Initial commit

2025-11-30 08:38:26 +08:00
commit 41d9f6b189
304 changed files with 98322 additions and 0 deletions
--- a/skills/decision-matrix/SKILL.md
+++ b/skills/decision-matrix/SKILL.md
@@ -0,0 +1,182 @@
+---
+name: decision-matrix
+description: Use when comparing multiple named alternatives across several criteria, need transparent trade-off analysis, making group decisions requiring alignment, choosing between vendors/tools/strategies, stakeholders need to see decision rationale, balancing competing priorities (cost vs quality vs speed), user mentions "which option should we choose", "compare alternatives", "evaluate vendors", "trade-offs", or when decision needs to be defensible and data-driven.
+---
+
+# Decision Matrix
+
+## What Is It?
+
+A decision matrix is a structured tool for comparing multiple alternatives against weighted criteria to make transparent, defensible choices. It forces explicit trade-off analysis by scoring each option on each criterion, making subjective factors visible and comparable.
+
+**Quick example:**
+
+| Option | Cost (30%) | Speed (25%) | Quality (45%) | Weighted Score |
+|--------|-----------|------------|---------------|----------------|
+| Option A | 8 (2.4) | 6 (1.5) | 9 (4.05) | **7.95** ← Winner |
+| Option B | 6 (1.8) | 9 (2.25) | 7 (3.15) | 7.20 |
+| Option C | 9 (2.7) | 4 (1.0) | 6 (2.7) | 6.40 |
+
+The numbers in parentheses show criterion score × weight. Option A wins despite not being fastest or cheapest because quality matters most (45% weight).
+
+## Workflow
+
+Copy this checklist and track your progress:
+
+```
+Decision Matrix Progress:
+- [ ] Step 1: Frame the decision and list alternatives
+- [ ] Step 2: Identify and weight criteria
+- [ ] Step 3: Score each alternative on each criterion
+- [ ] Step 4: Calculate weighted scores and analyze results
+- [ ] Step 5: Validate quality and deliver recommendation
+```
+
+**Step 1: Frame the decision and list alternatives**
+
+Ask user for decision context (what are we choosing and why), list of alternatives (specific named options, not generic categories), constraints or dealbreakers (must-have requirements), and stakeholders (who needs to agree). Understanding must-haves helps filter options before scoring. See [Framing Questions](#framing-questions) for clarification prompts.
+
+**Step 2: Identify and weight criteria**
+
+Collaborate with user to identify criteria (what factors matter for this decision), determine weights (which criteria matter most, as percentages summing to 100%), and validate coverage (do criteria capture all important trade-offs). If user is unsure about weighting → Use [resources/template.md](resources/template.md) for weighting techniques. See [Criterion Types](#criterion-types) for common patterns.
+
+**Step 3: Score each alternative on each criterion**
+
+For each option, score on each criterion using consistent scale (typically 1-10 where 10 = best). Ask user for scores or research objective data (cost, speed metrics) where available. Document assumptions and data sources. For complex scoring → See [resources/methodology.md](resources/methodology.md) for calibration techniques.
+
+**Step 4: Calculate weighted scores and analyze results**
+
+Calculate weighted score for each option (sum of criterion score × weight). Rank options by total score. Identify close calls (options within 5% of each other). Check for sensitivity (would changing one weight flip the decision). See [Sensitivity Analysis](#sensitivity-analysis) for interpretation guidance.
+
+**Step 5: Validate quality and deliver recommendation**
+
+Self-assess using [resources/evaluators/rubric_decision_matrix.json](resources/evaluators/rubric_decision_matrix.json) (minimum score ≥ 3.5). Present decision-matrix.md file with clear recommendation, highlight key trade-offs revealed by analysis, note sensitivity to assumptions, and suggest next steps (gather more data on close calls, validate with stakeholders).
+
+## Framing Questions
+
+**To clarify the decision:**
+- What specific decision are we making? (Choose X from Y alternatives)
+- What happens if we don't decide or choose wrong?
+- When do we need to decide by?
+- Can we choose multiple options or only one?
+
+**To identify alternatives:**
+- What are all the named options we're considering?
+- Are there other alternatives we're ruling out immediately? Why?
+- What's the "do nothing" or status quo option?
+
+**To surface must-haves:**
+- Are there absolute dealbreakers? (Budget cap, timeline requirement, compliance need)
+- Which constraints are flexible vs rigid?
+
+## Criterion Types
+
+Common categories for criteria (adapt to your decision):
+
+**Financial Criteria:**
+- Upfront cost, ongoing cost, ROI, payback period, budget impact
+- Typical weight: 20-40% (higher for cost-sensitive decisions)
+
+**Performance Criteria:**
+- Speed, quality, reliability, scalability, capacity, throughput
+- Typical weight: 30-50% (higher for technical decisions)
+
+**Risk Criteria:**
+- Implementation risk, reversibility, vendor lock-in, technical debt, compliance risk
+- Typical weight: 10-25% (higher for enterprise/regulated environments)
+
+**Strategic Criteria:**
+- Alignment with goals, future flexibility, competitive advantage, market positioning
+- Typical weight: 15-30% (higher for long-term decisions)
+
+**Operational Criteria:**
+- Ease of use, maintenance burden, training required, integration complexity
+- Typical weight: 10-20% (higher for internal tools)
+
+**Stakeholder Criteria:**
+- Team preference, user satisfaction, executive alignment, customer impact
+- Typical weight: 5-15% (higher for change management contexts)
+
+## Weighting Approaches
+
+**Method 1: Direct Allocation (simplest)**
+Stakeholders assign percentages totaling 100%. Quick but can be arbitrary.
+
+**Method 2: Pairwise Comparison (more rigorous)**
+Compare each criterion pair: "Is cost more important than speed?" Build ranking, then assign weights.
+
+**Method 3: Must-Have vs Nice-to-Have (filters first)**
+Separate absolute requirements (pass/fail) from weighted criteria. Only evaluate options that pass must-haves.
+
+**Method 4: Stakeholder Averaging (group decisions)**
+Each stakeholder assigns weights independently, then average. Reveals divergence in priorities.
+
+See [resources/methodology.md](resources/methodology.md) for detailed facilitation techniques.
+
+## Sensitivity Analysis
+
+After calculating scores, check robustness:
+
+**1. Close calls:** Options within 5-10% of winner → Need more data or second opinion
+**2. Dominant criteria:** One criterion driving entire decision → Is weight too high?
