Microsoft Purview Data Governance

Microsoft Purview data catalog

Purview's Data Map scans sources across clouds, and Unified Catalog adds data products, critical data elements and data quality.

Compare: data catalog tools

• Data Catalog: Microsoft Purview Unified Catalog provides AI-enabled data discovery, governance domains, data products, glossary, lineage, and data quality — a full SaaS catalog experience (docs).

• Automated Metadata Harvesting: Data Map scans multicloud, on-premises, and SaaS sources to automatically capture technical, business, and semantic metadata (docs).

• Search & Discovery: Search by governance domain, data product, keyword, or AI copilot — full catalog search with role-based faceting (docs).

• AI-Powered Search: Security Copilot integration enables natural language catalog queries — AI-powered copilot finds data assets via conversational search (docs).

• Custom Fields & Templates: Custom metadata (Preview) adds attribute groups for business concepts and data assets (docs).

• Catalog Usage Analytics: Health controls and health actions provide governance program health scores and estate-wide data governance coverage dashboards (docs).

• Business Glossary: Unified Catalog glossary terms are active objects: terms applied to data products pass their attached policies down to the underlying data assets (docs).

• Metadata Propagation: Glossary term policies trickle down from governance domains to data products to individual assets — automatic policy propagation confirmed (docs).

• Usage & Popularity Statistics: No per-asset usage or popularity view is documented in Unified Catalog (docs).

• Asset Certification: No certified/endorsed status for assets found in the Unified Catalog docs.

Microsoft Purview data lineage

Purview shows table, column and end-to-end lineage across Azure and connected sources.

Compare: data lineage tools

• Data Lineage: Lineage tracking between data products and assets — Data Map captures lineage from Azure Data Factory, Synapse, Power BI, and 200+ connectors (docs).

• Table-Level Lineage: Dataset-to-dataset lineage mapping across Azure, multicloud, and on-premises sources via Data Map scanning (docs).

• Column-Level Lineage: Column or attribute level lineage identifies which source attributes create or derive target attributes, e.g. Azure Data Factory copies Table1/ColumnA -> Table2/ColumnA (docs).

• End-to-End Lineage: End-to-end lineage from source systems through Azure pipelines to Power BI dashboards and reports — confirmed for Microsoft ecosystem (docs).

• Impact Analysis: Impact analysis identifies downstream assets affected by upstream changes — supported in Data Map lineage visualization (docs).

• OpenLineage Support: The Data Map source table lists Airflow lineage but doesn't mention OpenLineage (docs).

Microsoft Purview data stewardship

Compare: data stewardship tools

• Data Stewardship: Data owners and stewards manage curation, quality, access, and lifecycle within governance domains — federated stewardship model confirmed (docs).

• Critical Data Elements: Critical data elements are explicitly listed as a Unified Catalog feature — logical groupings of key business data with attached quality rules and access policies (docs).

• Review & Approval Workflows: Workflows automate approvals for data product access and for publishing data products and glossary terms, with pending-on-all or pending-on-any approvers and conditions (docs).

• Governance Scorecards: Health controls measure governance practices against standards and give the data estate a maturity score — equivalent to governance scorecards (docs).

• Metadata Automation: Health actions, access policies, and governance workflows automate data curation, access provisioning, and quality improvement at scale (docs).

• Governance Campaigns: Bulk glossary edit/move exists (Preview, Apr 2026), but there are no asset-wide campaigns (docs).

Microsoft Purview data classification

Compare: data classification tools

• Sensitive Data Discovery: Automated classifiers and sensitivity labels scan data assets for PII, PHI, PCI, and confidential data — trainable and built-in classifiers available (docs).

• Data Policy: Unified Catalog access policies manage self-service data product access with right-use requirements and security controls (docs).

• Native Policy Enforcement: Data owner/DevOps policies are enforced in Azure Storage and SQL Server on Azure Arc (Policy column = Yes), but Azure SQL Database support was dropped in Jul 2026. This is a judgment call, since the targets are Microsoft products (narrow) (docs).

• Data Masking: Sensitivity labels apply encryption and masking to classified PII/PHI fields; DLP prevents unauthorized sharing or export of sensitive data (docs).

Microsoft Purview AI features

Purview's AI features in Unified Catalog are previews: natural-language data product search, asset suggestions and data quality rule suggestions. There's no first-party MCP server.

• AI Features: Security Copilot for natural language catalog queries, AI-powered recommendations, and automated classification all confirmed (docs).

• AI Chat Assistant: Security Copilot enables conversational data discovery and governance queries across the Purview catalog (docs).

• AI Documentation Generation: The application card lists no description generation (docs).

• Agentic Metadata Enrichment: All AI suggestions need human review. Same source.

Microsoft Purview data compliance

Purview covers data subject rights, retention policies, activity monitoring and compliance reports, the broadest compliance coverage in the matrix.

• Data Compliance: Compliance Manager, Purview Audit, Records Management, and eDiscovery provide full regulatory compliance coverage (docs).

• Audit Trail: Microsoft Purview Audit is a dedicated product — immutable, time-stamped audit trail for all user and admin actions across Microsoft 365 and data assets (docs).

