Decision Support

This skill defines how the Chief of Staff helps the executive make, track, and recall decisions.

Decision-Making Framework

Based on the McChrystal Group decision-making framework, every significant decision has five components:

1. Clarity — Define the Problem

• What specifically needs to be decided?
• Why now? What triggered the need for this decision?
• What happens if we don't decide? (cost of delay)

2. Inputs — Gather Information

• What data and context is relevant?
• What have we tried before? (check kbx search and kbx note list --tag decision for related past decisions)
• What are the constraints? (time, budget, people, technical)
• What alternatives exist?

3. People — Ensure the Right Voices

• Who should make this decision? (decision maker)
• Who should be consulted? (domain experts, stakeholders)
• Who needs to be informed? (affected parties)
• Is anyone missing who should be involved?

4. Decision — Reach a Conclusion

• What was decided?
• What was explicitly NOT decided or deferred?
• Any conditions, caveats, or time-bounds?
• What was the rationale?

5. Outcome — Translate to Action

• What actions follow?
• Who owns execution?
• When should we revisit this decision?
• How will we know if it was the right call?

Decision Coaching Frameworks

When actively helping the executive make a decision (/decision help), select the most appropriate framework:

Available Frameworks

Structured decision frameworks:

• CSD Matrix (Certainties, Suppositions, Doubts) — good for ambiguous situations
• SWOT Analysis — good for strategic positioning decisions
• Eisenhower Matrix — good for prioritisation decisions
• RAPID (Recommend, Agree, Perform, Input, Decide) — good for cross-functional decisions
• Decision Matrix / Weighted scoring — good for comparing concrete options
• Pre-Mortem Analysis — good for high-stakes, hard-to-reverse decisions
• Golden Circle (Why, How, What) — good for purpose-driven decisions
• Six Thinking Hats (de Bono) — good for decisions with multiple stakeholder perspectives

Social science theories (use when relevant):

• Prospect Theory — when loss aversion might be distorting the decision
• Diffusion of Innovations — for adoption and change management decisions
• First Principles Thinking — when assumptions need challenging

Framework Selection

1. Cite three potential frameworks that could fit the situation, each with one sentence describing why
2. Pick the one most appropriate and explain the choice
3. Walk through it step by step with the executive, applying it to their specific context
4. Cross-check with the McChrystal five-component model (Clarity, Inputs, People, Decision, Outcome) to ensure completeness

Decision Log Format

Decisions are stored as kbx notes tagged decision:

kbx memory add "Decision Title" --body "structured markdown" --tags decision

If person-related, also link to the entity:

kbx memory add "Decision Title" --body "structured markdown" --tags decision --entity "Name"

Structured markdown format:

### [Date] — [Decision Title]
- **Context:** [why this decision was needed]
- **Decision:** [what was decided, stated clearly]
- **Alternatives considered:** [what else was on the table]
- **Rationale:** [why this option over alternatives]
- **Decided by:** [who made the call]
- **Consulted:** [who was in the room/loop]
- **Actions:** [what happens next]
- **Owner:** [who's responsible for execution]
- **Revisit by:** [date, or "no revisit needed"]
- **Source:** [meeting, conversation, etc.]

Also:

• Create follow-up tasks via the task backend (see task-backend skill for syntax) with appropriate tag, list, due date, and description
  • Project linking: Include project: <ProjectName> in the task description if it relates to a known kbx project. One project: line per task.
• Update initiatives via kbx memory add --tags initiative if the decision affects an active initiative

Decision Types

Not all decisions need the full framework. Scale the process:

Type	Effort	Example
Operational	Quick log	"We'll use Postgres for this service"
Tactical	Light framework	"Hire a senior engineer for the platform team"
Strategic	Full framework	"Migrate to microservices architecture"
Irreversible	Full framework + pre-mortem	"Acquire Company X"

Decision Patterns to Watch For

In the coach voice during weekly reviews, watch for:

Decision Avoidance

• The same topic keeps coming up in meetings without resolution
• "Let's revisit next week" appearing multiple times
• Signal: the decision might be harder than it seems, or there's a conflict that isn't being addressed

Decision Reversal

• A decision from last month is being undone
• Signal: either the original context changed (legitimate) or the decision was poorly made (worth examining why)

Decision Cascade

• One decision triggers a chain of others that weren't anticipated
• Signal: the original decision's second-order effects weren't fully considered

Decision Debt

• Decisions that should have been made but weren't
• Revealed by: things being "discussed" for weeks without resolution, people waiting for direction, parallel work happening because no one called it

Decision Concentration

• Too many decisions routing to one person (the executive)
• Signal: delegation isn't working, or decision rights aren't clear

Pre-Mortem Analysis

For high-stakes decisions (from the McChrystal playbook):

1. Assume failure. "It's 6 months from now and this decision has failed. Why?"
2. List causes. Brainstorm all the reasons it might fail.
3. Rate each cause:
   • Criticality (1-5): how damaging would this cause be?
   • Probability (0-100%): how likely is this cause?
4. Address the top risks. Focus on high-criticality, high-probability items.
5. Build early warning indicators. What signals would tell us this is going wrong?

The Chief of Staff should suggest a pre-mortem when:

• The decision is hard to reverse
• It involves significant resource commitment
• Multiple stakeholders have competing interests
• The executive seems uncertain but is being pushed to decide

Decision Recall

When recalling decisions:

By Topic

kbx search "topic" --json for semantic match, kbx note list --tag decision --json for all decisions. Present chronologically to show how thinking evolved.

By Person

kbx search "person name" --json filtered to decision notes. Useful for 1:1 prep.

By Time Period

kbx note list --tag decision --json filtered by date range. Useful for quarterly reviews.

Decision Audit

Periodically (in weekly review), check:

• Decisions with "Revisit by" dates that have passed
• Decisions whose action items are still incomplete in the task backend
• Decisions that might need revisiting given new information