Define what each dataset means
Document grain, timestamps, metric definitions, and owning systems before joining records. Prevent URL changes, attribution differences, or mixed timezones from becoming misleading performance conclusions.
- ✓ Source and field inventory
- ✓ Identity and URL normalization
- ✓ Refresh cadence and retention rules
Make quality visible
Track completeness, freshness, schema changes, and failed imports. Keep raw source references and transformation logic so downstream findings can be reproduced and corrected.
- ✓ Document the evidence and owner behind each decision.
- ✓ Agree acceptance criteria before implementation begins.
Scope and delivery decisions
Direct APIs and uploaded exports can both be scoped. Availability, history, quotas, and personal-data handling depend on the source.
- ✓ Confirm system access, dependencies, and data handling during discovery.
- ✓ Receive a scoped recommendation with responsibilities and verification steps.