Data Certification Advisor
Description
This workflow orchestrates a process to validate and certify data governance across multiple tables. It evaluates every table against five governance signals — ownership, documentation, quality, lineage, and access control — and automatically recommends a certification type, so users know which data they can trust, which needs attention, and which is a compliance risk.
Details
Today, nobody can quickly tell which data in the catalog is safe to use and which isn't. Stewards either have to manually check ownership, documentation, quality, and access controls one table at a time — or nobody checks at all before a report or pipeline gets built on top of it. This recipe closes that gap: it looks at every table and gives it one clear, defensible certification with a plain-English reason attached.
Business value
- Creates a single, reusable trust verdict for every table in the workspace — no more guessing or manual spot-checks.
- Flags hard governance violations directly — unmasked sensitive/PII columns, and critical data with no real accountable owner.
- Distinguishes a genuine owner from a generic admin account, so ownership reflects accountability rather than a placeholder.
- Rolls raw findings up into a summary and chart showing exactly which connections carry the most risk.
- Gives every table a plain-English reason for its status, making each finding immediately actionable rather than a black-box score.
Who benefits
- Data Stewards
- Analysts and BI Teams
- Governance and Compliance Stakeholders
- Leadership
What you receive
How the recipe works (guided flow)
Step 1 — Build the Master Table List
Scan every table and consolidate schema, source connection, owner, documentation, data quality, lineage connectivity, and sensitive-data signals into one complete inventory.
Step 2 — Assess the Five Trust Signals
Evaluate ownership, documentation, data quality, lineage connectivity, and sensitive-data protection, while verifying that ownership belongs to a genuine non-admin user and assigning each signal a Strong or Weak status.
Step 3 — Apply the Governance Rules
Check for critical data without a real accountable owner, exposed sensitive data without masking, or insufficient information for assessment, then calculate a composite trust score out of 100 for tables that clear the governance checks.
Step 4 — Certify, Explain, and Report
Assign every table one clear certification — Certify, Caution, Violation, or None — with a plain-English reason, then produce the business-facing report, connection-level summary, and risk distribution chart.
Sample insights (illustrative)
| Insight Category | What the recipe discovered | Business Impact |
|---|---|---|
| Ownership Quality | A meaningful share of "owned" tables were actually attributed only to admin service accounts rather than real people. | Filtering these out provides a more honest picture of true accountability across the catalog. |
| Risk Concentration by Connection | Violations and stale-data risk cluster heavily in a small number of source connections rather than spreading evenly. | Teams can prioritize governance outreach to the connections driving most of the risk instead of conducting a catalog-wide sweep. |
| Certify Remains Rare Early On | Very few tables clear all five trust signals at once in an early assessment. | This quantifies the governance work ahead, table by table, rather than representing a failure of the certification process. |
Make sure the following are available in your workspace:
- Core catalog tables are crawled: oetable, oeschema, connectioninfo, oecolumn, lineagecolumn
- Business documentation is maintained in wiki, separate from technical table/column descriptions
- User and role metadata is available: user, user_role, roles, so ownership can be verified against real, non-admin accounts
- Governance fields, including owner, steward, custodian, governance roles, and critical-data/masking flags, are populated during crawling