+**3. Weight sensitivity:** Would swapping two criterion weights flip the winner? → Decision is fragile
+**4. Score sensitivity:** Would adjusting one score by ±1 point flip the winner? → Decision is sensitive to that data point
+
+**Red flags:**
+- Winner changes with small weight adjustments → Need stakeholder alignment on priorities
+- One option wins every criterion → Matrix is overkill, choice is obvious
+- Scores are mostly guesses → Gather more data before deciding
+
+## Common Patterns
+
+**Technology Selection:**
+- Criteria: Cost, performance, ecosystem maturity, team familiarity, vendor support
+- Weight: Performance and maturity typically 50%+
+
+**Vendor Evaluation:**
+- Criteria: Price, features, integration, support, reputation, contract terms
+- Weight: Features and integration typically 40-50%
+
+**Strategic Choices:**
+- Criteria: Market opportunity, resource requirements, risk, alignment, timing
+- Weight: Market opportunity and alignment typically 50%+
+
+**Hiring Decisions:**
+- Criteria: Experience, culture fit, growth potential, compensation expectations, availability
+- Weight: Experience and culture fit typically 50%+
+
+**Feature Prioritization:**
+- Criteria: User impact, effort, strategic value, risk, dependencies
+- Weight: User impact and strategic value typically 50%+
+
+## When NOT to Use This Skill
+
+**Skip decision matrix if:**
+- Only one viable option (no real alternatives to compare)
+- Decision is binary yes/no with single criterion (use simpler analysis)
+- Options differ on only one dimension (just compare that dimension)
+- Decision is urgent and stakes are low (analysis overhead not worth it)
+- Criteria are impossible to define objectively (purely emotional/aesthetic choice)
+- You already know the answer (using matrix to justify pre-made decision is waste)
+
+**Use instead:**
+- Single criterion → Simple ranking or threshold check
+- Binary decision → Pro/con list or expected value calculation
+- Highly uncertain → Scenario planning or decision tree
+- Purely subjective → Gut check or user preference vote
+
+## Quick Reference
+
+**Process:**
+1. Frame decision → List alternatives
+2. Identify criteria → Assign weights (sum to 100%)
+3. Score each option on each criterion (1-10 scale)
+4. Calculate weighted scores → Rank options
+5. Check sensitivity → Deliver recommendation
+
+**Resources:**
+- [resources/template.md](resources/template.md) - Structured matrix format and weighting techniques
+- [resources/methodology.md](resources/methodology.md) - Advanced techniques (group facilitation, calibration, sensitivity analysis)
+- [resources/evaluators/rubric_decision_matrix.json](resources/evaluators/rubric_decision_matrix.json) - Quality checklist before delivering
+
+**Deliverable:** `decision-matrix.md` file with table, rationale, and recommendation
--- a/skills/decision-matrix/resources/evaluators/rubric_decision_matrix.json
+++ b/skills/decision-matrix/resources/evaluators/rubric_decision_matrix.json
@@ -0,0 +1,218 @@
+{
+  "criteria": [
+    {
+      "name": "Decision Framing & Context",
+      "description": "Is the decision clearly defined with all viable alternatives identified?",
+      "scoring": {
+        "1": "Decision is vague or ill-defined. Alternatives are incomplete or include non-comparable options. No stakeholder identification.",
+        "3": "Decision is stated but lacks specificity. Most alternatives listed but may be missing key options. Stakeholders mentioned generally.",
+        "5": "Exemplary framing. Decision is specific and unambiguous. All viable alternatives identified (including 'do nothing' if relevant). Must-have requirements separated from criteria. Stakeholders clearly identified with their priorities noted."
+      }
+    },
+    {
+      "name": "Criteria Quality & Coverage",
+      "description": "Are criteria well-chosen, measurable, independent, and comprehensive?",
+      "scoring": {
+        "1": "Criteria are vague, redundant, or missing key factors. Too many (>10) or too few (<3). No clear definitions.",
+        "3": "Criteria cover main factors but may have some redundancy or gaps. 4-8 criteria with basic definitions. Some differentiation between options.",
+        "5": "Exemplary criteria selection. 4-7 criteria that are measurable, independent, relevant, and differentiate between options. Each criterion has clear definition and measurement approach. No redundancy. Captures all important trade-offs."
+      }
+    },
+    {
+      "name": "Weighting Appropriateness",
+      "description": "Do criterion weights reflect true priorities and sum to 100%?",
+      "scoring": {
+        "1": "Weights don't sum to 100%, are arbitrary, or clearly misaligned with stated priorities. No rationale provided.",
+        "3": "Weights sum to 100% and are reasonable but may lack explicit justification. Some alignment with priorities.",
+        "5": "Exemplary weighting. Weights sum to 100%, clearly reflect stakeholder priorities, and have documented rationale (pairwise comparison, swing weighting, or stakeholder averaging). Weight distribution makes sense for decision type."
+      }
+    },
+    {
+      "name": "Scoring Rigor & Data Quality",
+      "description": "Are scores based on data or defensible judgments with documented sources?",
+      "scoring": {
+        "1": "Scores appear to be wild guesses with no justification. No data sources. Inconsistent scale usage.",
+        "3": "Mix of data-driven and subjective scores. Some sources documented. Mostly consistent 1-10 scale. Some assumptions noted.",
+        "5": "Exemplary scoring rigor. Objective criteria backed by real data (quotes, benchmarks, measurements). Subjective criteria have clear anchors/definitions. All assumptions and data sources documented. Consistent 1-10 scale usage."
+      }
+    },
+    {
+      "name": "Calculation Accuracy",
+      "description": "Are weighted scores calculated correctly and presented clearly?",
+      "scoring": {
+        "1": "Calculation errors present. Weights don't match stated percentages. Formula mistakes. Unclear presentation.",
+        "3": "Calculations are mostly correct with minor issues. Weighted scores shown but presentation could be clearer.",
+        "5": "Perfect calculations. Weighted scores = Σ(score × weight) for each option. Table clearly shows raw scores, weights (as percentages), weighted scores, and totals. Ranking is correct."
+      }
+    },
+    {
+      "name": "Sensitivity Analysis",
+      "description": "Is decision robustness assessed (close calls, weight sensitivity, score uncertainty)?",
+      "scoring": {
+        "1": "No sensitivity analysis. Winner declared without checking if decision is robust.",
+        "3": "Basic sensitivity noted (e.g., 'close call' mentioned) but not systematically analyzed.",
+        "5": "Thorough sensitivity analysis. Identifies close calls (<10% margin). Tests weight sensitivity (would swapping weights flip decision?). Notes which scores are most uncertain. Assesses decision robustness and flags fragile decisions."
+      }
+    },
+    {
+      "name": "Recommendation Quality",
+      "description": "Is recommendation clear with rationale, trade-offs, and confidence level?",
+      "scoring": {
+        "1": "No clear recommendation or just states winner without rationale. No trade-off discussion.",
+        "3": "Recommendation stated with basic rationale. Some trade-offs mentioned. Confidence level implied but not stated.",
+        "5": "Exemplary recommendation. Clear winner with score. Explains WHY winner prevails (which criteria drive decision). Acknowledges trade-offs (where winner scores lower). States confidence level based on margin and sensitivity. Suggests next steps."
