| name | metrics-review |
| description | Run a weekly, monthly, or quarterly metrics review: scorecard, trends, bright spots, concerns, and recommended actions. Use when investigating a spike or drop, comparing performance against targets, or preparing a metrics readout. Do NOT use for stakeholder status updates (stakeholder-update) or product strategy (product).
|
| license | MIT |
| allowed-tools | Read Write Glob Grep |
| argument-hint | <time period or metric focus> |
| metadata | {"author":"Carinya Parc","version":"0.1.0","owner":"delivery","review_cadence":"quarterly","work_shape":"review-and-gate","output_class":"decision-support"} |
Metrics review
You review product metrics, identify trends, and surface actionable insights.
Pass the time period or metric focus after the skill name.
See CONNECTORS.md for analytics connectors. For North Star /
L1 hierarchy, OKRs, cadences, and dashboard design, read
metrics-frameworks.md when needed.
Steps
1. Gather data
If product analytics is connected, pull key metrics, prior-period comparison,
targets, and segments. Otherwise ask the user for values, comparisons, and known
events (launches, outages, campaigns, seasonality).
Confirm: time period, metric focus (or full suite), targets, known events.
2. Organize
Structure around North Star → L1 health indicators (acquisition, activation,
engagement, retention, monetization, satisfaction) → L2 diagnostics. If the user
has no hierarchy, help identify North Star and key L1s first.
3. Analyze
For each key metric: current value, trend, vs target, rate of change, anomalies.
Note correlations, leading indicators, and segment drivers of aggregate moves.
4. Generate the review
- Summary — 2–3 sentences: overall health, notable changes, key callout
- Metric scorecard — table: Metric | Current | Previous | Change | Target | Status
- Trend analysis — what happened, likely why, one-time vs sustained
- Bright spots — beating targets, positive trends, strong segments
- Areas of concern — misses, early warnings, visibility gaps
- Recommended actions — investigations, experiments, investments, alerts
- Context and caveats — data quality, comparability events, missing metrics
5. Follow up
Offer deeper investigation, dashboard spec, experiment proposals, or a recurring
review template.
Quality rules
- Lead with the "so what" — absolute numbers without comparison are useless
- Attribution is uncertain; say so when correlating events to metric moves
- Every review should drive at least one action
- Focus on meaningful changes; small fluctuations are noise
Output format
Tables for the scorecard. Clear status indicators (On track / At risk / Miss).
Summary scannable in ~30 seconds.
