Write the decision you will make if the experiment disappoints
A review date is useful only when the team knows what evidence to bring and what would change the plan. Extend the strategy map with a stopping rule. This invented service example is editorial teaching material, not a result or a new account of Richard Durnall’s work.

1. Choose one outcome and the observation behind the gap. Define the denominator and time window before setting a target. Keep waiting time separate from processing time and avoid inventing a baseline from memory.
2. Write a proposed mechanism: what small change should affect which part of the work, and why? Name a counter-observation that would challenge that explanation. A target is not a test design.
3. Agree the review evidence and a guardrail. Include unintended effects and changes in demand. Use the same definitions across periods; a favourable count with a different denominator is not a comparable result.
4. Choose among continue, change, stop or collect more evidence at review. Preserve the original prediction so a disappointing result cannot be edited into an apparent success. Share the decision with affected people.

Fictional change: Add an explicit “next action” line to a small service’s acknowledgement message, hoping to reduce repeated clarification contacts.
Evidence to collect: Number of eligible messages; number of clarification contacts; waiting time definition; whether recipients misunderstood the action. No observations have yet been collected.
Counter-explanation: A different case mix or a support campaign could change contact counts. Log those changes rather than attributing the whole difference to wording.
Stop / modify rule: If the new wording directs a recipient to an incorrect action, pause it and correct the message even if the raw contact count falls. If observations are too few, report insufficient evidence.
Review card: Outcome __; baseline source/window __; change/mechanism __; disconfirming observation __; guardrail __; review owner/date __; evidence __; decision __.

Fictional change: ____________________
Evidence to collect: ____________________
Counter-explanation: ____________________
Stop / modify rule: ____________________
Review card: ____________________
Hoshin provides an alignment context; this card makes a small proposed change reviewable. It does not claim a statistically established improvement or replace investigation of the actual work.

Lean Enterprise Institute — Hoshin Kanri — https://www.lean.org/lexicon-terms/hoshin-kanri/