Metrics Review

Help me review metrics and identify what needs attention for: $ARGUMENTS

Instructions

Conduct a structured metrics review. The goal is not to report numbers — it's to identify what's changed, why it matters, and what action to take.

Process

1. Scan for changes — What metrics moved significantly (up or down) since last review?
2. Diagnose causes — For each change, identify likely causes (release, seasonality, external event, instrumentation)
3. Assess impact — Quantify who's affected and how much it matters
4. Recommend action — For each finding: investigate further, fix now, monitor, or accept

Tools to Use

• Query available data sources (BigQuery, dashboards, analytics tools) via MCP
• Pull recent release notes or deploy logs for correlation
• Check for known incidents or outages that could explain metric changes
• Compare across dimensions: platform, tier, region, app version

Output Format

# Metrics Review: [Area/Date]

## Headlines
- [Most important metric change — 1 sentence]
- [Second most important — 1 sentence]
- [Third — 1 sentence]

## Dashboard

| Metric | Current | Previous | Change | Trend | Status |
|--------|---------|----------|--------|-------|--------|
| [metric] | [value] | [value] | [delta] | [direction] | OK / Watch / Act |

## Deep Dives

### [Metric that needs attention]
- **What changed:** [description with numbers]
- **Likely cause:** [hypothesis with evidence]
…

Principles

• Trends over snapshots — Always show direction, not just current value
• Slice before concluding — Break every aggregate by platform, version, tier before trusting it
• Correlation is not causation — A metric moving after a release is suspicious, not proven
• Verify the instrument — Before investigating a metric drop, confirm the logging didn't break
• Recommend, don't just report — Every finding should end with "so what should we do?"