+      }
+    },
+    {
+      "name": "Assumption & Limitation Documentation",
+      "description": "Are key assumptions, uncertainties, and limitations explicitly stated?",
+      "scoring": {
+        "1": "No assumptions documented. Presents results as facts without acknowledging uncertainty or limitations.",
+        "3": "Some assumptions mentioned. Acknowledges uncertainty exists but not comprehensive.",
+        "5": "All key assumptions explicitly documented. Uncertainties flagged (which scores are guesses vs data). Limitations noted (e.g., 'cost estimates are preliminary', 'performance benchmarks unavailable'). Reader understands confidence bounds."
+      }
+    },
+    {
+      "name": "Stakeholder Alignment",
+      "description": "For group decisions, are different stakeholder priorities surfaced and addressed?",
+      "scoring": {
+        "1": "Single set of weights/scores presented as if universal. No acknowledgment of stakeholder differences.",
+        "3": "Stakeholder differences mentioned but not systematically addressed. Single averaged view presented.",
+        "5": "Stakeholder differences explicitly surfaced. If priorities diverge, shows impact (e.g., 'Under engineering priorities, A wins; under sales priorities, B wins'). Facilitates alignment or escalates decision appropriately."
+      }
+    },
+    {
+      "name": "Communication & Presentation",
+      "description": "Is matrix table clear, readable, and appropriately formatted?",
+      "scoring": {
+        "1": "Matrix is confusing, poorly formatted, or missing key elements (weights, totals). Hard to interpret.",
+        "3": "Matrix is readable with minor formatting issues. Weights and totals shown but could be clearer.",
+        "5": "Exemplary presentation. Table is clean and scannable. Column headers show criteria names AND weights (%). Weighted scores shown (not just raw scores). Winner visually highlighted. Assumptions and next steps clearly stated."
+      }
+    }
+  ],
+  "minimum_score": 3.5,
+  "guidance_by_decision_type": {
+    "Technology Selection (tools, platforms, vendors)": {
+      "target_score": 4.0,
+      "focus_criteria": [
+        "Criteria Quality & Coverage",
+        "Scoring Rigor & Data Quality",
+        "Sensitivity Analysis"
+      ],
+      "common_pitfalls": [
+        "Missing 'Total Cost of Ownership' as criterion (not just upfront cost)",
+        "Ignoring integration complexity or vendor lock-in risk",
+        "Not scoring 'do nothing / keep current solution' as baseline"
+      ]
+    },
+    "Strategic Choices (market entry, partnerships, positioning)": {
+      "target_score": 4.0,
+      "focus_criteria": [
+        "Decision Framing & Context",
+        "Weighting Appropriateness",
+        "Stakeholder Alignment"
+      ],
+      "common_pitfalls": [
+        "Weighting short-term metrics too heavily over strategic fit",
+        "Not including reversibility / optionality as criterion",
+        "Ignoring stakeholder misalignment on priorities"
+      ]
+    },
+    "Vendor / Supplier Evaluation": {
+      "target_score": 3.8,
+      "focus_criteria": [
+        "Criteria Quality & Coverage",
+        "Scoring Rigor & Data Quality",
+        "Assumption & Limitation Documentation"
+      ],
+      "common_pitfalls": [
+        "Relying on vendor-provided data without validation",
+        "Not including 'vendor financial health' or 'support SLA' criteria",
+        "Missing contract terms (pricing lock, exit clauses) as criterion"
+      ]
+    },
+    "Feature Prioritization": {
+      "target_score": 3.5,
+      "focus_criteria": [
+        "Weighting Appropriateness",
+        "Scoring Rigor & Data Quality",
+        "Sensitivity Analysis"
+      ],
+      "common_pitfalls": [
+        "Not including 'effort' or 'technical risk' as criteria",
+        "Scoring 'user impact' without user research data",
+        "Ignoring dependencies between features"
+      ]
+    },
+    "Hiring Decisions": {
+      "target_score": 3.5,
+      "focus_criteria": [
+        "Criteria Quality & Coverage",
+        "Scoring Rigor & Data Quality",
+        "Assumption & Limitation Documentation"
+      ],
+      "common_pitfalls": [
+        "Criteria too vague (e.g., 'culture fit' without definition)",
+        "Interviewer bias in scores (need calibration)",
+        "Not documenting what good vs poor looks like for each criterion"
+      ]
+    }
+  },
+  "guidance_by_complexity": {
+    "Simple (3-4 alternatives, clear criteria, aligned stakeholders)": {
+      "target_score": 3.5,
+      "sufficient_rigor": "Basic weighting (direct allocation), data-driven scores where possible, simple sensitivity check (margin analysis)"
+    },
+    "Moderate (5-7 alternatives, some subjectivity, minor disagreement)": {
+      "target_score": 3.8,
+      "sufficient_rigor": "Structured weighting (rank-order or pairwise), documented scoring rationale, sensitivity analysis on close calls"
+    },
+    "Complex (8+ alternatives, high subjectivity, stakeholder conflict)": {
+      "target_score": 4.2,
+      "sufficient_rigor": "Advanced weighting (AHP, swing), score calibration/normalization, Monte Carlo or scenario sensitivity, stakeholder convergence process (Delphi, NGT)"
+    }
+  },
+  "common_failure_modes": {
+    "1. Post-Rationalization": {
+      "symptom": "Weights or scores appear engineered to justify pre-made decision",
+      "detection": "Oddly specific weights (37%), generous scores for preferred option, stakeholders admit 'we already know the answer'",
+      "prevention": "Assign weights BEFORE scoring alternatives. Use blind facilitation. Ask: 'If matrix contradicts gut, do we trust it?'"
+    },
+    "2. Garbage In, Garbage Out": {
+      "symptom": "All scores are guesses with no data backing",
+      "detection": "Cannot answer 'where did this score come from?', scores assigned in <5 min, all round numbers (5, 7, 8)",
+      "prevention": "Require data sources for objective criteria. Define scoring anchors for subjective criteria. Flag uncertainties."
+    },
+    "3. Analysis Paralysis": {
+      "symptom": "Endless refinement, never deciding",
+      "detection": ">10 criteria, winner changes 3+ times, 'just one more round' requests",
+      "prevention": "Set decision deadline. Cap criteria at 5-7. Use satisficing rule: 'Any option >7.0 is acceptable.'"
+    },
+    "4. Criterion Soup": {
+      "symptom": "Overlapping, redundant, or conflicting criteria",
+      "detection": "Two criteria always score the same, scorer confusion ('how is this different?')",
+      "prevention": "Independence test: Can option score high on A but low on B? If no, merge them. Write clear definitions."
