SQL, data modeling, data quality, EDA, visualization · insight-0062
Agree a data contract at the producer boundary before building on a shared table
- Evidence type
- practitioner heuristic
- Difficulty
- medium
- Roles
- Data analyst, ML engineer, Data scientist
- Industries
- cross-industry
Summary: situation, action, reason, boundary
Situation
Your forecast or KPI starts consuming a table owned by another team, with no written agreement about it.
Action
Write a short contract with the producer: schema and types, meaning and units of each field, grain and keys, freshness and completeness expectations, owner and contact, and how breaking changes are announced and versioned. Enforce the checkable parts automatically on each load.
Reason
Contracts turn implicit assumptions into explicit, testable promises and give both sides a process for change.
Boundary
Cross-team tables, vendor feeds, event schemas and feature stores.
Symptom or risk
A column is renamed, its meaning changes or its refresh slips, and you find out from a wrong number in a leadership meeting.
Exceptions
Exploratory one-off analyses; write the assumptions in the analysis instead.
Trade-offs
Contracts add coordination overhead and can slow producers; keep them to the fields you consume.
Pitfalls
Writing a contract document that no automated check enforces.