Skip to content

florianbonnet14/ThePowerOfAnalytics_ClaudeSkills / kpi-tree-architect

kpi-tree-architect

Verified

Build comprehensive KPI trees from North Star metrics. Use when users need to decompose metrics into hierarchical driver structures, create MECE (Mutually Exclusive, Collectively Exhaustive) breakdowns, identify leverage points, map performance drivers, or prepare for root cause analysis. Essential for setting up analytical foundations after defining a North Star metric.

FileKindSize
SKILL.mdskill md10.2 KB
kpi_tree_architect_tutorial.mdmarkdown6.2 KB
references/advanced-techniques.mdmarkdown12.6 KB
references/examples.mdmarkdown14.4 KB
references/templates.mdmarkdown13.9 KB

references/advanced-techniques.md

View on GitHub
# Advanced KPI Tree Techniques

Advanced patterns and methods for complex KPI tree situations.

## Technique 1: Multiple Trees for Same North Star

**When to use**: When different valuable perspectives exist for the same metric.

### Example: Revenue

**Tree 1: By Customer Segment**
```
Total Revenue
├─ Enterprise Revenue
│   ├─ New Enterprise
│   └─ Existing Enterprise
├─ SMB Revenue
│   ├─ New SMB
│   └─ Existing SMB
└─ Self-Serve Revenue
    ├─ New Self-Serve
    └─ Existing Self-Serve
```

**Tree 2: By Product Line**
```
Total Revenue
├─ Core Product Revenue
│   ├─ Tier 1 (Basic)
│   ├─ Tier 2 (Pro)
│   └─ Tier 3 (Enterprise)
├─ Add-on Revenue
│   ├─ Feature Add-ons
│   └─ Capacity Add-ons
└─ Professional Services Revenue
    ├─ Implementation
    ├─ Training
    └─ Support
```

**Tree 3: By Lifecycle Stage**
```
Total Revenue
├─ New Customer Revenue (Month 1-3)
├─ Growing Customer Revenue (Month 4-12)
└─ Mature Customer Revenue (12+ months)
    ├─ Base Revenue
    ├─ Expansion Revenue
    └─ Renewal Revenue
```

### Usage Guidelines
- **Maintain all trees**: Each serves different analytical purposes
- **Use situationally**: Pick tree based on current question
  - Customer dynamics → Tree 1
  - Product portfolio decisions → Tree 2
  - Retention/expansion strategy → Tree 3
- **Cross-reference**: Sometimes combine insights from multiple trees
- **Don't force consistency**: Trees don't need to match structure

### Benefits
- Different stakeholders prefer different views
- Reveals different optimization opportunities
- Supports various decision types
- Comprehensive understanding of metric

## Technique 2: Time-Based Decomposition

**When to use**: For cohort analysis or when metric behavior changes over time.

### Pattern 1: Lifecycle Stages
```
Annual Customer Value
├─ Acquisition Cost (Month 0)
│   └─ Negative contribution
├─ Early Stage Value (M1-3)
│   ├─ Lower usage
│   └─ Higher support costs
├─ Ramped Value (M4-6)
│   ├─ Increased usage
│   └─ Lower support needs
├─ Mature Value (M7-12)
│   ├─ Stable usage
│   └─ Expansion opportunities
└─ Renewal Decision (Month 12)
    └─ 🔍 Customer success engagement
```

### Pattern 2: Seasonal Decomposition
```
Annual GMV (Gross Merchandise Value)
├─ Q1 GMV (Post-holiday low)
│   └─ 🔍 Post-holiday promotions
├─ Q2 GMV (Spring recovery)
│   └─ 🔍 Seasonal inventory
├─ Q3 GMV (Summer baseline)
└─ Q4 GMV (Holiday spike)
    ├─ Black Friday/Cyber Monday
    ├─ Holiday Shopping
    └─ 🔍 Holiday marketing campaigns
```

### Pattern 3: Cohort-Based
```
30-Day Retention
├─ Day 0-7 Retention
│   ├─ Day 1 Return Rate
│   ├─ Day 3 Activation Rate
│   └─ Day 7 Habit Formation
├─ Day 8-14 Retention
│   └─ Feature Discovery Rate
├─ Day 15-21 Retention
│   └─ Social Connection Rate
└─ Day 22-30 Retention
    └─ Power User Conversion
```

### Implementation Tips
- **Choose time granularity carefully**: Too fine = noise, too coarse = miss insights
- **Consider business cycles**: Align with natural business rhythms
- **Track cohorts separately**: Different cohorts may behave differently
- **Watch for temporal patterns**: Seasonality, day-of-week effects, etc.

