Skip to main content
AI/MLcoalesce-labs

decision-doc

Document important product decisions. Creates decision logs with rationale, alternatives, and trade-offs.

Stars
12
Source
coalesce-labs/catalyst
Updated
2026-05-31
Slug
coalesce-labs--catalyst--decision-doc
View on GitHubRaw SKILL.md

// install — copy + paste into any project

mkdir -p .claude/skills && curl -fsSL https://raw.githubusercontent.com/coalesce-labs/catalyst/HEAD/plugins/pm/skills/decision-doc/SKILL.md -o .claude/skills/decision-doc.md

Drops the SKILL.md into .claude/skills/decision-doc.md. Works with Claude Code, Cursor, and any agent that loads SKILL.md files from .claude/skills/.

/decision-doc - Document Strategic Decisions That Stick

When the PM types /decision-doc, create decision documents that get stakeholder alignment, prevent future debates, and create institutional memory.

When to Use

  • Making a significant product direction choice
  • Evaluating build vs. buy decisions
  • Choosing between competing feature approaches
  • Deciding on architecture or technical direction
  • Resolving team disagreements on strategy
  • Making tradeoffs between conflicting goals
  • Any decision that will be questioned 3 months from now

How It Works

This is a 4-step process:

Step 1: Frame the Decision

Step 2: Present Options with Tradeoffs

Step 3: Make a Recommendation

Step 4: Document the Decision


Step 1: Frame the Decision

When the PM types /decision-doc, I'll start with:

Let's document this decision so it sticks.

**What decision are you making?**

Frame it as a clear choice:
- "Should we build feature X in-house or integrate with Partner Y?"
- "Should we launch with A/B test or full rollout?"
- "Should we prioritize mobile app or web improvements in Q2?"

**Why does this decision matter?**
- What's the impact if we get it right?
- What's the cost if we get it wrong?
- Who else needs to be involved?

**What's your timeline?**
- When do you need to decide?
- What forces the decision? (deadline, customer commitment, etc.)
- What happens if we delay?

Tell me about the decision and I'll help structure the document.

What Makes a Good Decision to Document

Do document:

  • ✅ Decisions with significant resource commitment
  • ✅ Decisions with long-term consequences
  • ✅ Decisions where reasonable people disagree
  • ✅ Decisions that will be questioned later
  • ✅ Decisions that set precedent for future choices
  • ✅ Decisions involving multiple stakeholders

Don't document:

  • ❌ Trivial tactical choices
  • ❌ Decisions that can be easily reversed
  • ❌ Decisions only affecting one person
  • ❌ Decisions that are obvious/uncontroversial

Step 2: Present Options with Tradeoffs

Once I understand the decision:

Great. Now let's map out your options.

**How many options are you considering?**
(Usually 2-4 options. More than 4 means you need to narrow first)

For each option, I need to understand:
- What it is (clear description)
- Why it's appealing (upsides)
- Why it's risky (downsides)
- What it costs (time, money, opportunity cost)
- What assumptions it requires

You can either:
- Tell me about each option
- Upload analysis you've already done
- Let me help you brainstorm options you might have missed

I'll structure this into a clear comparison framework.

The Options Framework

For each option, I'll create:

## Option [1/2/3]: [Name of Option]

### Description

[Clear, jargon-free explanation of what this option means]

**One-sentence summary:** [What this is in plain language]

### Pros (Why This Could Work)

- **[Pro 1]:** [Explanation and impact]
- **[Pro 2]:** [Explanation and impact]
- **[Pro 3]:** [Explanation and impact]

### Cons (Why This Could Fail)

- **[Con 1]:** [Explanation and risk]
- **[Con 2]:** [Explanation and risk]
- **[Con 3]:** [Explanation and risk]

### Costs & Resources Required

**Engineering:** [X weeks/people]
**Design:** [X weeks/people]
**Other Teams:** [Who else needs to be involved]
**Timeline:** [How long until we see results]
**Budget:** [Any direct costs - tools, services, licenses]

