Skip to content
X-Dreamers

Startup ideas · Lesson 4 of 6

How to Test a Business Idea: Design a Test That Can Fail

Write a small experiment with a decision rule before you see the results.

20 min · Free · no sign-up

Do this lesson in Equip →

What you will learn

Make the risky assumption explicit

State who you believe will do what, in what situation. “People want automation” is hard to disprove. “Three of five interviewed shop owners will book a trial of a weekly reconciliation service” is observable, although a small sample remains uncertain.

Match the test to the claim

An interview tests the existence and context of a problem. A landing page can test message response. A paid, clearly described manual pilot can test a stronger commitment. Clicks alone cannot test repeat use or profitability.

Set a stopping rule

Write a time limit, effort budget, success threshold and what you will do if it fails. Avoid changing the threshold after seeing the numbers. If the result is ambiguous, explain why and design a better test.

An example

Hypothesis: independent bookkeepers will reserve time for a trial of weekly invoice checking. Test: ask five relevant people for a 20-minute pilot session over one week. Threshold: three bookings. This is a learning test, not a statistically reliable estimate of the market.

The trap to avoid

Do not use fake testimonials, misleading availability or undisclosed charges to manufacture validation.

Try it: the exercise

Answer these three prompts. In the Equip learning desk your answers are saved in your browser, with a checklist to review them.

  1. Your riskiest assumptionWho will do what? Which belief could make the project fail?
  2. Experiment and timeboxSpecify a low-cost, honest test and the relevant people.
  3. Decision ruleDefine pass, fail and ambiguous outcomes before collecting results.

A good answer has

  • An observable behaviour
  • A fixed time or effort budget
  • An explicit decision for failure
Do this lesson in Equip →

Check yourself

Your experiment aims to test willingness to pay. Which signal is closest?
  1. Likes on a product announcement.
  2. A transparent paid pilot accepted by the target buyer.
  3. A long list of email subscribers.

Answer: B. A paid pilot tests a financial commitment more directly, while still not establishing retention or scalability.

Only one person books a trial; your prewritten threshold was three. What next?
  1. Call it a pass because one is better than none.
  2. Delete the threshold from your notes.
  3. Record the miss and examine audience, problem and offer before retesting.

Answer: C. Preserving the original rule prevents hindsight bias and makes the failed test useful.

From the course library

  • Startup School ↗

    Y Combinator · 7 weeks · 1–2 hr/week

    A practical starting point for evaluating ideas, planning an MVP, finding users and building a company.

  • Entrepreneurship – from ideas to reality ↗

    The Open University · OpenLearn · Self-paced · guided activities

    Explore entrepreneurial thinking and develop an idea into a business proposition through guided activities.