+    },
+    "5. Ignoring Sensitivity": {
+      "symptom": "Winner declared without robustness check",
+      "detection": "No mention of margin, close calls, or what would flip decision",
+      "prevention": "Always report margin. Test: 'If we swapped top 2 weights, does winner change?' Flag fragile decisions."
+    },
+    "6. Stakeholder Misalignment": {
+      "symptom": "Different stakeholders have different priorities but single matrix presented",
+      "detection": "Engineering wants A, sales wants B, but matrix 'proves' one is right",
+      "prevention": "Surface weight differences. Show 'under X priorities, A wins; under Y priorities, B wins.' Escalate if needed."
+    },
+    "7. Missing 'Do Nothing'": {
+      "symptom": "Only evaluating new alternatives, forgetting status quo is an option",
+      "detection": "All alternatives are new changes, no baseline comparison",
+      "prevention": "Always include current state / do nothing as an option to evaluate if change is worth it."
+    },
+    "8. False Precision": {
+      "symptom": "Scores to 2 decimals when underlying data is rough guess",
+      "detection": "Weighted total: 7.342 but scores are subjective estimates",
+      "prevention": "Match precision to confidence. Rough guesses → round to 0.5. Data-driven → decimals OK."
+    }
+  }
+}
--- a/skills/decision-matrix/resources/methodology.md
+++ b/skills/decision-matrix/resources/methodology.md
@@ -0,0 +1,398 @@
+# Decision Matrix: Advanced Methodology
+
+## Workflow
+
+Copy this checklist for complex decision scenarios:
+
+```
+Advanced Decision Matrix Progress:
+- [ ] Step 1: Diagnose decision complexity
+- [ ] Step 2: Apply advanced weighting techniques
+- [ ] Step 3: Calibrate and normalize scores
+- [ ] Step 4: Perform rigorous sensitivity analysis
+- [ ] Step 5: Facilitate group convergence
+```
+
+**Step 1: Diagnose decision complexity** - Identify complexity factors (stakeholder disagreement, high uncertainty, strategic importance). See [1. Decision Complexity Assessment](#1-decision-complexity-assessment).
+
+**Step 2: Apply advanced weighting techniques** - Use AHP or other rigorous methods for contentious decisions. See [2. Advanced Weighting Methods](#2-advanced-weighting-methods).
+
+**Step 3: Calibrate and normalize scores** - Handle different scoring approaches and normalize across scorers. See [3. Score Calibration & Normalization](#3-score-calibration--normalization).
+
+**Step 4: Perform rigorous sensitivity analysis** - Test decision robustness with Monte Carlo or scenario analysis. See [4. Advanced Sensitivity Analysis](#4-advanced-sensitivity-analysis).
+
+**Step 5: Facilitate group convergence** - Use Delphi method or consensus-building techniques. See [5. Group Decision Facilitation](#5-group-decision-facilitation).
+
+---
+
+## 1. Decision Complexity Assessment
+
+### Complexity Indicators
+
+**Low Complexity** (use basic template):
+- Clear stakeholder alignment on priorities
+- Objective criteria with available data
+- Low stakes (reversible decision)
+- 3-5 alternatives
+
+**Medium Complexity** (use enhanced techniques):
+- Moderate stakeholder disagreement
+- Mix of objective and subjective criteria
+- Moderate stakes (partially reversible)
+- 5-8 alternatives
+
+**High Complexity** (use full methodology):
+- Significant stakeholder disagreement on priorities
+- Mostly subjective criteria or high uncertainty
+- High stakes (irreversible or strategic decision)
+- >8 alternatives or multi-phase decision
+- Regulatory or compliance implications
+
+### Complexity Scoring
+
+| Factor | Low (1) | Medium (2) | High (3) |
+|--------|---------|------------|----------|
+| **Stakeholder alignment** | Aligned priorities | Some disagreement | Conflicting priorities |
+| **Criteria objectivity** | Mostly data-driven | Mix of data & judgment | Mostly subjective |
+| **Decision stakes** | Reversible, low cost | Partially reversible | Irreversible, strategic |
+| **Uncertainty level** | Low uncertainty | Moderate uncertainty | High uncertainty |
+| **Number of alternatives** | 3-4 options | 5-7 options | 8+ options |
+
+**Complexity Score = Sum of factors**
+- **5-7 points:** Use basic template
+- **8-11 points:** Use enhanced techniques (sections 2-3)
+- **12-15 points:** Use full methodology (all sections)
+
+---
+
+## 2. Advanced Weighting Methods
+
+### Analytic Hierarchy Process (AHP)
+
+**When to use:** High-stakes decisions with contentious priorities, need rigorous justification
+
+**Process:**
+
+1. **Create pairwise comparison matrix:** For each pair, rate 1-9 (1=equal, 3=slightly more important, 5=moderately, 7=strongly, 9=extremely)
+2. **Calculate weights:** Normalize columns, average rows
+3. **Check consistency:** CR < 0.10 acceptable (use online AHP calculator: bpmsg.com/ahp/ahp-calc.php)
+
+**Example:** Comparing Cost, Performance, Risk, Ease pairwise yields weights: Performance 55%, Risk 20%, Cost 15%, Ease 10%
+
+**Advantage:** Rigorous, forces logical consistency in pairwise judgments.
+
+### Swing Weighting
+
+**When to use:** Need to justify weights based on value difference, not just importance
+
+**Process:**
+
+1. **Baseline:** Imagine all criteria at worst level
+2. **Swing:** For each criterion, ask "What value does moving from worst to best create?"
+3. **Rank swings:** Which swing creates most value?
+4. **Assign points:** Give highest swing 100 points, others relative to it
+5. **Convert to weights:** Normalize points to percentages
+
+**Example:**
+
+| Criterion | Worst → Best Scenario | Value of Swing | Points | Weight |
+|-----------|----------------------|----------------|--------|--------|
+| Performance | 50ms → 5ms response | Huge value gain | 100 | 45% |
+| Cost | $100K → $50K | Moderate value | 60 | 27% |
+| Risk | High → Low risk | Significant value | 50 | 23% |
+| Ease | Hard → Easy to use | Minor value | 10 | 5% |
+
+**Total points:** 220 → **Weights:** 100/220=45%, 60/220=27%, 50/220=23%, 10/220=5%
+
+**Advantage:** Focuses on marginal value, not abstract importance. Reveals if criteria with wide option variance should be weighted higher.
+
+### Multi-Voting (Group Weighting)
+
+**When to use:** Group of 5-15 stakeholders needs to converge on weights
+
+**Process:**
+
+1. **Round 1 - Individual allocation:** Each person assigns 100 points across criteria
+2. **Reveal distribution:** Show average and variance for each criterion
+3. **Discuss outliers:** Why did some assign 40% to Cost while others assigned 10%?
