13.19 Analytics

User stories

• As a CRO, I want a real-time risk dashboard.
• As a CEO, I want monthly unit-economics dashboards.
• As a data scientist, I want to define metrics once and reuse.

API requirements

• GET /analytics/dashboards/{id}.
• POST /analytics/metrics.
• GET /analytics/anomalies.

Data model

• metric, dashboard, experiment, anomaly.

UI screens

• BI tool (Metabase / Superset) bound to warehouse.
• Internal dashboards for live operational data.

Backend services

• Warehouse + ETL.
• BI layer.
• Anomaly detector.

External integrations

• Warehouse vendor (Snowflake / ClickHouse / Redshift).

Test cases

• Metric defined once, used by multiple dashboards.
• Anomaly alert triggers notification.
• Backfill on data correction.

Edge cases

• Stale dashboard with no alert.
• Sensitive PII in dashboard for unauthorised role.

Acceptance criteria

• Daily refresh works for all critical dashboards.
• Anomaly detection live for key metrics.
• Self-serve analytics for business users at year 2.