### Key Assumptions

[What needs to be true for this to work]

**Example assumptions:**

- "Assumes customers will adopt new workflow within 30 days"
- "Assumes partner API is stable and well-documented"
- "Assumes we can hire 2 senior engineers by Q2"

### Success Criteria

[How we'll know this option worked]

**Example criteria:**

- "Adoption >60% within first month"
- "Implementation completed in <8 weeks"
- "No degradation in core metrics"

### Risks & Mitigation

| Risk     | Likelihood   | Impact       | Mitigation            |
| -------- | ------------ | ------------ | --------------------- |
| [Risk 1] | High/Med/Low | High/Med/Low | [How we'd address it] |
| [Risk 2] | High/Med/Low | High/Med/Low | [How we'd address it] |

### Dependencies

- [Other team/project that must happen first]
- [External factor we can't control]
- [Technical prerequisite]

### Example: Mobile App vs Web

**Option 1: Build Native Mobile App**

- **Pro:** Better performance, offline support, push notifications
- **Con:** 6 month build time, need to hire mobile engineers, maintenance overhead
- **Cost:** 2 iOS engineers + 2 Android engineers for 6 months ($500K+)
- **Assumption:** 40% of usage will shift to mobile within 6 months
- **Risk:** Mobile engineers are hard to hire, could delay 3+ months

Step 3: Make a Recommendation

After laying out options:

You've mapped the options clearly. Now it's time to make a call.

**What's your recommendation?**

Remember: The goal isn't to make everyone happy. It's to make the right
decision for the business and clearly explain why.

A strong recommendation includes:
- **The choice:** Which option you're choosing
- **The why:** Your reasoning (data + logic)
- **The confidence level:** How certain you are (high/medium/low)
- **The failure mode:** What you'd do if you're wrong

Don't hedge. Make a clear call, even if you're not 100% certain.

I'll help you structure a compelling recommendation.

Recommendation Structure

## Recommendation: [Option Name]

### The Decision

I recommend we **[clear statement of choice]**.

**Example:** "I recommend we build a native mobile app starting Q2,
rather than continuing to improve the mobile web experience."

### Rationale

**Why this option wins:**

1. **[Primary reason]** - [Data or logic supporting this]
2. **[Secondary reason]** - [Data or logic supporting this]
3. **[Tertiary reason]** - [Data or logic supporting this]

**Why alternatives fall short:**

- **[Alternative 1]** - [Specific reason it's not as good]
- **[Alternative 2]** - [Specific reason it's not as good]

**Key tradeoffs we're accepting:**

- We're trading [X] for [Y] because [reason]
- We're deprioritizing [A] to focus on [B] because [reason]

**Example:**

- We're trading faster time-to-market for better performance because
  our user research shows performance is the #1 complaint
- We're deprioritizing web improvements to focus on mobile because
  45% of usage is now mobile and growing 15% QoQ

### Confidence Level

**Confidence:** [High / Medium / Low]

**What would increase confidence:**

- [Test/data that would make you more certain]
- [Validation that's still needed]

**Example:**

- **Confidence:** Medium-High (75%)
- **Would increase to High if:** We validate mobile engineers can be
  hired within 6 weeks, and tech lead confirms 6-month timeline is realistic

### What If We're Wrong?

**How we'll know:**

- [Leading indicator that this isn't working]
- [Timeline for when we should see results]

**Pivot plan:**

- [What we'd do if this fails]
- [How easily we could reverse course]

**Example:**

- **How we'll know:** If after 2 months we haven't hired mobile engineers,
  or if recruitment timeline slips past 12 weeks
- **Pivot plan:** Pause mobile app, redirect resources to mobile web
  improvements as interim solution

### What This Unlocks

**Positive outcomes if we're right:**

- [Business impact]
- [User impact]
- [Strategic advantage]

**Example:**

- Unlock 45% of our user base who prefer native apps
- Enable offline mode, which is #2 feature request
- Position us competitively against Competitor X who just launched mobile

Step 4: Document the Decision

Finally, I'll create the complete decision document:

I'll create your decision document in the format that works best for
your culture and stakeholders.

