Mmcp.market

kpi-tree-architect skill

by florianbonnet14·florianbonnet14/ThePowerOfAnalytics_ClaudeSkills·29 stars·MIT

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.

A100/100content scan

Is the kpi-tree-architect skill safe?

Clean: nothing in its files matched our rules. We read 5 files in the folder on 2026-09-28.

No findings.

Install the kpi-tree-architect skill

A skill is a folder. Copy it into your agent's skills folder and the agent loads it when the task matches its description. 

git clone --depth 1 https://github.com/florianbonnet14/ThePowerOfAnalytics_ClaudeSkills.git /tmp/ThePowerOfAnalytics_ClaudeSkills
mkdir -p ~/.claude/skills
cp -r /tmp/ThePowerOfAnalytics_ClaudeSkills/kpi-tree-architect ~/.claude/skills/kpi-tree-architect
available in every project

In the Claude apps, zip the folder and upload it from the Skills settings. The folder on GitHub

The instructions your agent would load

SKILL.md as published, without the frontmatter. Read it on GitHub

KPI Tree Architect

Build hierarchical KPI structures that decompose North Star metrics into actionable drivers using MECE principles.

Core Principle: MECE Decomposition

Mutually Exclusive & Collectively Exhaustive

  • Mutually Exclusive: No overlapping drivers (no double-counting)
  • Collectively Exhaustive: No missing drivers (sum of parts = whole)

Visual analogy: Tangram puzzle pieces that don't overlap and cover the entire area.

Three Decomposition Methods

1. Mathematical Decomposition

Use when metric has clear mathematical formula.

Example: Upselling Rate = # Customers Buying Higher Tier / Total # Customers

  • Decomposes to numerator and denominator
  • Common patterns: Rate = Numerator/Denominator, Average = Sum/Count

2. Process Decomposition

Use when metric depends on sequential steps (funnels, workflows).

Example: # Customers Buying = Visits × Click Rate × Checkout Completion

  • Decomposes by funnel stages
  • Common patterns: Conversion funnels, customer journeys

3. Segmentation Decomposition

Use when metric varies significantly by groups.

Example: Total Visits = Organic + Email Campaign + Other Channels

  • Decomposes by segments
  • Common patterns: Customer types, product categories, regions, channels

Workflow Choice

First, always ask the user: "Would you like me to:

  1. Generate the complete tree at once (I'll build the full decomposition and present it), or
  2. Build it collaboratively (we'll work through each level together with your input)?"

Then follow the appropriate workflow below.

Workflow A: Complete Tree Generation

Use when user chooses option 1.

Process:

  1. Ask clarifying questions about the North Star metric and business context
  2. Build the complete tree using MECE principles
  3. Present the full tree with validation and insights
  4. Discuss and refine based on feedback

Steps:

  1. Understand North Star: Ask about metric definition, formula, business context
  2. Choose decomposition methods: Determine best approach for each level
  3. Build complete structure: Decompose to 3-5 levels
  4. Validate MECE: Check each level
  5. Identify influential factors: Mark qualitative drivers
  6. Present the tree with insights: Show full tree, validation, key findings

Workflow B: Collaborative Iterative Building

Use when user chooses option 2. Build the tree level by level with user input.

Iterative Process:

For each metric to decompose:

Step 1: Select Metric

Ask: "Which metric would you like to decompose next?"

  • Start with North Star metric
  • Then move to user-selected drivers from previous level

Step 2: Propose Decomposition Method

Based on the metric, propose 2-3 decomposition approaches:

Template: "For [Metric Name], I see [2-3] possible decomposition approaches:

Option 1: [Method Name] (e.g., Mathematical)

  • Break down as: [Component A] [operator] [Component B]
  • Best for: [when to use]
  • Example: [simple example]

Option 2: [Method Name] (e.g., Process)

  • Break down by: [stages/steps]
  • Best for: [when to use]
  • Example: [simple example]

Which approach makes more sense for your business? Or would you like me to recommend one?"

Step 3: Ask Clarifying Questions

Before proposing the decomposition, ask 1-2 questions:

  • About the business process or customer journey
  • About how the metric is calculated
  • About natural segments or stages
  • About data availability

Step 4: Propose Specific Decomposition

Present the specific breakdown:

"Based on [your answers], here's how I'd break down [Metric]:

[Metric Name] ├─ Component A: [Name and brief definition] ├─ Component B: [Name and brief definition] └─ Component C: [Name and brief definition]

MECE Validation:

  • Mutually Exclusive: [Why no overlap]
  • Collectively Exhaustive: [Why sum equals whole]

Does this breakdown make sense to you?"