## Technique 3: Hierarchical Segmentation

**When to use**: When multiple dimensions matter simultaneously.

### Pattern: Nested Segmentation
```
Total Monthly Active Users
├─ By Geography (Dimension 1)
│   ├─ North America
│   │   ├─ By Plan Tier (Dimension 2)
│   │   │   ├─ Free Users
│   │   │   │   ├─ By Activity Level (Dimension 3)
│   │   │   │   │   ├─ High Activity (Daily)
│   │   │   │   │   ├─ Medium Activity (Weekly)
│   │   │   │   │   └─ Low Activity (Monthly)
│   │   │   └─ Paid Users
│   │   │       └─ [Same activity breakdown]
│   ├─ Europe
│   │   └─ [Same tier/activity breakdown]
│   └─ Asia-Pacific
│       └─ [Same tier/activity breakdown]
```

### Dimensionality Guidelines

**Two dimensions**: Usually manageable
```
Metric
├─ Dimension 1: Segment A
│   ├─ Dimension 2: Sub-segment 1
│   └─ Dimension 2: Sub-segment 2
└─ Dimension 1: Segment B
    ├─ Dimension 2: Sub-segment 1
    └─ Dimension 2: Sub-segment 2
```

**Three dimensions**: Maximum recommended
```
Metric
└─ Dim 1 → Dim 2 → Dim 3
   (Total combinations = options across all dimensions)
```

**More than three**: Consider alternative approaches
- Use multiple separate trees
- Focus on most important dimension
- Use data exploration tools instead

### Warning Signs
- Tree becomes too wide (>7 branches at any level)
- Can't explain path in <1 minute
- Same segments repeated multiple times
- Analysis paralysis from too many combinations

### Better Alternative: Matrix View
Instead of deeply nested tree:
```
         | Free  | Paid  |
---------|-------|-------|
Americas | 1,000 | 5,000 |
Europe   | 800   | 3,000 |
APAC     | 500   | 2,000 |
```

## Technique 4: Ratio Decomposition

**When to use**: For efficiency metrics (costs, rates, productivities).

### Pattern 1: Cost Efficiency
```
Cost per Acquisition (CPA)
= Total Marketing Spend / Total Acquisitions

Decompose both:
├─ Total Marketing Spend
│   ├─ Fixed Costs
│   │   ├─ Brand/Awareness
│   │   ├─ Salaries/Overhead
│   │   └─ Tools/Infrastructure
│   └─ Variable Costs
│       ├─ Paid Ads Spend
│       │   ├─ Search Ads
│       │   ├─ Social Ads
│       │   └─ Display Ads
│       └─ Campaign-Specific Costs
└─ Total Acquisitions
    ├─ Organic Acquisitions
    │   ├─ Direct Traffic
    │   ├─ SEO
    │   └─ Word-of-Mouth
    └─ Paid Acquisitions
        └─ [Matches paid spend breakdown]

Insight: Can calculate CPA by channel
CPA(Search) = Search Spend / Search Acquisitions
```

### Pattern 2: Productivity Ratios
```
Revenue per Employee
= Total Revenue / Total Employees

├─ Total Revenue (Numerator)
│   ├─ Product Revenue
│   └─ Services Revenue
└─ Total Employees (Denominator)
    ├─ Sales Team
    │   └─ Revenue per Sales Rep
    ├─ Engineering Team
    │   └─ Revenue per Engineer
    └─ Operations Team
        └─ Revenue per Ops Staff

Can calculate departmental productivity
```

### Pattern 3: Conversion Efficiency
```
Application Success Rate
= Applications Approved / Applications Submitted

├─ Applications Approved (Numerator)
│   ├─ Auto-Approved
│   └─ Manual-Approved
│       └─ 🔍 Review process quality
└─ Applications Submitted (Denominator)
    ├─ Complete Applications
    │   ├─ Approved
    │   └─ Rejected
    │       └─ 🔍 Rejection reasons
    └─ Incomplete Applications
        └─ 🔍 Form complexity

Can identify where funnel loses people
```

### Analysis Benefits
- **Numerator focus**: Increase good outcomes
- **Denominator focus**: Reduce waste/costs
- **Ratio optimization**: Sometimes tradeoffs exist
- **Segment analysis**: Ratios often vary by segment

## Technique 5: Weighted Contribution Models

**When to use**: When segments contribute differently to overall metric.

### Pattern: Size × Performance
```
Total Customer Lifetime Value (LTV)
= Sum of (Segment Size × Segment LTV)

├─ Enterprise Contribution
│   = % Enterprise × Enterprise LTV
│   ├─ % Enterprise (15% of customers)
│   └─ Enterprise LTV ($50,000)
│   = 15% × $50K = $7,500 contribution
├─ Mid-Market Contribution
│   = % Mid-Market × Mid-Market LTV
│   ├─ % Mid-Market (35% of customers)
│   └─ Mid-Market LTV ($5,000)
│   = 35% × $5K = $1,750 contribution
└─ SMB Contribution
    = % SMB × SMB LTV
    ├─ % SMB (50% of customers)
    └─ SMB LTV ($500)
    = 50% × $500 = $250 contribution

Total Weighted LTV = $7,500 + $1,750 + $250 = $9,500
```

