The Test Card is a compact planning artefact for one test. It connects hypothesis, test, metric, and success threshold before execution. This lets the team later determine whether the evidence sufficiently supports the assumption or calls for another learning step.

1. 1

   Hypothesis

   State the critical assumption

   

   Goal

   Define one specific, falsifiable claim that matters to the next decision.

   Inputs for this step

   • Prioritized assumption
   • Product or business model context
   • Upcoming decision

   Actions

   • State what must be true for the idea to work.
   • Name the audience, expected behaviour, and relevant context.
   • Limit the statement to exactly one testable assumption.

   Output

   Testable hypothesis

   Audience · expected behaviour · context

   Quality check

   Can an observable result clearly weaken the claim?

   Pitfall

   The statement bundles several assumptions or merely describes a desired solution. Reduce the statement to one decision-relevant assumption and add observable behaviour.

   Timing: 5 minutes

2. 2

   Test

   Choose the smallest meaningful test

   

   Goal

   Design a responsible procedure that produces evidence specifically for the hypothesis.

   Inputs for this step

   • Testable hypothesis
   • Access to the audience or test environment
   • Time, cost, privacy, and risk constraints

   Actions

   • Choose the smallest intervention with sufficient evidence strength.
   • Specify participants, procedure, duration, and data capture.
   • Check consent, privacy, deception risks, and stop authority.

   Output

   Executable test

   Participants · intervention · procedure · duration · safeguards

   Quality check

   Does the test produce evidence that directly fits the hypothesis?

   Pitfall

   The test measures convenient opinions although the hypothesis concerns actual behaviour. Match the test signal to the hypothesis and observe behaviour where required.

   Timing: 10 minutes

3. 3

   Metric

   Define the observable signal

   

   Goal

   Specify which observation or calculation counts as evidence and how it will be captured.

   Inputs for this step

   • Hypothesis and test procedure
   • Available data source
   • Relevant observation window

   Actions

   • Choose one primary metric that represents the expected behaviour.
   • Specify numerator, denominator, target segment, and window where the metric is calculated.
   • Set the data source, capture point, and quality check.

   Output

   Unambiguous metric

   Signal · data source · calculation · window

   Quality check

   Can two people derive the same value from the same raw data?

   Pitfall

   The metric counts activity without representing the behaviour claimed in the hypothesis. Align the metric with the claimed behaviour and record its calculation and data source.

   Timing: 5 to 10 minutes

4. 4

   Success threshold

   Set the threshold in advance

   

   Goal

   Define a quantitative or clearly observable boundary at which the result is considered sufficient.

   Inputs for this step

   • Defined metric
   • Baseline or justified reference value
   • Consequence of the later decision

   Actions

   • Set the success threshold before viewing results.
   • Add minimum data or required observation quality.
   • Record safety, ethics, or data-quality boundaries for pausing and stopping.

   Output

   Pre-set success threshold

   Boundary · minimum data · pause and stop limits

   Quality check

   Can the team determine whether the threshold was met without introducing a new interpretation?

   Pitfall

   The threshold remains vague or is selected to fit the observed result. Justify, date, and freeze the boundary and minimum data before launch.

   Timing: 5 to 10 minutes