jobs, pains, and gains.

OutcomeThe data-driven approach to naming the book demonstrated how cheap online experiments can validate even seemingly subjective decisions, replacing internal debate with customer evidence.
Dropbox False Negative from Wrong Experiment Design

Dropbox initially tested customer interest using Google AdWords. The ads performed poorly, suggesting a lack of market interest. However, the real reason was that Dropbox represented a new market category that people were not yet searching for, not that interest was absent.

OutcomeThis cautionary example illustrates the false-negative trap: a poorly designed experiment can incorrectly invalidate a valid hypothesis. Dropbox eventually used an explainer video that went viral, demonstrating massive interest that the AdWords test failed to capture.

Common mistakes

5 traps
Refining a Business Plan Instead of Testing
Business plans create an illusion of certainty through polished projections and detailed spreadsheets. But numbers that are entirely made up do not become more reliable by being presented in a prettier format. Testing produces actual evidence while planning produces assumptions.
Testing the Solution Before Validating the Problem
If you test your value proposition without first establishing that customers care about the jobs, pains, and gains you are targeting, you never know if rejection means a bad solution or an irrelevant problem. Always test the circle before the square.
Ignoring Disconfirming Evidence
When invested in an idea, teams tend to explain away negative results or shift success criteria after seeing data. Predefined thresholds on Test Cards exist precisely to prevent this bias. Honor the criteria you set before seeing results.
Running Expensive Tests Too Early
Pilots and market studies are expensive and slow. They make sense when you have some certainty about the right direction, but deploying them when uncertainty is at maximum wastes resources. Start with cheap experiments like landing page tests or interview sprints.
Treating All Evidence as Equally Reliable
What customers say in interviews differs from what they do when spending real money. A click on a button indicates less commitment than a credit card purchase. Design experiments with awareness of evidence strength and do not overweight weak signals.

Origin story

How this framework came to be

This framework synthesizes Steve Blank's customer development process and Eric Ries's Lean Startup methodology into practical tools. Blank's core insight was that there are no facts inside the building, so entrepreneurs must get out and test assumptions with real customers. Ries added the build-measure-learn loop to make this iterative and systematic.

Osterwalder and team designed the Test Card and Learning Card as simple, visual artifacts that any team could use to impose discipline on their experimentation process, bridging the gap between lean startup theory and everyday practice in workshops and organizations worldwide.

Source

Traced to primary
Source · BOOK
Value Proposition Design: How to Create Products and Services Customers Want (Strategyzer)
Alexander Osterwalder, Yves Pigneur, Gregory Bernarda, Alan Smith · 2014
Open source →