### Analysis Insights
- **Contribution != Size**: Small segment can contribute more
- **Optimization paths**: Improve performance OR shift mix
- **Strategic choices**: 
  - Improve Enterprise LTV (high impact per customer)
  - Grow Enterprise % (shift customer mix)
  - Both (multiplicative effect)

### Implementation
```
For any metric that's a weighted average:
1. Identify segments
2. Calculate segment size (%)
3. Calculate segment performance (metric value)
4. Multiply: Contribution = Size × Performance
5. Validate: Sum of contributions = Total metric
```

## Technique 6: Variance Analysis Trees

**When to use**: To understand what drives changes in metrics over time.

### Pattern: Metric Change Decomposition
```
Revenue Change (Q1 to Q2)
= ΔRevenue from Volume + ΔRevenue from Price

├─ Volume Effect
│   = (Q2 Units - Q1 Units) × Q1 Price
│   ├─ New Customer Units
│   ├─ Churned Customer Units (negative)
│   └─ Existing Customer Change
└─ Price Effect
    = Q2 Units × (Q2 Price - Q1 Price)
    ├─ Price Increases
    ├─ Promotional Discounts (negative)
    └─ Mix Shift Effects

Total Change = Volume Effect + Price Effect
```

### Mix vs Rate Analysis
```
Conversion Rate Change
= Mix Effect + Rate Effect

Example: Overall conversion dropped 5%
├─ Mix Effect (+3%): More traffic from lower-converting channels
└─ Rate Effect (-8%): All channels converting worse

Insight: Fix channel performance, not channel mix
```

## Technique 7: Leading vs Lagging Indicators

**When to use**: To create actionable, real-time trees.

### Pattern: Leading Indicator Focus
```
North Star: 6-Month Revenue (Lagging)

Leading Indicator Tree:
├─ Current Month Pipeline (1-month lead)
│   ├─ Qualified Opportunities
│   └─ Average Deal Size
├─ Demo Requests (2-month lead)
│   ├─ Demo Request Rate
│   └─ Demo-to-Opp Conversion
└─ Website Engagement (3-month lead)
    ├─ High-Intent Page Visits
    └─ Content Download Rate

Actionable today → Impacts revenue in 1-6 months
```

### Implementation
- **Mix timeframes**: Combine current, leading, and lagging
- **Lead time mapping**: Document how long each factor takes to impact North Star
- **Early warning system**: Leading indicators signal future problems
- **Intervention timing**: Know when you can still affect outcomes

## Technique 8: Constraint-Based Trees

**When to use**: When capacity or resource constraints matter.

### Pattern: Throughput Analysis
```
Orders Fulfilled per Day
= min(Capacity Constraints)

Constrained by:
├─ Warehouse Capacity
│   ├─ Available Labor Hours
│   │   └─ Current: 1,000 orders/day
│   └─ Space Utilization
│       └─ Current: 1,200 orders/day
├─ Packing Capacity
│   └─ Packing Station Throughput
│       └─ Current: 800 orders/day ← BOTTLENECK
└─ Shipping Capacity
    └─ Carrier Pickup Capacity
        └─ Current: 1,500 orders/day

Actual Throughput = 800 orders/day (limited by packing)

Next constraint after fixing packing: Labor (1,000/day)
```

### Bottleneck Optimization
1. Identify current constraint
2. Focus improvement efforts there
3. Track how bottleneck shifts
4. Prepare for next constraint

## Common Patterns Summary

| Pattern | When to Use | Structure |
|---------|------------|-----------|
| Multiple Trees | Different analytical perspectives needed | Parallel trees, same North Star |
| Time-Based | Temporal patterns or lifecycle matters | Time periods as segments |
| Hierarchical Segmentation | Multiple dimensions important | Nested dimensions (max 3) |
| Ratio Decomposition | Efficiency or rate metrics | Decompose numerator AND denominator |
| Weighted Contribution | Segments contribute differently | Size × Performance for each segment |
| Variance Analysis | Understanding metric changes | ΔMetric = ΔFactor1 + ΔFactor2 |
| Leading Indicators | Need predictive/actionable tree | Leading factors organized by lead time |
| Constraint-Based | Capacity/resource limitations | min(Constraints) model |

## Choosing the Right Technique

Ask these questions:
1. **One perspective enough?** → Single tree. **Multiple perspectives needed?** → Multiple trees
2. **Time patterns matter?** → Time-based decomposition
3. **Multiple dimensions critical?** → Hierarchical segmentation (careful with complexity)
4. **Ratio or efficiency metric?** → Ratio decomposition
5. **Segments perform differently?** → Weighted contribution
6. **Need to explain changes?** → Variance analysis
7. **Need early warning?** → Leading indicators
8. **Capacity constrained?** → Constraint-based

Often combine multiple techniques in one tree.
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 194kUpdated Claude CodeCodexVerifiedInstall