Decision Document

A decision document records a material choice so teams stop relitigating it in meetings. Standard sections:

• Decision statement — one sentence, yes/no or A vs B.
• Context — why now, what happens if we defer.
• Criteria — what "good" means; weighted when stakes are high.
• Options — 2–4 real alternatives, including do-nothing or defer.
• Analysis — evidence per option against criteria.
• Recommendation — chosen path, dissent, and risks.
• Reversibility — cost to undo; review date and decision owner.

It is not a slide deck narrative or a PRD — it is an audit trail for strategy forks.

Unwritten decisions get revisited every sprint. Stakeholders remember different conversations; sunk cost grows. A decision doc speeds consensus, surfaces dissent early, and links to risk registers and OKRs. Consultants use it to show clients why a path was chosen — critical for trust.

Build vs buy, pricing model, market entry, platform bet, org design, or any fork with material cost and multiple viable options. Skip for reversible experiments with cheap rollback.

1. Write the decision statement — falsifiable; avoid vague "improve platform."
2. List criteria — 3–6; assign weights if helpful (must sum to 100%).
3. Define options — mutually exclusive; include status quo.
4. Score or narrate each option per criterion with evidence.
5. State recommendation — one option; name dissent and residual risk.
6. Record reversibility — low / medium / high; what would trigger reversal.
7. Assign owner and review date — calendar the revisit.

• Options are real, not straw men to justify a pre-baked answer.
• Dissent and risks are named with mitigations.
• Links to RSK-*, financial model, or validation evidence when available.
• Review date is on the calendar, not "someday."

• Straw-man options — "do nothing forever" vs your favourite plan.
• Recommendation without criteria — politics dressed as analysis.
• No reversibility — teams overcommit to sunk cost.
• Too long — executives stop reading; use exec summary + appendix.

Northvale Systems — build portal vs buy SRM module. Criteria: time-to-value (40%), integration fit (35%), 3yr TCO (25%). Recommendation: extend in-house portal with phased Ariba read-only feed. Reversibility: medium — data model coupling. Owner: CIO; review: 9 months.

PulseWell — freemium vs sales-led. Criteria: CAC payback, GDPR risk, enterprise pipeline quality. Recommendation: constrained freemium ≤50 employees; sales owns ≥200 seats. Dissent: CEO fears wrong ICP. Reversibility: high pre-launch.

Harbor — hire analysts vs partner diagnostic. Recommendation: one hire + partner overflow for Q4. Reversibility: low on hire. Review: two quarters of diagnostic conversion data.

Clearwater — CommCare vs custom SMS. Recommendation: CommCare for 120-site pilot. Reversibility: medium — field migration cost. Board decision by month-end.

Run /decide for DEC-001 with refs to RSK-* and strategy artefacts when present. Pair with /execupdate when leadership needs SCQA framing of the same fork.