Strong master data management interviews test decision-making, not memorized vendor screens. Candidates should explain assumptions, trade-offs, evidence, failure modes, ownership, and operational controls. EBX-specific details depend on the deployed product version and configuration; verify them in current TIBCO documentation.

1. What problem does MDM solve?

MDM coordinates authoritative data for shared entities such as customers, products, suppliers, or locations. It establishes governance, identifiers, quality rules, matching, survivorship, hierarchies, distribution, and lifecycle controls. It does not automatically create one perfect truth.

2. Registry, consolidation, coexistence, or centralized?

Describe where mastered attributes live, which systems can author them, latency and synchronization needs, identifiers, failure recovery, and migration. These are architecture patterns, not maturity rankings.

3. How do you define a golden record?

State the entity, attributes, sources, provenance, survivorship rules, confidence, exceptions, and stewardship. “Golden” means approved for specified uses, not universally correct.

4. How do matching and deduplication work?

Use deterministic and probabilistic signals appropriate to the entity. Measure precision, recall, review workload, subgroup effects, false merges, false splits, and reversibility. Protect sensitive attributes and prevent feedback from unreviewed merges.

5. What are survivorship rules?

Rules select or derive mastered values using source authority, recency, validation, completeness, confidence, or stewardship. Record lineage and allow exceptions; “most recent wins” is unsafe when timestamps or sources are unreliable.

6. How do hierarchies differ from relationships?

Hierarchies organize parent-child structures with validity periods and constraints. Relationships may be many-to-many, directional, typed, and temporal. Test cycles, orphan nodes, effective dates, and authorization.

7. How do you measure data quality?

Define rules from intended use: validity, completeness, uniqueness, consistency, timeliness, and referential integrity. Report denominator, severity, exceptions, trend, and outcome impact. See data-cleaning best practices.

8. What is stewardship?

Stewards resolve defined exceptions under decision rights and service objectives. They need evidence, workload limits, escalation, conflict handling, and audit trails—not responsibility for every defect.

9. How do you secure mastered data?

Classify attributes, minimize collection, enforce least privilege and segregation, protect administrative and integration identities, encrypt appropriately, log access and changes, test deletion, and plan incidents and recovery.

10. How do you design integration?

Specify contracts, identifiers, schemas, validation, ordering, idempotency, retries, dead-letter handling, reconciliation, versioning, and rollback. “Real time” requires a freshness objective and failure behavior.

11. How do you migrate to MDM?

Profile sources, define ownership and target semantics, map and cleanse, test matching, rehearse backfill, reconcile, dual-run where needed, cut over with acceptance criteria, and retain rollback.

12. How would you troubleshoot a production defect?

Trace the record from source through validation, matching, survivorship, workflow, publication, and consumer. Preserve evidence, contain impact, correct safely, reconcile downstream systems, and address root cause.

Connect these answers to MDM domains and patterns and a sample governance policy. In an interview, use a concrete scenario and quantify the trade-off.

Originally published February 25, 2021; technically reviewed and substantially updated September 4, 2026.