AgentSkillsCN

metrics-design

设计完整的指标框架,构建结果—中间—领先层级体系,搭配激活漏斗与参与度维度。适用于定义产品成功的衡量方式时使用。

SKILL.md
--- frontmatter
name: metrics-design
description: Design complete metrics frameworks with outcome-intermediate-leading hierarchy, activation funnels, and engagement dimensions. Use when defining how to measure product success.

Metrics Design Skill

Purpose

Provide structured guidance for designing complete, measurable metrics frameworks. Adapts to the PM's preferred hierarchy from their profile.

Metrics Hierarchy

Default: Outcome → Intermediate → Leading

Outcome Metrics measure business impact:

  • Revenue, retention, expansion
  • Measured over longer periods (quarterly, annually)
  • What the business ultimately cares about

Intermediate Metrics measure product outcomes:

  • Self-service rate, task completion rate, satisfaction score
  • Bridge between user behavior and business impact
  • Measured over medium periods (monthly, quarterly)

Leading Metrics measure user behavior:

  • Feature adoption, session frequency, engagement depth
  • Early signals that predict intermediate and outcome metrics
  • Measured continuously (daily, weekly)

Alternative Hierarchies

North Star → L1 → L2:

  • North Star: Single metric that captures core value delivered
  • L1: Direct drivers of the North Star
  • L2: Inputs to L1 metrics

Business → Product → Feature:

  • Business: Revenue and retention metrics
  • Product: User outcome metrics
  • Feature: Usage and adoption metrics

Activation Funnel

Default: Setup → Aha → Habit

StageDefinitionHow to Measure
SetupUser has completed initial configuration and can use the product% of users who complete setup; time to complete
AhaUser has experienced the core value for the first time% of users who perform the key value action; time to Aha
HabitUser has developed a repeated pattern of value extraction% of users who perform the value action [X] times in [Y] period

Defining Good Thresholds

  • Setup completion should be > 80% (if lower, onboarding is broken)
  • Aha rate should be > 50% of those who set up (if lower, value prop is unclear)
  • Habit rate should be > 30% of those who reached Aha (if lower, value doesn't sustain)
  • Time to Aha should be < 1 week for most B2B products

Engagement Dimensions

Default: Breadth, Depth, Intensity

DimensionWhat It MeasuresExample
BreadthHow many features are usedAvg features used per week
DepthHow frequently features are usedAvg actions per week
IntensityHow much time/effort is investedAvg session duration

User Segmentation

SegmentDefault DefinitionImplications
Casual1-2 sessions per weekAt risk of churn, need engagement
Core3-4 sessions per weekHealthy engagement, focus on deepening
Power5+ sessions per weekChampions, leverage for expansion

Tradeoff Metrics

Every optimization has a cost. Always define what might get worse:

If OptimizingWatch For Decline In
Adoption rateQuality of users adopted
Engagement depthBreadth of feature usage
Self-service rateSatisfaction with complex cases
Speed to marketCode quality, tech debt
Revenue per userTotal addressable users

Anti-Patterns

  • Metrics without baselines (can't measure improvement)
  • Outcome metrics only (no early warning system)
  • Vanity metrics (total users, page views without context)
  • Too many metrics (if everything is measured, nothing is prioritized)
  • Metrics without owners (who acts on this data?)
  • Metrics without thresholds (what's good vs bad?)