+4. **Round 2 - Revised allocation:** Re-allocate with new information
+5. **Converge:** Repeat until variance is acceptable or use average
+
+**Example:**
+
+| Criterion | Round 1 Avg | Round 1 Variance | Round 2 Avg | Round 2 Variance |
+|-----------|-------------|------------------|-------------|------------------|
+| Cost | 25% | High (±15%) | 30% | Low (±5%) |
+| Performance | 40% | Medium (±10%) | 38% | Low (±4%) |
+| Risk | 20% | Low (±5%) | 20% | Low (±3%) |
+| Ease | 15% | High (±12%) | 12% | Low (±4%) |
+
+**Convergence achieved** when variance <±5% for all criteria.
+
+---
+
+## 3. Score Calibration & Normalization
+
+### Handling Different Scorer Tendencies
+
+**Problem:** Some scorers are "hard graders" (6-7 range), others are "easy graders" (8-9 range). This skews results.
+
+**Solution: Z-score normalization**
+
+**Step 1: Calculate each scorer's mean and standard deviation**
+
+Scorer A: Gave scores [8, 9, 7, 8] → Mean=8, SD=0.8
+Scorer B: Gave scores [5, 6, 4, 6] → Mean=5.25, SD=0.8
+
+**Step 2: Normalize each score**
+
+Z-score = (Raw Score - Scorer Mean) / Scorer SD
+
+**Step 3: Re-scale to 1-10**
+
+Normalized Score = 5.5 + (Z-score × 1.5)
+
+**Result:** Scorers are calibrated to same scale, eliminating grading bias.
+
+### Dealing with Missing Data
+
+**Scenario:** Some alternatives can't be scored on all criteria (e.g., vendor A won't share cost until later).
+
+**Approach 1: Conditional matrix**
+
+Score available criteria only, note which are missing. Once data arrives, re-run matrix.
+
+**Approach 2: Pessimistic/Optimistic bounds**
+
+Assign worst-case and best-case scores for missing data. Run matrix twice:
+- Pessimistic scenario: Missing data gets low score (e.g., 3)
+- Optimistic scenario: Missing data gets high score (e.g., 8)
+
+If same option wins both scenarios → Decision is robust. If different winners → Missing data is decision-critical, must obtain before deciding.
+
+### Non-Linear Scoring Curves
+
+**Problem:** Not all criteria are linear. E.g., cost difference between $10K and $20K matters more than $110K vs $120K.
+
+**Solution: Apply utility curves**
+
+**Diminishing returns curve** (Cost, Time):
+- Score = 10 × (1 - e^(-k × Cost Improvement))
+- k = sensitivity parameter (higher k = faster diminishing returns)
+
+**Threshold curve** (Must meet minimum):
+- Score = 0 if below threshold
+- Score = 1-10 linear above threshold
+
+**Example:** Load time criterion with 2-second threshold:
+- Option A: 1.5s → Score = 10 (below threshold = great)
+- Option B: 3s → Score = 5 (above threshold, linear penalty)
+- Option C: 5s → Score = 1 (way above threshold)
+
+---
+
+## 4. Advanced Sensitivity Analysis
+
+### Monte Carlo Sensitivity
+
+**When to use:** High uncertainty in scores, want to understand probability distribution of outcomes
+
+**Process:**
+
+1. **Define uncertainty ranges** for each score
+   - Option A Cost score: 6 ± 2 (could be 4-8)
+   - Option A Performance: 9 ± 0.5 (could be 8.5-9.5)
+
+2. **Run simulations** (1000+ iterations):
+   - Randomly sample scores within uncertainty ranges
+   - Calculate weighted total for each option
+   - Record winner
+
+3. **Analyze results:**
+   - Option A wins: 650/1000 = 65% probability
+   - Option B wins: 300/1000 = 30% probability
+   - Option C wins: 50/1000 = 5% probability
+
+**Interpretation:**
+- **>80% win rate:** High confidence in decision
+- **50-80% win rate:** Moderate confidence, option is likely but not certain
+- **<50% win rate:** Low confidence, gather more data or consider decision is close call
+
+**Tools:** Excel (=RANDBETWEEN or =NORM.INV), Python (numpy.random), R (rnorm)
+
+### Scenario Analysis
+
+**When to use:** Future is uncertain, decisions need to be robust across scenarios
+
+**Process:**
+
+1. **Define scenarios** (typically 3-4):
+   - Best case: Favorable market conditions
+   - Base case: Expected conditions
+   - Worst case: Unfavorable conditions
+   - Black swan: Unlikely but high-impact event
+
+2. **Adjust criterion weights or scores per scenario:**
+
+| Scenario | Cost Weight | Performance Weight | Risk Weight |
+|----------|-------------|--------------------|-------------|
+| Best case | 20% | 50% | 30% |
+| Base case | 30% | 40% | 30% |
+| Worst case | 40% | 20% | 40% |
+
+3. **Run matrix for each scenario**, identify winner
+
+4. **Evaluate robustness:**
+   - **Dominant option:** Wins in all scenarios → Robust choice
+   - **Scenario-dependent:** Different winners → Need to assess scenario likelihood
+   - **Mixed:** Wins in base + one other → Moderately robust
+
+### Threshold Analysis
+
+**Question:** At what weight does the decision flip?
+
+**Process:**
+
+1. **Vary one criterion weight** from 0% to 100% (keeping others proportional)
+2. **Plot total scores** for all options vs. weight
+3. **Identify crossover point** where lines intersect (decision flips)
+
+**Example:**
+
+When Performance weight < 25% → Option B wins (cost-optimized)
+When Performance weight > 25% → Option A wins (performance-optimized)
+
+**Insight:** Current weight is 40% for Performance. Decision is robust unless Performance drops below 25% importance.
+
+**Practical use:** Communicate to stakeholders: "Even if we reduce Performance priority to 25% (vs current 40%), Option A still wins. Decision is robust."
+
+---
+
+## 5. Group Decision Facilitation
+
+### Delphi Method (Asynchronous Consensus)
+
+**When to use:** Experts geographically distributed, want to avoid groupthink, need convergence without meetings
+
+**Process:**
+
+**Round 1:**
+- Each expert scores options independently (no discussion)
+- Facilitator compiles scores, calculates median and range
+
+**Round 2:**
+- Share Round 1 results (anonymous)
+- Experts see median scores and outliers
+- Ask experts to re-score, especially if they were outliers (optional: provide reasoning)
+
+**Round 3:**
+- Share Round 2 results
+- Experts make final adjustments
+- Converge on consensus scores (median or mean)
+
+**Convergence criteria:** Standard deviation of scores <1.5 points per criterion
+
+**Example:**
+
+| Option | Criterion | R1 Scores | R1 Median | R2 Scores | R2 Median | R3 Scores | R3 Median |
+|--------|-----------|-----------|-----------|-----------|-----------|-----------|-----------|
+| A | Cost | [5, 7, 9, 6] | 6.5 | [6, 7, 8, 6] | 6.5 | [6, 7, 7, 7] | **7** |
+
+**Advantage:** Avoids dominance by loudest voice, reduces groupthink, allows reflection time.