**Choose format:**
1. **Standard** - Comprehensive doc for important decisions
2. **DACI** - When you need clear roles and accountability
3. **One-Way Door** - For irreversible decisions (à la Amazon)
4. **Lightweight** - For smaller decisions that still need documentation

**Distribution:**
- Who needs to approve this?
- Who needs to be consulted?
- Who just needs to be informed?

I'll also create:
- Email/Slack summary to announce the decision
- FAQ section for common objections
- Timeline and next steps

Let's build this...

Standard Decision Doc Template

# Decision Document: [Decision Title]

**Date:** [Date]
**Owner:** [Your name]
**Status:** [Proposed / Approved / Implemented]
**Decision Deadline:** [When we need to decide by]

---

## TL;DR

**Decision:** [One sentence: What we're deciding]

**Recommendation:** [One sentence: What you recommend]

**Impact:** [One sentence: Why this matters]

---

## Context

### Why We're Making This Decision

[Background and situation that created the need for this decision]

**What triggered this:**

- [Event/data/feedback that forced the decision]

**Current state:**

- [Where things stand today]
- [What's not working]
- [What's at stake]

**Example:**

- **Trigger:** 45% of our traffic is now from mobile devices, up from 25%
  a year ago
- **Current state:** Mobile web experience has 35% bounce rate vs. 18% on
  desktop
- **At stake:** Losing 2,000+ potential users per month due to poor mobile
  experience

### Scope

**In scope:**

- [What this decision covers]

**Out of scope:**

- [What this decision does NOT cover]
- [Related decisions that will be made separately]

**Example:**

- **In scope:** Whether to build native iOS/Android apps
- **Out of scope:** Specific features for v1 (separate PRD), launch timing
  (separate decision)

### Constraints

**Technical:**

- [Technology limitations]

**Resource:**

- [Budget/people limitations]

**Timeline:**

- [Time constraints]

**Business:**

- [Strategic or market constraints]

---

## Options Considered

[Full breakdown of each option using the framework from Step 2]

### Option 1: [Name]

[Complete analysis]

### Option 2: [Name]

[Complete analysis]

### Option 3: [Name]

[Complete analysis]

---

## Decision Criteria

**How we're evaluating options:**

| Criteria      | Weight | Option 1 | Option 2 | Option 3 |
| ------------- | ------ | -------- | -------- | -------- |
| [Criterion 1] | High   | 8/10     | 5/10     | 6/10     |
| [Criterion 2] | High   | 6/10     | 9/10     | 7/10     |
| [Criterion 3] | Medium | 7/10     | 7/10     | 9/10     |
| **Total**     |        | **XXX**  | **XXX**  | **XXX**  |

**Criteria definitions:**

- **[Criterion 1]:** [What this means and why it matters]
- **[Criterion 2]:** [What this means and why it matters]

**Example Criteria:**

- **User Impact** (High): How much this improves core user experience
- **Time to Value** (High): How quickly we can ship and see results
- **Resource Efficiency** (Medium): How efficiently this uses our limited resources
- **Strategic Alignment** (Medium): How well this supports our Q2 strategy

---

## Recommendation

[Complete recommendation using the framework from Step 3]

---

## Stakeholder Input

### Consulted

**Engineering:**

- [Name]: [Their input/concerns]
- [Name]: [Their input/concerns]

**Design:**

- [Name]: [Their input/concerns]

**Leadership:**

- [Name]: [Their input/concerns]

### Key Concerns Raised

**Concern 1:** [Stakeholder concern]
**Response:** [How you're addressing it]

**Concern 2:** [Stakeholder concern]
**Response:** [How you're addressing it]

### Unresolved Disagreements

[If anyone strongly disagrees, document it]

**[Name] believes:** [Their position]
**My reasoning for proceeding anyway:** [Why you're still recommending this]

---

## Success Metrics

**How we'll measure success:**

**Primary Metric:**

- [Metric]: [Current] → [Target] by [Date]

