The experiment card

The experiment card brings the team's agreement into one place before work starts. It follows the comparison and adoption conversation, helping everyone see what they're testing, what it could mean, and when they'll decide. Aim for one page, using Appendix A.3; keep supporting evidence alongside it when needed.

Friction. What current problem are we addressing?

Hypothesis. What do we believe will improve, for whom, and why?

Owner. Who is accountable for the business outcome?

Users. Who will use the changed process in real work?

Scope. What narrow slice will be tested, and what is excluded?

Measure. What one primary measure will decide whether the result matters?

Evidence standard. What result would justify continuing or investing?

Guardrails. What data, actions, or outputs are prohibited or require review?

Decision date. When will we stop testing and decide?

Decision choices. Stop, continue, invest, or industrialize.

Make the learning plan complete by also recording the outcome's link to the P&L or balance sheet, the bounded change, the baseline or the condition for obtaining one, a target or decision range, the time or sample boundary, cost and capacity limits, and the stop rule. Leave space for the observed result and next decision. If the financial link is still uncertain, say so; a plausible connection isn't a measured return.

If the card exposes information or agreement you don't have, choose the preparation needed to obtain it. A test intended to show improvement needs an appropriate baseline or measurement; supplying a narrative about the work doesn't establish one. A baseline-only step can record what will be observed and what would make a later test possible. It doesn't need an invented improvement target. Once the conditions are in place, use the card to keep the experiment within its agreed boundary.