decision-memo-builder

Installation
SKILL.md

Decision Memo Builder

A 1-page memo that forces a decision. Not a status update. Not a brief. Not a FYI. The reader walks away knowing exactly what they are being asked to approve, by when, and why.


Memo DNA — four non-negotiables

Every decision memo must pass all four. If the draft fails one, rewrite. Do not ship a memo that fails Memo DNA.

  1. Brutal brevity — 1 page, ~400 words max. Every sentence earns its place or gets cut.
  2. Storyline clarity (SCP) — Situation → Complication → Resolution flow. Reader never asks "so what?"
  3. Decision-forcing — Last line is a yes/no ask with owner and date. Never "further analysis needed."
  4. Evidence density — Every claim backed by a number, quote, or source. No adjectives doing load-bearing work.

Structure

Title (≤10 words)

Name the decision, not the topic.

  • ✅ "Sunset legacy reporting engine by Q3"
  • ❌ "Reporting strategy update"

Context (3–4 sentences)

State of the world before the ask. Facts only. Baseline numbers. No history-of-the-universe framing.

Complication (2–3 sentences)

What changed, or what's at risk if we don't decide now. Include the number that makes this urgent.

  • ✅ "Usage down 15% QoQ; 3 enterprise deals ($1.2M) flagged the gap in Q1 QBRs."
  • ❌ "Usage has been declining and customers are concerned."

Options (2–3 options, 1–2 lines each)

Mutually exclusive. Each option gets:

  • Name
  • One-line description
  • The single most important trade-off (cost, time, or risk)

Do not sneak a recommendation in here.

Recommendation (1 short paragraph)

State the chosen option and why it wins on the decision criteria that matter most. Cite the evidence that tips the call.

Risks & Mitigations (≤3 bullets)

The top 3 things that could go wrong, each paired with a 1-line mitigation. Not a risk inventory — the three the reader would actually ask about.

The Ask (1 line)

A specific yes/no request with an owner and a date.

  • ✅ "Approve sunset by EOD Friday; Sarah owns migration plan by 5/1."
  • ❌ "Let's discuss next steps."

Process

  1. Interview before drafting. Ask the user:

    • Who is the reader? (CEO, VP, peer)
    • What decision do they need to make?
    • What's the deadline?
    • What are the real 2–3 options, and which do you think wins?
    • What evidence do you have? (Numbers, quotes, benchmarks)
  2. Draft in Memo DNA order. Title → Context → Complication → Options → Recommendation → Risks → Ask.

  3. Run the four tests. For each section, confirm it passes Memo DNA. Flag any that fail.

  4. Cut 30%. First drafts are always too long. Strip adjectives, hedging ("potentially," "may," "could possibly"), and throat-clearing.

  5. Stress-test the ask. Can the reader answer yes or no from the memo alone? If not, sharpen.


Worked Example 1: Tooling Migration

Title Switch engineering + product from Jira to Linear by 5/30

Context ~60 engineers and PMs on Jira today. $48K/yr in licenses. 4 active projects, custom workflows, 6 years of history. Engineering NPS on tooling = 4/10 in the Q1 survey.

Complication Jira admin role has been open 4 months; no single owner maintaining workflows. Two teams (Growth, Platform) ran Linear pilots in Q1 — 30% faster issue creation, 50% less "which field is required" confusion. Renewal is due 5/30 and auto-renews at +12%.

Options

  1. Stay on Jira — $54K next year, zero migration cost, continued tooling friction.
  2. Migrate to Linear — $36K/yr, ~6 weeks of migration work, cleaner UX, loses some Jira-specific reports.
  3. Run Linear + Jira in parallel for 6 months — hedged, doubled cost ($90K), splits team context.

Recommendation Migrate to Linear. Saves $18K/yr, ends the 4-month admin vacancy, and two pilot teams already validate the productivity win (60 people × 1 hr/wk saved ≈ $300K/yr loaded). The "lost reports" risk is low — exec dashboards already live in Looker, not Jira.

Risks & Mitigations

  • 6 years of Jira history → export to a read-only archive; no lift-and-shift.
  • Custom workflow loss → document the top 3 workflows, rebuild in Linear before cutover.
  • Team resistance → give Growth + Platform leads ownership of internal comms.

The Ask Approve Linear migration by 5/30. Priya owns migration plan; Alex cancels Jira renewal.


Worked Example 2: Scope Cut

Title Cut AI-tagging from v2 launch — approve by EOD

Context v2 is committed for 5/15. Current scope: 8 features, ~1,800 engineer-hours estimated. Team capacity = 1,400 hours remaining.

Complication We're 400 hours over. AI-tagging alone = 350 hours and blocks 3 other teams' integrations if it slips.

Options

  1. Ship all 8, slip 3 weeks — breaks GTM commits to 4 design partners.
  2. Cut AI-tagging, ship 5/15 — scope fits, saves 350 hrs, feature defers 6 weeks.
  3. Cut 2 smaller features, keep AI-tagging — fits scope but breaks onboarding funnel.

Recommendation Cut AI-tagging. Launch date > this feature. AI-tagging ranked #6 in customer survey (n=42). Cutting the 2 smaller features would break a conversion funnel we validated last quarter (+18% activation).

Risks & Mitigations

  • GTM already promised AI-tagging → update positioning memo this week.
  • Team morale dip → announce v2.1 date (7/1) alongside the cut.
  • Competitor parity → no competitor ships it today; 6-week delay = low risk.

The Ask Approve cutting AI-tagging from v2 by EOD. Alex updates GTM narrative; Jordan re-plans v2.1.


Anti-patterns

  • ❌ Bullet-point soup with no storyline
  • ❌ "We should consider..." — decision-averse hedging
  • ❌ Missing numbers — "significant impact," "major risk," "strong signal"
  • ❌ Recommendation leaked into the Options section
  • ❌ Ask that's just "discuss next steps"
  • ❌ > 1 page — if you need more, split into memo + appendix

When to Use

  • Leadership decision requests (buy/build, hire, cut, pivot)
  • Roadmap calls (ship / slip / cut)
  • Scope trade-offs (cut X to save Y)
  • Strategic pivots
  • Any moment a peer or exec needs a yes/no from you

When NOT to Use

  • FYI updates → use a status note
  • Brainstorms → use issue-tree-builder
  • Broader strategic narratives → use scpr-framework
  • Presentations → use storyline-builder
  • Post-review quality check → pair with mckinsey-critic
Related skills

More from sruthir28/enterprise-ai-skills

Installs
1
GitHub Stars
58
First Seen
7 days ago