/setup-metrics

/setup-metrics -- Product Metrics Dashboard Design

Design a comprehensive metrics framework for your product or feature — from selecting the right North Star to defining alert thresholds that catch problems early.

Invocation

/setup-metrics SaaS project management tool
/setup-metrics New checkout flow we just launched
/setup-metrics             # asks what you're measuring

Workflow

Step 1: Understand What to Measure

Ask the user:

• What product or feature area are you setting up metrics for?
• What stage is it in? (pre-launch, recently launched, mature)
• What are the current business goals or OKRs?
• Do you have existing metrics? What's missing or broken?
• What analytics tools are you using? (helps tailor implementation advice)

Step 2: Define the Metrics Framework

Apply the product-metrics skill (in pm-data-analytics):

Ask the user upfront: Is this for an AI product or feature? The skill adds a four-layer AI metrics stack (model quality, operational, product-level, business) when yes.

North Star Metric:

• Identify the single metric that best captures the value your product delivers to users
• Validate against criteria: measures value delivery, is a leading indicator, is actionable
• Define the metric precisely (formula, data source, time window)

Input Metrics (3-5):

• Identify the levers that drive the North Star
• Each input metric should be directly actionable by a team
• Map the causal chain: Input → North Star → Business Outcome

Health Metrics (3-5):

• Metrics that should stay stable — if they degrade, something is wrong
• Examples: error rates, latency, support ticket volume, NPS, churn rate
• Define "healthy" ranges and degradation thresholds

Counter-Metrics (1-2):

• Metrics that could indicate you're optimizing the wrong way
• Example: if North Star is "daily active users", counter-metric is "session quality" to prevent empty engagement

Step 3: Design Alert Thresholds

For each metric:

Metric	Green	Yellow	Red	Check Frequency
[metric]	[healthy range]	[warning]	[critical]	[daily/weekly]

• Yellow: Investigate — something may be off
• Red: Act immediately — page someone or escalate

Step 4: Create Dashboard Spec

## Metrics Dashboard: [Product/Feature]

**North Star**: [metric name]
**Definition**: [precise formula]
**Current value**: [if known]
**Target**: [goal]

### Input Metrics
| Metric | Definition | Owner | Target | Current |
|--------|-----------|-------|--------|---------|

### Health Metrics
| Metric | Healthy Range | Yellow Threshold | Red Threshold |
|--------|-------------|-----------------|---------------|

### Counter-Metrics
| Metric | Why It Matters | Watch For |
|--------|---------------|-----------|

### Metrics Tree
North Star: [metric]
├── Input: [metric 1] → driven by [team/action]
├── Input: [metric 2] → driven by [team/action]
├── Input: [metric 3] → driven by [team/action]
└── Counter: [metric] → watch for [degradation signal]

### Implementation Notes
- Data sources: [where each metric comes from]
- Refresh frequency: [real-time / hourly / daily]
- Tool recommendations: [based on user's stack]

### Review Cadence
- **Daily**: Glance at North Star and health metrics
- **Weekly**: Review input metrics trends, discuss in team standup
- **Monthly**: Deep dive — are inputs driving the North Star as expected?
- **Quarterly**: Reassess the metrics framework itself

Save as a markdown file to the user's workspace.

Step 5: Offer Next Steps

• "Want me to write SQL queries to compute these metrics?"
• "Should I create OKRs based on this metrics framework?"
• "Want me to build a cohort analysis to set realistic baselines?"
• "Should I set up a weekly metrics review template?"

Notes

• A good North Star is rare — most teams pick vanity metrics. Push for a metric that captures user value delivered, not just engagement
• Input metrics should be MECE (mutually exclusive, collectively exhaustive) in explaining the North Star
• If the product is pre-launch, define metrics now but note that baselines will need calibration after launch
• Counter-metrics prevent Goodhart's Law — when a metric becomes a target, it ceases to be a good metric
• Recommend starting with fewer metrics, well-instrumented, over a sprawling dashboard nobody checks