The Learning Card documents what a team actually learned from a test. It carries over the hypothesis and success threshold from a Test Card or Experiment Canvas, separates observation from interpretation, and turns the evidence into a next step.

1. 1

   Hypothesis

   Carry over the tested hypothesis

   

   Goal

   Preserve a traceable evaluation standard for the test.

   Inputs for this step

   • Test Card or Experiment Canvas
   • Versioned hypothesis
   • Predefined success threshold

   Actions

   • Copy the hypothesis verbatim from the test artifact.
   • Reference test population, period, and measurement rule.
   • Mark deviations from the planned test visibly.

   Output

   Referenced test basis

   Hypothesis · test · threshold · version · deviations

   Quality check

   Were the hypothesis and success threshold defined before the results were reviewed?

   Pitfall

   The hypothesis or success threshold is rephrased after the data is seen to make the result look more favorable. Restore the original wording and label later interpretations as new hypotheses.

   Timing: 3–5 minutes

2. 2

   Observation

   Document observed facts

   

   Goal

   Create a verifiable factual basis for evaluation.

   Inputs for this step

   • Raw data and observation notes
   • Measurement rule and success threshold
   • Deviation log

   Actions

   • Calculate the result against the metric and threshold.
   • Record concrete observations with source and scope.
   • Disclose missing data, failures, and test deviations.

   Output

   Observation record

   Measurement · scope · source · deviation · data gap

   Quality check

   Can another person trace the observation to the data?

   Pitfall

   Explanations such as lack of interest already appear as observed facts. Enter only visible events and measurements and move every explanation to the insight field.

   Timing: 5–10 minutes

3. 3

   Insight

   Infer a bounded insight

   

   Goal

   Form an actionable insight bounded by the evidence.

   Inputs for this step

   • Referenced test basis
   • Observation record
   • Known data and design limitations

   Actions

   • Review plausible meaning of the observations together.
   • State whether the hypothesis is supported, weakened, or remains open.
   • Make alternative explanations and limits of transfer visible.

   Output

   Bounded insight

   Interpretation · support level · uncertainty · limit

   Quality check

   Does the evidence support the stated insight without generalizing beyond the test population?

   Pitfall

   A single test result is presented as general proof or a certain cause. Limit the statement to the tested context and document competing explanations.

   Timing: 10–15 minutes

4. 4

   Action

   Commit to the next action

   

   Goal

   Turn the learning into a traceable next action.

   Inputs for this step

   • Bounded insight
   • Pending product or business decision
   • Available resources

   Actions

   • Choose persevere, adapt, retest, or stop with a rationale.
   • Name affected assumptions or model fields.
   • Set an owner, date, and expected outcome for the next step.

   Output

   Justified follow-up action

   Decision · change · owner · date · new validation need

   Quality check

   Does the action follow from the insight, and is ownership unambiguous?

   Pitfall

   The card ends with a vague recommendation or retrospectively replans the same test. Set a concrete follow-up action with an owner and create a separate test for new validation needs.

   Timing: 5–10 minutes