**Secondary Metrics:**

- [Metric]: [Current] → [Target] by [Date]
- [Metric]: [Current] → [Target] by [Date]

**Guardrail Metrics** (must not decline):

- [Metric]: Must stay above [threshold]
- [Metric]: Must stay above [threshold]

**Example:**

- **Primary:** Mobile user retention: 45% → 65% by end of Q3
- **Secondary:**
  - Mobile DAU: 50K → 80K by end of Q3
  - Time in app: 12 min → 20 min by end of Q3
- **Guardrails:**
  - Overall retention must stay >60%
  - NPS must stay >40

### Leading Indicators (Early Signals)

**After 1 month:**

- [What we should see]

**After 3 months:**

- [What we should see]

**Example:**

- **After 1 month:** 2 mobile engineers hired, tech stack decided
- **After 3 months:** MVP in beta with 1,000 users, NPS >45

---

## Implementation Plan

### Timeline

| Phase   | Milestone   | Owner    | Target Date |
| ------- | ----------- | -------- | ----------- |
| Phase 1 | [Milestone] | @[Owner] | [Date]      |
| Phase 2 | [Milestone] | @[Owner] | [Date]      |
| Phase 3 | [Milestone] | @[Owner] | [Date]      |

### Next Steps

**Immediate (This Week):**

1. [Action item] - @[Owner]
2. [Action item] - @[Owner]

**Short-term (Next 2 Weeks):**

1. [Action item] - @[Owner]
2. [Action item] - @[Owner]

**Medium-term (Next Month):**

1. [Action item] - @[Owner]

### Dependencies

**Blockers:**

- [What must happen before we start]

**Parallel Work:**

- [What can happen at the same time]

---

## Risks & Mitigation

| Risk     | Impact       | Likelihood   | Mitigation | Owner    |
| -------- | ------------ | ------------ | ---------- | -------- |
| [Risk 1] | High/Med/Low | High/Med/Low | [Plan]     | @[Owner] |
| [Risk 2] | High/Med/Low | High/Med/Low | [Plan]     | @[Owner] |

---

## Decision Log

**Proposed:** [Date] by [Name]
**Discussed:** [Date] in [Meeting/Channel]
**Approved:** [Date] by [Name]
**Implemented:** [Date]
**Reviewed:** [Date] - [Outcome]

---

## Appendix

### Supporting Data

- [Link to research]
- [Link to analysis]
- [Link to competitive intel]

### References

- [Related PRDs]
- [Related decisions]
- [Research documents]

### FAQ

**Q: [Common question]**
A: [Answer]

**Q: [Common question]**
A: [Answer]

Alternative Formats

DACI Format (When You Need Clear Roles)

# DACI: [Decision Title]

## Roles

**Driver (owns the decision):** @[Name]
**Approver (makes the final call):** @[Name]
**Contributors (provide input):** @[Name], @[Name], @[Name]
**Informed (need to know outcome):** @[Name], @[Name]

## Decision

[What we're deciding]

## Recommendation

[Driver's recommendation]

## Context & Options

[Summary of situation and options]

## Approval Status

- [ ] Approver has reviewed
- [ ] Approver approves recommendation
- [ ] Decision communicated to Informed group
- [ ] Implementation started

**Date Approved:** [Date]
**Approver Comments:** [Any caveats or conditions]

One-Way Door Format (For Irreversible Decisions)

# One-Way Door Decision: [Title]

**Type:** ONE-WAY DOOR (Hard to reverse)

## Why This Is One-Way

[Explanation of why this decision is hard or impossible to reverse]

**Examples:**

- Architecture choices that lock us into a tech stack
- Partnerships with long-term contracts
- Decisions that create customer expectations
- Resource commitments that can't be unwound

## Extra Scrutiny Required

Because this is one-way, we're applying extra rigor:

**Questions we MUST answer:**

1. [Critical question 1]
2. [Critical question 2]
3. [Critical question 3]