• Compliance Reporting: Compliance Manager generates score-based compliance reports and evidence packs against 300+ regulatory frameworks (docs).

• Regulatory Framework Mapping: Compliance Manager includes 300+ regulation templates — GDPR, HIPAA, SOX, CCPA, NIST CSF, FedRAMP — with gap analysis and evidence collection (docs).

• Data Activity Monitoring: Activity Explorer monitors user interactions with sensitive data across Microsoft 365 — showing who accessed, modified, or shared governed data (docs).

• Data Retention Policies: Microsoft Purview Data Lifecycle Management is a dedicated product — enforces retention and deletion policies across Microsoft 365 and Azure data assets (docs).

• Data Subject Rights: Priva Subject Rights Requests (M365 GA; beyond-M365 incl. Snowflake in Preview, licensed per request) (docs).

• ROPA: Priva Privacy Assessments may cover processing records; not verified (docs).

Microsoft Purview collaboration

• Collaboration: OKRs, governance domain collaboration, self-service requests, and Microsoft Teams notification integration confirmed (docs).

• Alert Notifications: Microsoft Teams and email notifications for governance events and access request updates confirmed via Microsoft 365 integration (docs).

• Ticketing Integrations: Workflows are built in. No Jira or ServiceNow connector is documented (docs).

• Discussions & Annotations: No comments or threads on assets are documented (docs).

Microsoft Purview data marketplace

• Self-Service Access Requests: Self-service access requests are a confirmed Unified Catalog feature — consumers request access to data products and requests route to owners (docs).

• Data Products: Data products group related assets with use cases for discovery — explicitly listed as a core Unified Catalog feature (docs).

• Data Mesh Support: Governance domains plus a stated "federated governance approach" (docs).

• Subscription Management: Users can subscribe to data products to receive updates — mentioned in Unified Catalog overview as a key discovery capability (docs).

• Data Marketplace: Data products exist in Unified Catalog with access requests, but a dedicated consumer-facing marketplace storefront was not confirmed (docs).

Microsoft Purview data quality

• Data Quality: Built-in data quality rules, health scores, and health actions enable data owners to monitor and improve quality across the data estate (docs).

• SQL Table Tests: Data quality rules can be set at the data product and asset level — including custom quality conditions and rules (docs).

• Data Health Scores: Data quality scores at asset, data product, and governance domain levels — providing full estate-wide quality health visibility (docs).

• Data Profiling: AI-assisted profiling with distribution, min, max, stddev, uniqueness and completeness (docs).

• Anomaly Detection: DQ is rule- and threshold-based, with no ML anomaly detection (docs).

• Data Contracts: Not documented. Data products have access policies but no contracts (docs).

Microsoft Purview AI governance

• AI Governance: DSPM for AI discovers and governs the AI apps and agents in use across the tenant — Copilot experiences, Microsoft Foundry, Entra-registered apps and third-party LLM tools — with policy, audit and compliance controls applied to AI interactions (docs).

• AI Asset Inventory: DSPM for AI inventories the AI apps and agents connected to the organization and categorizes them (Copilot experiences, Enterprise AI apps, other AI apps); it is an inventory of AI in use rather than a registry of models the organization trains (docs).

• AI Regulatory Compliance: Compliance Manager ships regulatory assessment templates for the EU AI Act, NIST AI RMF, ISO 42001 and ISO 23894, translating each into controls with evidence workflows (docs).

• AI Agent Access Governance: Purview controls ensure data 'is never returned to the user or used by a large language model (LLM) if the user doesn't have access to that data'. Sensitivity labels apply at retrieval — encrypted content requires the EXTRACT usage right before an AI app returns it — and endpoint DLP blocks sensitive data from being shared with generative AI apps. Covers Copilot experiences and agents, Entra-registered AI apps, ChatGPT Enterprise, and Anthropic Claude (Enterprise) (docs).

• AI Lineage: Purview governs AI usage — prompts, responses and the data accessed during interactions — but documents no lineage from training datasets to a model (docs).

• AI Model Health Score: No composite per-model trust score; DSPM for AI reports risk at the data and application level rather than scoring individual models (docs).

Microsoft Purview workflows

• Workflows: REST APIs, Python SDK (Azure SDK), and Power Automate integrations all confirmed (docs).

• REST / GraphQL API: Microsoft Purview provides extensive REST APIs via the Azure SDK — Data Map, Catalog, and Scanning APIs all documented (docs).

• Python SDK: Azure SDK for Python includes azure-purview-datamap and azure-purview-catalog packages for programmatic catalog access (docs).

• Webhooks: Atlas Kafka/Event Hubs notifications existed in classic Data Map; not re-verified for Unified Catalog.

• MCP Server: There's no first-party MCP server in the docs or the 2026 what's new (community packages only) (docs).

• dbt Integration: dbt isn't in the Data Map supported sources (docs).

• Browser Extension: No browser extension confirmed for Microsoft Purview (docs).

Microsoft Purview data context layer

• Knowledge Graph: The Data Map is Apache Atlas-based, but no property-graph feature is documented for users.

• Semantic Models: Not documented (docs).