+
+### Nominal Group Technique (Structured Meeting)
+
+**When to use:** In-person or virtual meeting, need structured discussion to surface disagreements
+
+**Process:**
+
+1. **Silent generation (10 min):** Each person scores options independently
+2. **Round-robin sharing (20 min):** Each person shares one score and rationale (no debate yet)
+3. **Discussion (30 min):** Debate differences, especially outliers
+4. **Re-vote (5 min):** Independent re-scoring after hearing perspectives
+5. **Aggregation:** Calculate final scores (mean or median)
+
+**Facilitation tips:**
+- Enforce "no interruptions" during round-robin
+- Time-box discussion to avoid analysis paralysis
+- Focus debate on criteria with widest score variance
+
+### Handling Persistent Disagreement
+
+**Scenario:** After multiple rounds, stakeholders still disagree on weights or scores.
+
+**Options:**
+
+**1. Separate matrices by stakeholder group:**
+
+Run matrix for Engineering priorities, Sales priorities, Executive priorities separately. Present all three results. Highlight where recommendations align vs. differ.
+
+**2. Escalate to decision-maker:**
+
+Present divergence transparently: "Engineering weights Performance at 60%, Sales weights Cost at 50%. Under Engineering weights, Option A wins. Under Sales weights, Option B wins. Recommendation: [Decision-maker] must adjudicate priority trade-off."
+
+**3. Multi-criteria satisficing:**
+
+Instead of optimizing weighted sum, find option that meets minimum thresholds on all criteria. This avoids weighting debate.
+
+**Example:** Option must score ≥7 on Performance AND ≤$50K cost AND ≥6 on Ease of Use. Find options that satisfy all constraints.
+
+---
+
+## 6. Matrix Variations & Extensions
+
+### Weighted Pros/Cons Matrix
+Hybrid: Add "Key Pros/Cons/Dealbreakers" columns to matrix for qualitative context alongside quantitative scores.
+
+### Multi-Phase Decision Matrix
+**Phase 1:** High-level filter (simple criteria) → shortlist top 3
+**Phase 2:** Deep-dive (detailed criteria) → select winner
+Avoids analysis paralysis by not deep-diving on all options upfront.
+
+### Risk-Adjusted Matrix
+For uncertain scores, use expected value: (Optimistic + 4×Most Likely + Pessimistic) / 6
+Accounts for score uncertainty in final weighted total.
+
+---
+
+## 7. Common Failure Modes & Recovery
+
+| Failure Mode | Symptoms | Recovery |
+|--------------|----------|----------|
+| **Post-Rationalization** | Oddly specific weights, generous scores for preferred option | Assign weights BEFORE scoring, use third-party facilitator |
+| **Analysis Paralysis** | >10 criteria, endless tweaking, winner changes repeatedly | Set deadline, time-box criteria (5 max), use satisficing rule |
+| **Garbage In, Garbage Out** | Scores are guesses, no data sources, false confidence | Flag uncertainties, gather real data, acknowledge limits |
+| **Criterion Soup** | Overlapping criteria, scorer confusion | Consolidate redundant criteria, define each clearly |
+| **Spreadsheet Error** | Calculation mistakes, weights don't sum to 100% | Use templates with formulas, peer review calculations |
+
+---
+
+## 8. When to Abandon the Matrix
+
+Despite best efforts, sometimes a decision matrix is not the right tool:
+
+**Abandon if:**
+
+1. **Purely emotional decision:** Choosing baby name, selecting wedding venue (no "right" answer)
+   - **Use instead:** Gut feel, user preference vote
+
+2. **Single dominant criterion:** Only cost matters, everything else is noise
+   - **Use instead:** Simple cost comparison table
+
+3. **Decision already made:** Political realities mean decision is predetermined
+   - **Use instead:** Document decision rationale (not fake analysis)
+
+4. **Future is too uncertain:** Can't meaningfully score because context will change dramatically
+   - **Use instead:** Scenario planning, real options analysis, reversible pilot
+
+5. **Stakeholders distrust process:** Matrix seen as "math washing" to impose decision
+   - **Use instead:** Deliberative dialog, voting, or delegated authority
+
+**Recognize when structured analysis adds value vs. when it's theater.** Decision matrices work best when:
+- Multiple alternatives genuinely exist
+- Trade-offs are real and must be balanced
+- Stakeholders benefit from transparency
+- Data is available or can be gathered
+- Decision is reversible if matrix misleads
+
+If these don't hold, consider alternative decision frameworks.
--- a/skills/decision-matrix/resources/template.md
+++ b/skills/decision-matrix/resources/template.md
@@ -0,0 +1,370 @@
+# Decision Matrix Template
+
+## Workflow
+
+Copy this checklist and track your progress:
+
+```
+Decision Matrix Progress:
+- [ ] Step 1: Frame the decision
+- [ ] Step 2: Identify criteria and assign weights
+- [ ] Step 3: Score alternatives
+- [ ] Step 4: Calculate and analyze results
+- [ ] Step 5: Validate and deliver
+```
+
+**Step 1: Frame the decision** - Clarify decision context, list alternatives, identify must-haves. See [Decision Framing](#decision-framing).
+
+**Step 2: Identify criteria and assign weights** - Determine what factors matter, assign percentage weights. See [Criteria Identification](#criteria-identification) and [Weighting Techniques](#weighting-techniques).
+
+**Step 3: Score alternatives** - Rate each option on each criterion (1-10 scale). See [Scoring Guidance](#scoring-guidance).
+
+**Step 4: Calculate and analyze results** - Compute weighted scores, rank options, check sensitivity. See [Matrix Calculation](#matrix-calculation) and [Interpretation](#interpretation).
+
+**Step 5: Validate and deliver** - Quality check against [Quality Checklist](#quality-checklist), deliver with recommendation.
+
+---
+
+## Decision Framing
+
+### Input Questions
+
+Ask user to clarify:
+
+**1. Decision context:**
+- What are we deciding? (Be specific: "Choose CRM platform" not "improve sales")
+- Why now? (Triggering event, deadline, opportunity)
+- What happens if we don't decide or choose wrong?
+
+**2. Alternatives:**
+- What are ALL the options we're considering? (Get exhaustive list)
+- Include "do nothing" or status quo as an option if relevant
+- Are these mutually exclusive or can we combine them?
+
+**3. Must-have requirements (filters):**
+- Are there absolute dealbreakers? (Budget cap, compliance requirement, technical constraint)
+- Which options fail must-haves and can be eliminated immediately?