**Validation we MUST do before deciding:**

- [Proof point 1]
- [Proof point 2]
- [Proof point 3]

**Dissenting voices we MUST hear from:**

- [Skeptic 1]: [Their concern]
- [Skeptic 2]: [Their concern]

**Example:**

- **Must answer:** Can we build this with our current team's skill set?
- **Must validate:** Talk to 3 companies who made similar decision - what
  do they wish they knew?
- **Must hear from:** VP Eng (concerns about tech debt) and Lead Designer
  (concerns about user impact)

## [Rest of standard decision doc]

Lightweight Format (For Smaller Decisions)

# Quick Decision: [Title]

**Date:** [Date]
**Owner:** [Name]

## The Decision

[What we're deciding in 1-2 sentences]

## Options

1. **[Option A]:** [Pro/Con in one line each]
2. **[Option B]:** [Pro/Con in one line each]

## Recommendation

**Go with [Option X]** because [one sentence reasoning].

## Next Steps

- [Action 1] - @[Owner] - [Date]
- [Action 2] - @[Owner] - [Date]

## How We'll Know It Worked

[One metric or signal]

Common Mistakes to Avoid

❌ Mistake 1: Analysis Without Recommendation

Bad: "Here are 3 options, all with pros and cons. Thoughts?" Good: "I recommend Option 2 because [reasoning]. Here's why I'm not choosing the others."

❌ Mistake 2: Hiding the Tradeoffs

Bad: Only showing upsides of your preferred option Good: Honest about what you're giving up: "We're trading speed for quality because..."

❌ Mistake 3: Decision by Committee

Bad: Trying to get everyone to agree Good: Get input from stakeholders, but make a clear call as the owner

❌ Mistake 4: Vague Success Criteria

Bad: "Success means users are happier" Good: "Success means mobile NPS increases from 40 to 55 within 3 months"

❌ Mistake 5: No Failure Plan

Bad: Not addressing what happens if you're wrong Good: "If after 2 months we haven't hit 60% adoption, we'll pivot to..."

❌ Mistake 6: Too Much Detail

Bad: 20-page document that no one reads Good: TL;DR at top, details in appendix, clear recommendation


After the Decision

Announcing the Decision

Use `/catalyst-pm-ops:slack-message` to announce your decision.

I'll create:
- **For leadership:** Executive summary of decision and rationale
- **For team:** Full context and next steps
- **For stakeholders:** How this affects their work

**Announcement should include:**
- What was decided
- Why (brief rationale)
- What happens next
- Who owns implementation
- How to provide feedback if people disagree

Following Up

## Decision Review Checkpoints

**After 1 month:**

- Review leading indicators
- Adjust course if needed
- Document what we learned

**After 3 months:**

- Measure success metrics
- Compare to predictions
- Share results with stakeholders

**After 6 months:**

- Full retrospective
- Update decision doc with "What We Learned"
- Use insights for future decisions

When to Revisit

## Triggers to Revisit This Decision

**Automatic review if:**

- [ ] Success metrics aren't tracking toward targets
- [ ] Key assumption proves false
- [ ] Major market change makes this less relevant
- [ ] Resource constraints significantly change
- [ ] Stakeholder requests formal review

**Scheduled review:**

- [Date] - First checkpoint
- [Date] - Full review

Integration With Other Commands

Informed by Competitive Analysis

Use `/competitor-analysis` first to understand market context.

Your decision doc should reference:
- Competitive positioning
- Market gaps you're exploiting
- Risks from competitive moves

Feeds Into PRD

After the decision is approved, use `/prd-draft`.

I'll auto-populate:
- Strategic rationale from your decision doc
- Success metrics
- Non-goals (from rejected options)
- Open questions

Tracked in Status Updates

Use `/catalyst-pm-ops:status-update` to track implementation progress.

I'll reference:
- Timeline from decision doc
- Success metrics to track
- Risks to monitor

Pro Tips

1. Write It Before the Meeting

Don't start the decision doc during the meeting. Write it before, circulate for async review, use meeting time for discussion.

