Model Drift Monitoring Dashboard Builder

You are an ML analytics lead designing model monitoring for a production machine learning system. Build a model drift monitoring dashboard and response plan for:

**Model use case:** [fraud scoring, churn prediction, lead scoring, recommendations, etc.]
**Model type:** [classification / regression / ranking / generative]
**Prediction frequency:** [real-time / hourly / daily / batch]
**Key input features:** [important features]
**Target or outcome delay:** [when ground truth becomes available]
**Business risk of bad predictions:** [financial, customer, compliance, safety]
**Available data:** [training baseline, live predictions, labels, segments]

Return:

1. **Monitoring Goals** — what the dashboard must detect and why
2. **Metric Framework** — input drift, prediction drift, performance drift, data quality, segment health
3. **Dashboard Layout** — panels, filters, charts, and alert summaries
4. **Threshold Recommendations** — warning and critical levels with rationale
5. **Segment Analysis Plan** — geography, plan, device, customer type, or other relevant cohorts
6. **Investigation Playbook** — steps to diagnose drift and decide whether to retrain, roll back, or adjust thresholds
7. **Stakeholder Summary Template** — plain-English weekly update for non-technical leaders

Make the plan practical for teams that may not have perfect labels immediately.