Design note
A plausible story is not yet a cause. The session should increase the quality of the next test, not merely the number of actions.
All playbooks
Diagnose a recurring problemMedium
Problem Solving: agree the mechanism before choosing actions
Build a shared problem statement, separate evidence from explanation and choose tests for the most credible causes.
Output
Causal hypothesis map with corrective experiments
Time
4 hours
People
6–12
Preparation
3–10 days
Facilitator
Medium
Decision owner
Situational
Output
Causal hypothesis map with corrective experiments
Use when
- A measurable problem recurs but functions explain it differently.
- The group jumps from symptoms to favourite solutions.
- Evidence can be inspected and corrective tests are possible.
Do not use when
- An active incident requires command and containment first.
- The issue is a personal conflict requiring mediation.
Preparation
- Define expected versus actual performance.
- Collect time-series, examples and counterexamples.
- Map the process boundary and affected roles.
- Separate containment already taken from causal claims.
Roles
Problem owner
Defines the outcome and owns the corrective cycle.
Process experts
Bring direct evidence from the work.
Facilitator
Separates observation, inference, cause and action.
Evidence recorder
Marks confidence and missing data visibly.
Run of show
1
30 min
Problem frame
Agree the observable gap, boundary and consequence.
Output: Problem statement
2
60 min
Evidence gallery
Inspect facts, exceptions and missing data.
Output: Evidence register
3
75 min
Causal hypotheses
Build mechanisms that could explain the pattern.
Output: Hypothesis map
4
45 min
Challenge and tests
Seek disconfirming evidence and practical tests.
Output: Prioritised tests
5
30 min
Corrective cycle
Assign owners, measures and review dates.
Output: Experiment cards
Exercises
Evidence Gallery
Turn a pre-read into a shared, inspectable evidence base before opinions dominate.
45–75 minAssumption Stress Test
Identify what must be true, what is known and what should be tested first.
35–60 minConstructive Dissent
Challenge a proposal through explicit roles, evidence and revision rules.
30–50 minArtifacts
- Problem statement
- Evidence register
- Causal hypothesis map
- Experiment cards
- Review calendar
Follow-up
- Run the smallest discriminating test before a broad rollout.
- Review evidence and mechanism separately from task completion.
- Retire explanations that fail their stated test.
Failure modes
- A cause is declared because it sounds plausible.
- The group produces a fishbone but no test.
- Corrective actions are judged by completion rather than outcome.
Remote and hybrid
Collect evidence asynchronously, use one shared causal map and time-box verbal advocacy after silent hypothesis work.
Accessibility
- Use plain-language evidence cards and define technical terms.
- Allow asynchronous evidence contribution.
- Provide a textual alternative to causal diagrams.