2. Make Your Recommendation Clear

Don't hedge. Even if you're only 60% confident, make a clear call and explain your confidence level.

3. Steel-Man the Alternatives

Make the strongest possible case for options you're NOT choosing. This builds credibility and shows you thought it through.

4. Use Data, But Tell a Story

Numbers matter, but humans make decisions based on narratives. Weave data into a compelling story.

5. Document Disagreement

If someone strongly disagrees, document their position respectfully. Don't pretend everyone agrees.

6. Set Review Dates Up Front

Decide now when you'll check if this was the right call. Put it on the calendar.

7. Keep It Alive

Link to this doc in PRDs, Slack threads, and other places. Make it easy to find when people ask "why did we do this?"


Remember: The point of a decision doc isn't to be right—it's to be clear about what you're deciding, why you're deciding it, and how you'll know if it worked.


Context Routing Strategy

When the PM uses /decision-doc, I automatically:

1. Check Strategic Alignment

Source: thoughts/shared/pm/frameworks/, thoughts/shared/pm/prds/

  • What I look for: Current roadmap, strategic pillars, OKRs, company direction
  • How I use it: Ensure recommendation aligns with broader strategy
  • Example: If your strategy is "focus on activation," I'll note if this decision supports or conflicts with that

2. Identify Affected Stakeholders

Source: thoughts/shared/pm/context/stakeholder-template.md + profiles

  • What I look for: Who has input on this decision, who needs to approve, who will be affected
  • How I use it: Build stakeholder consultation section automatically
  • Example: If this decision affects Sales, I'll ask you to consult with VP Sales and note their input

3. Research Past Related Decisions

Source: thoughts/shared/product/decisions/

  • What I look for: Similar decisions made before, how they were analyzed, what succeeded/failed
  • How I use it: Reference precedent, avoid contradictory decisions, build on learnings
  • Example: "In November we decided to focus on mobile. This decision respects that commitment by..."

4. Pull Success Metrics Framework

Source: thoughts/shared/pm/metrics/, this chat thread

  • What I look for: How you typically measure success, what metrics you care about
  • How I use it: Suggest relevant metrics for this decision
  • Example: If you've defined "activation rate" as key metric in past PRDs, I'll use that metric in success criteria

5. Extract Competitive Context

Source: thoughts/shared/pm/competitive-*.md, web search if needed

  • What I look for: Competitor moves, market trends, positioning implications
  • How I use it: Include competitive rationale in decision reasoning
  • Example: "If Competitor X launches this first, we'll lose positioning advantage"

6. Route Decision for Approval

Routing logic:

  • CEO/Board level decisions: Flag as one-way door, require extra scrutiny
  • Cross-functional decisions: Ensure all department heads are consulted
  • Tactical decisions: Route to relevant team owner
  • Reversible decisions: Use lightweight format

Output Quality Self-Check

Before presenting output to the PM, verify:

  • File saved to correct location: Output saved to thoughts/shared/product/decisions/decision-[topic]-[date].md
  • Context routing table was checked: Reviewed thoughts/shared/product/decisions/ for past decisions, thoughts/shared/pm/frameworks/ for alignment, and thoughts/shared/pm/context/stakeholder-template.md for stakeholder context
  • Decision framed as clear question: The decision is stated as a specific, answerable question with 2-4 distinct options (not open-ended or vague)
  • Each alternative has pros, cons, and trade-offs: Every option includes at least 2 pros, 2 cons, and explicit trade-offs with the other options
  • Recommendation includes explicit rationale: The recommendation states which option is chosen and provides numbered reasons why, with data or logic supporting each
  • Stakeholders who need to sign off are named: Specific people (not roles) are listed as approvers, contributors, and informed parties
  • Reversibility assessment included: The document explicitly states whether this is a one-way door or two-way door decision, with reasoning
  • Conflicts with existing strategy or past decisions flagged: Any tension with decisions in thoughts/shared/product/decisions/ or goals in thoughts/shared/pm/frameworks/ is called out with explanation of how the conflict is resolved