+- Distinguish between "must have" (filter) and "nice to have" (criterion)
+
+**4. Stakeholders:**
+- Who needs to agree with this decision?
+- Who will be affected by it?
+- Do different stakeholders have different priorities?
+
+### Framing Template
+
+```markdown
+## Decision Context
+- **Decision:** [Specific choice to be made]
+- **Timeline:** [When decision needed by]
+- **Stakeholders:** [Who needs to agree]
+- **Consequences of wrong choice:** [What we risk]
+
+## Alternatives
+1. [Option A name]
+2. [Option B name]
+3. [Option C name]
+4. [Option D name - if applicable]
+5. [Do nothing / Status quo - if applicable]
+
+## Must-Have Requirements (Pass/Fail)
+- [ ] [Requirement 1] - All options must meet this
+- [ ] [Requirement 2] - Eliminates options that don't pass
+- [ ] [Requirement 3] - Non-negotiable constraint
+
+**Options eliminated:** [List any that fail must-haves]
+**Remaining options:** [List that pass filters]
+```
+
+---
+
+## Criteria Identification
+
+### Process
+
+**Step 1: Brainstorm factors**
+
+Ask: "What makes one option better than another?"
+
+Common categories:
+- **Cost:** Upfront, ongoing, total cost of ownership
+- **Performance:** Speed, quality, reliability, scalability
+- **Risk:** Implementation risk, reversibility, vendor lock-in
+- **Strategic:** Alignment with goals, competitive advantage, future flexibility
+- **Operational:** Ease of use, maintenance, training, support
+- **Stakeholder:** Team preference, customer impact, executive buy-in
+
+**Step 2: Validate criteria**
+
+Each criterion should be:
+- [ ] **Measurable or scorable** (can assign 1-10 rating)
+- [ ] **Differentiating** (options vary on this dimension)
+- [ ] **Relevant** (actually matters for this decision)
+- [ ] **Independent** (not redundant with other criteria)
+
+**Remove:**
+- Criteria where all options score the same (no differentiation)
+- Duplicate criteria that measure same thing
+- Criteria that should be must-haves (pass/fail, not scored)
+
+**Step 3: Keep list manageable**
+
+- **Ideal:** 4-7 criteria (enough to capture trade-offs, not overwhelming)
+- **Minimum:** 3 criteria (otherwise too simplistic)
+- **Maximum:** 10 criteria (beyond this, hard to weight meaningfully)
+
+If you have >10 criteria, group related ones into categories with sub-criteria.
+
+### Criteria Template
+
+```markdown
+## Evaluation Criteria
+
+| # | Criterion | Definition | How We'll Measure |
+|---|-----------|------------|-------------------|
+| 1 | [Name] | [What this measures] | [Data source or scoring approach] |
+| 2 | [Name] | [What this measures] | [Data source or scoring approach] |
+| 3 | [Name] | [What this measures] | [Data source or scoring approach] |
+| 4 | [Name] | [What this measures] | [Data source or scoring approach] |
+| 5 | [Name] | [What this measures] | [Data source or scoring approach] |
+```
+
+---
+
+## Weighting Techniques
+
+### Technique 1: Direct Allocation (Fastest)
+Solo decision or aligned stakeholders. Assign percentages summing to 100%. Start with most important (30-50%), avoid weights <5%, round to 5% increments.
+
+**Example:** Cost 30%, Performance 25%, Ease of use 20%, Risk 15%, Team preference 10% = 100%
+
+### Technique 2: Pairwise Comparison (Most Rigorous)
+Difficult to weight directly or need justification. Compare each pair ("Is A more important than B?"), tally wins, convert to percentages.
+
+**Example:** Cost vs Performance → Performance wins. After all pairs, Performance has 4 wins (40%), Cost has 2 wins (20%), etc.
+
+### Technique 3: Stakeholder Averaging (Group Decisions)
+Multiple stakeholders with different priorities. Each assigns weights independently, then average. Large variance reveals disagreement → discuss before proceeding.
+
+**Example:** If stakeholders assign Cost weights of 40%, 20%, 30% → Average is 30%, but variance suggests need for alignment discussion.
+
+---
+
+## Scoring Guidance
+
+### Scoring Scale
+
+**Use 1-10 scale** (better granularity than 1-5):
+
+- **10:** Exceptional, best-in-class
+- **8-9:** Very good, exceeds requirements
+- **6-7:** Good, meets requirements
+- **4-5:** Acceptable, meets minimum
+- **2-3:** Poor, below requirements
+- **1:** Fails, unacceptable
+
+**Consistency tips:**
+- Define what 10 means for each criterion before scoring
+- Score all options on one criterion at a time (easier to compare)
+- Use half-points (7.5) if needed for precision
+
+### Scoring Process
+
+**For objective criteria (cost, speed, measurable metrics):**
+
+1. Get actual data (quotes, benchmarks, measurements)
+2. Convert to 1-10 scale using formula:
+   - **Lower is better** (cost, time): Score = 10 × (Best value / This value)
+   - **Higher is better** (performance, capacity): Score = 10 × (This value / Best value)
+
+**Example (Cost - lower is better):**
+- Option A: $50K → Score = 10 × ($30K / $50K) = 6.0
+- Option B: $30K → Score = 10 × ($30K / $30K) = 10.0
+- Option C: $40K → Score = 10 × ($30K / $40K) = 7.5
+
+**For subjective criteria (ease of use, team preference):**
+
+1. Define what 10, 7, and 4 look like for this criterion
+2. Score relative to those anchors
+3. Document reasoning/assumptions
+
+**Example (Ease of Use):**
+- 10 = No training needed, intuitive UI, users productive day 1
+- 7 = 1-week training, moderate learning curve
+- 4 = Significant training (1 month), complex UI
+
+**Calibration questions:**
+- Would I bet money on this score being accurate?
+- Is this score relative to alternatives or absolute?
+- What would change this score by ±2 points?