Step 5: Validate with User

Ask: "Does this decomposition work for you? Any adjustments needed?"

If user approves → Move to next step If user wants changes → Refine and re-validate

Step 6: Continue or Stop

Ask: "Would you like to decompose any of these components further, or shall we work on a different branch?"

Options:

More skills from florianbonnet14/ThePowerOfAnalytics_ClaudeSkills

  • Aanalysis-plannerStructure analysis investigations before diving into data, preventing wasted time and ensuring thoroughness. Use when users need to plan any significant analysis, investigate KPI changes, respond to stakeholder questions, plan feature/experiment analysis, or when previous analyses were unfocused. Helps define clear goals, generate testable hypotheses, create systematic analysis roadmaps, identify required data, estimate timelines, and prevent analysis paralysis.
  • Aanalysis-results-collectorTransform completed analysis work into structured, communication-ready documentation by conducting guided conversations that extract key findings, evidence, and recommendations. Use when a user has completed an analysis (typically following an analysis plan created by the analysis-planner skill) and needs to document results, create findings summaries, build executive reports, or prepare analysis outcomes for stakeholder communication. This skill systematically gathers what was tested, what was found, and what should be done next, then generates a professional markdown document ready for distribution.
  • Aanalytics-orchestratorIntelligent entry point to the analytics ecosystem based on "The Power of Analytics" book by Florian Bonnet. Guides users to the right analytics agent based on their needs, recommends workflow sequences, and provides clear instructions for launching agents. Use when users need help choosing between analytics agents, want to start an analytics project but aren't sure where to begin, or ask "what can you help me with" in an analytics context. Also use when users mention analytics tasks like defining metrics, planning analysis, collecting results, creating presentations, or writing executive summaries.
  • Acohort-analysis-specialistTrack customer behavior over time by grouping customers based on shared characteristics or experiences. Use when measuring retention rates, understanding how customers evolve over time, comparing new vs old user behavior, evaluating product changes by cohort, identifying at-risk cohorts early, or calculating LTV by cohort. Essential for understanding retention, lifecycle patterns, and how product changes affect different user vintages.
  • Aexecutive-summary-writerTransform analysis findings into clear, actionable executive summaries using the MAIN framework (Motive, Answer, Impact, Next steps) and Pyramidal Principle. Use when completing any analysis that needs stakeholder buy-in, preparing for leadership meetings, documenting investigation findings, writing business reviews, creating post-mortems, proposing initiatives based on data, or communicating insights to executives.
  • Ainfluential-factors-detectiveIdentify and analyze qualitative factors that influence KPIs but can't be measured as numbers. Use when users need to find non-quantifiable drivers affecting metrics (like copy quality, design aesthetics, timing), plan A/B tests for subjective elements, bridge the gap between data and design/content decisions, understand why quantitative analysis is incomplete, optimize messaging or UX elements, or identify the actual levers to pull for improving KPIs at the leaf nodes of a KPI tree.
  • Anorth-star-metric-advisorStrategic framing and KPI setting specialist that helps identify and validate the single most important metric capturing customer value delivery and predicting business success. Use when starting a new team or product, when existing product lacks clear success metric, when team seeks to define what success means, or when metrics don't align with customer value and/or business goals.
  • Apresentation-builderTransform analysis documents into structured presentation slide plans. Use when users need to convert markdown analysis documents into presentation decks, create slide-by-slide breakdowns with specific slide types and content structure, or plan presentation flow from analytical content. Outputs detailed slide specifications including slide types (Cover, Executive Summary, Text, Chart, Section) and all required content elements for each slide.
  • Asegmentation-expertIdentify, analyze, and act on customer segments to uncover hidden patterns and understand behavior differences. Use when investigating KPI changes to understand which customers drove them, planning targeted initiatives, building customer personas, or analyzing performance differences across groups. Helps select meaningful segmentation dimensions, create actionable segments, analyze performance differences, calculate segment contributions, and generate segment-specific recommendations.

All agent skills → · MCP servers