alexe-ev Design Product Metrics
Purpose
Help teams define a coherent metrics framework that connects product decisions to measurable outcomes.
Skill type
Conceptual skill
Use this skill when
- A product area lacks clear metrics
- A team is starting a new product or feature and needs instrumentation guidance
- Existing metrics are disconnected from goals or decisions
- OKRs need to be translated into measurable product metrics
Do not use this skill when
- Metrics are already defined and data needs to be analyzed (use analyze-funnel-retention-cohorts)
- The goal is experiment result analysis (use experimentation skills)
Required inputs
- Product area and context
- Business or product goals
Optional inputs
- Existing metrics or tracking
- User journey or funnel map
- OKRs
- Data infrastructure context
Upstream context
Works best when:
- Goals and OKRs are defined
- Product flow is understood
If upstream context is missing
Produce a metrics design framework with placeholder categories and flag that specific metrics require goal context.
Downstream handoff
Output can feed:
- build-decision-dashboard
- set-goals-okrs-kpis (metrics inform KR choices)
- define-success-metrics (experiment metrics derived from product metrics)
Instructions
- Map the user journey or funnel relevant to the product area.
- Identify the primary metric (the one metric that best reflects product success for this area).
- Identify secondary metrics (leading indicators, supporting signals).
- Define guardrail metrics (what must not degrade).
- Flag vanity metrics to avoid.
- Map each metric to a goal or OKR.
- Identify data requirements and instrumentation needs.
Output
Provide:
- Primary metric with definition
- Secondary metrics (3–5)
- Guardrail metrics
- Metric-to-goal mapping
- Instrumentation requirements
- Vanity metrics to avoid
- Gaps in current data coverage
Risks / caveats
- Measuring too many things dilutes focus — one primary metric per product area
- Vanity metrics (total signups, total pageviews) feel good and mean nothing without context
- If a metric can't inform a decision, it probably shouldn't be on the dashboard