Data governance is the system that decides who may change data, how changes are documented, and how the business proves both to auditors. It spans enterprise data catalogs, metadata management, data lineage, master data management (mdm), privacy controls, and the operating models around them. Demand keeps rising because analytics and AI consumption now ride on governed foundations, and regulators ask for evidence rather than assurances. Gartner's 2025 Magic Quadrant for Metadata Management Solutions frames the shift: the market is moving from augmented data catalogs to metadata orchestration platforms that reach across the whole estate, and data and analytics leaders are expected to weigh that evolution as they start data and AI journeys . Gartner defines master data management itself as a discipline in which business and IT work together to keep official shared master data uniform, accurate, and accountable . Hiring for this craft means finding people who have run those operating models, not just installed the software.
Challenges in Data Governance Recruiting
Compliance pressure turned data lineage from documentation into release gate
The UK GDPR's seven principles read as a governance requirements list. Personal data must be processed lawfully, fairly, and transparently; collected for specified purposes; minimised; kept accurate; and retained no longer than necessary, while the controller must be able to demonstrate compliance with all of it under the accountability principle . Failures reach the highest tier of fines, up to £17.5 million or 4% of worldwide turnover . Each principle lands as an engineering question. Lawfulness means a documented lawful basis attached to every pipeline; storage limitation means deletion logic that actually runs; accuracy means someone owns correction. Data lineage is the evidence layer for all of it. Snowflake's documentation positions lineage as impact analysis and troubleshooting, and explicitly as compliance, tracking how sensitive data flows between objects . A team that redraws lineage after a regulator asks has already failed the accountability principle. The interview question that matters is whether the candidate built lineage as a design property or drew it after the fact, and a subject-access request traced end to end exposes the difference within minutes.
Master data management (MDM) programs hire business stewards, not just architects
Gartner's definition of master data management is blunt about the operating model: business and IT work together to ensure the uniformity, accuracy, stewardship, governance, semantic consistency, and accountability of the enterprise's official shared master data assets . Its maturity model moves through five levels, from initial awareness that master data problems are hampering the business to treating master data as a strategic asset . Gartner's July 2025 research on ownership goes further: attempts to establish a single owner of master data routinely fail, and control must instead be exercised across three areas of the shared asset . The hiring consequence is that an MDM seat is rarely a pure architecture role. Survivorship rules, match and merge thresholds, golden record definitions, and domain steward committees are the daily material. Candidates who have implemented MDM on party, product, or location domains can name which match rules they tuned against false merges and which business unit resisted the golden record and why. That specificity is the craft; the platform badge is not.
Enterprise data catalogs evolved into metadata orchestration platforms
The 2025 Gartner Magic Quadrant for Metadata Management Solutions describes the market moving from augmented data catalogs toward metadata orchestration platforms that operate across the whole data environment . The difference is operational. A catalog is an inventory people browse; an orchestration platform collects, analyzes, and activates metadata across tools, feeding automation and impact analysis. The hiring market has not caught up with that shift. Plenty of candidates have administered a catalog product: loaded scanners, curated glossaries, ran certifications. Far fewer have wired metadata into pipelines, quality gates, and agent context so that definitions travel with the data instead of living in a portal nobody opens. The question that separates the populations is simple: when a field definition changed, which downstream systems learned about it automatically, and which ones the team had to go tell by hand. Orchestration-platform experience also shows in integration work: OpenLineage feeds, BI tool metadata connectors, and dbt or transformation metadata flowing into the catalog rather than sitting beside it.
Metadata management fails where stewardship carries no authority
DAMA's framework puts accountability in business hands. The DMBOK revision defines a data owner as a business person who is accountable for decisions about data within their domain . Catalog and glossary projects that skip this step produce shelves of definitions nobody is bound by. Snowflake's object tagging guidance reflects the same governance question at the platform level: a centralized tag administrator role versus decentralized tagging by individual teams . Both patterns work only while someone enforces naming and protects tags from drift. Practitioners who have run metadata management at scale can describe their stewardship model, which domains they onboarded in which order, and what happened when a steward left mid-program. Candidates who have merely used a catalog product describe features. The first group is far smaller than the second.
Data privacy obligations land inside masking and row-level policies
Snowflake's tag-based masking pattern is the practical shape of data privacy work. A masking policy attaches to a tag, and every column carrying that tag is automatically protected by data type, with new data staying protected until a data protection officer decides otherwise . One mechanism encodes default protection, minimisation, and a recorded decision point. Practitioners who have run privacy programs know the surrounding apparatus: sensitive data classification jobs, tag taxonomy design, PII detection, access reviews, and the quarterly evidence pack for auditors. The specialist who only knows a masking function cannot run the program. The one who can explain why a tag on the schema protects the table a team created last night understands the mechanism's point, and that understanding is what privacy compliance work actually consists of.
Data lineage evidence settles who actually governed the estate
Governance CVs converge on the same vocabulary: catalog, lineage, glossary, MDM, privacy. Verification separates practice from vocabulary with ownership questions. Which lineage graph did the candidate build, over how many systems, and what impact analysis ran before the last breaking change? Which catalog population did they own, and who reviewed new assets? For MDM, which match rules and survivorship decisions, and which business unit objected to the golden record? For privacy, which lawful basis was mapped to which pipeline, and what happened when a retention rule fired? A candidate who governed for real can narrate an escalation: a steward contested a definition, a masking policy broke a finance report, a subject-access request exposed a lineage gap. The cost of a miss is direct: audit findings that block launches, personal data processed without a documented lawful basis , and a governance office that exists on the org chart but not in the platform .
References
- Magic Quadrant for Metadata Management Solutions — Gartner. (accessed 2026-09-28)
- Master Data Management: Definition, Process, Framework and Template — Gartner. (accessed 2026-09-28)
- A guide to the data protection principles — Information Commissioner's Office (ICO). (accessed 2026-09-28)
- Data Lineage — Snowflake Documentation. (accessed 2026-09-28)
- Who Owns Master Data? — Gartner. (accessed 2026-09-28)
- DAMA-DMBOK 2.0 Revision — DAMA International. (accessed 2026-09-28)
- Tag-based masking policies — Snowflake Documentation. (accessed 2026-09-28)