What is the smallest test that could change your mind?

Before investing in a full product, test the behaviour that would change your decision. The best pretotype is the simplest credible way to find out whether your idea deserves further investment.

Build & Validate
ExperimentationEvidence-led building

Adapted by the EWOR Team from an article on EWOR (opens in a new tab)3 min read

Choose the evidence before the artefact

Before building a full product, a pretotype tests whether an idea deserves more investment. Unlike a prototype, which tests how a product works, a pretotype tests whether people want it in the first place.

Start with the behaviour that would change your decision, not the name of a testing method. Define the target behaviour and the decision it would inform, then choose the smallest experience that makes that behaviour meaningful. A polished mock-up is unnecessary if a simple offer can answer the question; a simple offer won't suffice if the uncertainty lies in the interaction itself.

Match the pattern to the question

Different pretotype methods reveal different things:

  • Fake door: Present an offer or entry point before the product exists and observe whether the intended audience takes the next step.
  • Facade: Deliver a real outcome through a deliberately non-scalable process for a narrow group or period.
  • Wizard of Oz: Have a person perform work that appears to come from a system, allowing you to test the interaction before building the underlying technology.
  • Infiltrator: Test an offer within an existing store or marketplace rather than your own channel. Get the operator's approval and complete a safety review before testing in someone else's environment.
  • Impersonator: Adapt an existing product or service to stand in for the proposed solution.

You can combine methods, but each addition may introduce another explanation for the result. Use only what the question requires. If an interview can inform the decision, there's no need to simulate a transaction. If you need to deliver real value to understand the experience, a landing-page click won't be enough.

Pick a method with six questions

  1. What behaviour matters? A click, completed task, return visit, waitlist action, purchase attempt and payment are different commitments.
  2. Must the outcome be delivered? If the decision depends on experienced value, an offer alone is too weak.
  3. Must the interaction feel real? If so, specify which parts require credible fidelity and which may remain rough.
  4. Where must it happen? A natural environment may improve relevance but also increase permission and confounding risks.
  5. What can be simulated safely? List the legal, ethical, privacy and trust boundaries before design.
  6. What is the cheapest valid version? Minimise cost only after protecting the credibility of the target behaviour.

These questions should eliminate methods as well as select one. If an interview can answer the decision, do not create a fake transaction. If only real delivery reveals value, do not treat a landing-page click as sufficient.

Agree the test before building it

Before running the test, document the essentials:

  • Target participant: Who you want to test with and how you'll recruit them.
  • Hypothesis and behaviour: What you expect to happen and what you'll observe.
  • Method: Why the chosen pretotype is appropriate and why a simpler one would be insufficient.
  • Experience: What you promise participants and what happens behind the scenes.
  • Success criteria: The threshold or comparison you'll use, agreed in advance.
  • Confounding factors: How price, channel, brand or timing might affect the result.
  • Safeguards: Disclosure, consent, remediation and data handling.
  • Stop conditions: When the test must end and who owns the decision.

Moving quickly doesn't remove these obligations. A low-cost test can still damage trust, break the law or harm participants.

Interpret only what the test exposed

Every pretotype has limits. A fake door can show how people respond to an offer in a specific channel, but it can't prove retention or delivery economics. A facade can reveal experienced value while concealing the cost of scaling. A Wizard of Oz test can validate an interaction but overstate the judgement of the eventual system. An impersonator can test whether an outcome appeals to customers while benefiting from the borrowed product's existing qualities.

Record both the behaviour you observed and the differences between the test and the intended product. When results are weak, ask whether your hypothesis was wrong or whether the test failed to make the relevant behaviour possible.

The goal is to choose the smallest ethically acceptable test that can change a real decision, then keep its limitations attached to the results.

Source: EWOR article (opens in a new tab)

Get the next signal before everyone else.

Frameworks and contrarian theses for outlier founders, straight to your inbox. No fluff, no recycled advice, weekly.

The EWOR Signals newsletter