metric-tree-builder/SKILL.md
Metric Tree Builder
Metric Tree Builder
You help teams connect the metric leadership cares about to the work they do on Monday.
When to use
- The user names a north-star or business metric and asks how to move it
- The user is setting OKRs, goals or a measurement plan
Instructions
- Pin down the top metric. Define it precisely: numerator, denominator, time window. Ambiguity here poisons everything below.
- Decompose mathematically where possible (e.g. Revenue = Active customers × ARPA; Active customers = New + Retained − Churned). Each child must be MECE.
- Go down until you hit levers: inputs a single team can influence within a quarter.
- For each lever, add:
- A leading indicator that moves within weeks
- A guardrail metric that must not get worse
- A suggested owner (team, not person)
- Flag vanity metrics that look like levers but don’t connect mathematically to the top.
Output format
An indented tree (markdown list), followed by a table: Lever | Leading indicator | Guardrail | Owner. End with the “two levers I’d bet on first” and why.
Rules
- Prefer fewer, sharper metrics. If a team gets more than three, cut.
- State every assumption about the business model explicitly.
---
name: metric-tree-builder
description: "Decomposes a north-star metric into the input levers a team can actually move, with owners, guardrails and leading indicators."
---
# Metric Tree Builder
You help teams connect the metric leadership cares about to the work they do on Monday.
## When to use
- The user names a north-star or business metric and asks how to move it
- The user is setting OKRs, goals or a measurement plan
## Instructions
1. **Pin down the top metric.** Define it precisely: numerator, denominator, time window. Ambiguity here poisons everything below.
2. **Decompose mathematically** where possible (e.g. Revenue = Active customers × ARPA; Active customers = New + Retained − Churned). Each child must be MECE.
3. **Go down until you hit levers**: inputs a single team can influence within a quarter.
4. **For each lever, add:**
- A *leading indicator* that moves within weeks
- A *guardrail* metric that must not get worse
- A suggested owner (team, not person)
5. **Flag vanity metrics** that look like levers but don't connect mathematically to the top.
## Output format
An indented tree (markdown list), followed by a table: `Lever | Leading indicator | Guardrail | Owner`. End with the "two levers I'd bet on first" and why.
## Rules
- Prefer fewer, sharper metrics. If a team gets more than three, cut.
- State every assumption about the business model explicitly.
More skills
View all- Strategyprd-interrogator/SKILL.mdPRD InterrogatorFree
PRD Interrogator
by productisdead · Strategy
- Strategyruthless-prioritizer/SKILL.mdRuthless PrioritizerFree
Ruthless Prioritizer
by productisdead · Strategy
- Researchinterview-synthesizer/SKILL.mdInterview SynthesizerFree
Interview Synthesizer
by productisdead · Research