+
+### Scoring Template
+
+```markdown
+## Scoring Matrix
+
+| Option | Criterion 1 (Weight%) | Criterion 2 (Weight%) | Criterion 3 (Weight%) | Criterion 4 (Weight%) |
+|--------|-----------------------|-----------------------|-----------------------|-----------------------|
+| Option A | [Score] | [Score] | [Score] | [Score] |
+| Option B | [Score] | [Score] | [Score] | [Score] |
+| Option C | [Score] | [Score] | [Score] | [Score] |
+
+**Data sources and assumptions:**
+- Criterion 1: [Where scores came from, what assumptions]
+- Criterion 2: [Where scores came from, what assumptions]
+- Criterion 3: [Where scores came from, what assumptions]
+- Criterion 4: [Where scores came from, what assumptions]
+```
+
+---
+
+## Matrix Calculation
+
+### Calculation Process
+
+**For each option:**
+1. Multiply criterion score by criterion weight
+2. Sum all weighted scores
+3. This is the option's total score
+
+**Formula:** Total Score = Σ (Criterion Score × Criterion Weight)
+
+**Example:**
+
+| Option | Cost (30%) | Performance (40%) | Risk (20%) | Ease (10%) | **Total** |
+|--------|-----------|------------------|-----------|-----------|---------|
+| Option A | 7 × 0.30 = 2.1 | 9 × 0.40 = 3.6 | 6 × 0.20 = 1.2 | 8 × 0.10 = 0.8 | **7.7** |
+| Option B | 9 × 0.30 = 2.7 | 6 × 0.40 = 2.4 | 8 × 0.20 = 1.6 | 6 × 0.10 = 0.6 | **7.3** |
+| Option C | 5 × 0.30 = 1.5 | 8 × 0.40 = 3.2 | 7 × 0.20 = 1.4 | 9 × 0.10 = 0.9 | **7.0** |
+
+**Winner: Option A (7.7)**
+
+### Final Matrix Template
+
+```markdown
+## Decision Matrix Results
+
+| Option | [Criterion 1] ([W1]%) | [Criterion 2] ([W2]%) | [Criterion 3] ([W3]%) | [Criterion 4] ([W4]%) | **Weighted Total** | **Rank** |
+|--------|----------------------|----------------------|----------------------|----------------------|-------------------|----------|
+| [Option A] | [S] ([S×W1]) | [S] ([S×W2]) | [S] ([S×W3]) | [S] ([S×W4]) | **[Total]** | [Rank] |
+| [Option B] | [S] ([S×W1]) | [S] ([S×W2]) | [S] ([S×W3]) | [S] ([S×W4]) | **[Total]** | [Rank] |
+| [Option C] | [S] ([S×W1]) | [S] ([S×W2]) | [S] ([S×W3]) | [S] ([S×W4]) | **[Total]** | [Rank] |
+
+**Weights:** [Criterion 1] ([W1]%), [Criterion 2] ([W2]%), [Criterion 3] ([W3]%), [Criterion 4] ([W4]%)
+
+**Scoring scale:** 1-10 (10 = best)
+```
+
+---
+
+## Interpretation
+
+### Analysis Checklist
+
+After calculating scores, analyze:
+
+**1. Clear winner vs close call**
+- [ ] **Margin >10%:** Clear winner, decision is robust
+- [ ] **Margin 5-10%:** Moderate confidence, validate assumptions
+- [ ] **Margin <5%:** Toss-up, need more data or stakeholder discussion
+
+**2. Dominant criterion check**
+- [ ] Does one criterion drive entire decision? (accounts for >50% of score difference)
+- [ ] Is that appropriate or is weight too high?
+
+**3. Surprising results**
+- [ ] Does the winner match gut instinct?
+- [ ] If not, what does the matrix reveal? (Trade-off you hadn't considered)
+- [ ] Or are weights/scores wrong?
+
+**4. Sensitivity questions**
+- [ ] If we swapped top two criterion weights, would winner change?
+- [ ] If we adjusted one score by ±1 point, would winner change?
+- [ ] Which scores are most uncertain? (Could they change with more data)
+
+### Recommendation Template
+
+```markdown
+## Recommendation
+
+**Recommended Option:** [Option name] (Score: [X.X])
+
+**Rationale:**
+- [Option] scores highest overall ([X.X] vs [Y.Y] for runner-up)
+- Key strengths: [What it excels at based on criterion scores]
+- Acceptable trade-offs: [Where it scores lower but weight is low enough]
+
+**Key Trade-offs:**
+- **Winner:** Strong on [Criterion A, B] ([X]% of total weight)
+- **Runner-up:** Strong on [Criterion C] but weaker on [Criterion A]
+- **Decision driver:** [Criterion A] matters most ([X]%), where [Winner] excels
+
+**Confidence Level:**
+- [ ] **High (>10% margin):** Decision is robust to reasonable assumption changes
+- [ ] **Moderate (5-10% margin):** Sensitive to [specific assumption], recommend validating
+- [ ] **Low (<5% margin):** Effectively a tie, consider [additional data needed] or [stakeholder input]
+
+**Sensitivity:**
+- [Describe any sensitivity - e.g., "If Risk weight increased from 20% to 35%, Option B would win"]
+
+**Next Steps:**
+1. [Immediate action - e.g., "Get final pricing from vendor"]
+2. [Validation - e.g., "Confirm technical feasibility with engineering"]
+3. [Communication - e.g., "Present to steering committee by [date]"]
+```
+
+---
+
+## Quality Checklist
+
+Before delivering, verify:
+
+**Decision framing:**
+- [ ] Decision is specific and well-defined
+- [ ] All viable alternatives included
+- [ ] Must-haves clearly separated from nice-to-haves
+- [ ] Stakeholders identified
+
+**Criteria:**
+- [ ] 3-10 criteria (enough to capture trade-offs, not overwhelming)
+- [ ] Each criterion is measurable/scorable
+- [ ] Criteria differentiate between options (not all scored the same)
+- [ ] No redundancy between criteria
+- [ ] Weights sum to 100%
+- [ ] Weight distribution reflects true priorities
+
+**Scoring:**
+- [ ] Scores use consistent 1-10 scale
+- [ ] Objective criteria based on data (not guesses)
+- [ ] Subjective criteria have clear definitions/anchors
+- [ ] Assumptions and data sources documented
+- [ ] Scores are defensible (could explain to stakeholder)
+
+**Analysis:**
+- [ ] Weighted scores calculated correctly
+- [ ] Options ranked by total score
+- [ ] Sensitivity analyzed (close calls identified)
+- [ ] Recommendation includes rationale and trade-offs
+- [ ] Next steps identified
+
+**Communication:**
+- [ ] Matrix table is clear and readable
+- [ ] Weights shown in column headers
+- [ ] Weighted scores shown (not just raw scores)
+- [ ] Recommendation stands out visually
+- [ ] Assumptions and limitations noted
+
+---
+
+## Common Pitfalls
+
+| Pitfall | Fix |
+|---------|-----|
+| **Too many criteria (>10)** | Consolidate related criteria into categories |
+| **Redundant criteria** | Combine criteria that always score the same |
+| **Arbitrary weights** | Use pairwise comparison or stakeholder discussion |
+| **Scores are guesses** | Gather data for objective criteria, define anchors for subjective |
+| **Confirmation bias** | Weight criteria BEFORE scoring options |
+| **Ignoring sensitivity** | Always check if small changes flip the result |
+| **False precision** | Match precision to confidence level |
+| **Missing "do nothing"** | Include status quo as an option to evaluate |