Microsoft Is Retiring Purview. Don’t Replace One Lock-In With Another.

On Nov. 12, 2026, Microsoft announced it will retire Purview Data Governance, with support ending Nov. 12, 2027. Microsoft is folding the Unified Catalog, Data Map, and governance workflows into OneLake Catalog, built directly into Microsoft Fabric. For organizations relying on Purview, this change creates a deadline to migrate critical governance capabilities or risk losing support for the tools and processes they depend on.

Gartner® published a first take on the announcement the same day, and its read is direct: treat OneLake Catalog as a Fabric discovery surface, not an enterprise governance platform. 

For organizations already standardized on Fabric, that distinction may not matter much. For the much larger group running Snowflake, Databricks, AWS, Google Cloud, Oracle, SAP, and on-premises systems alongside Microsoft, it changes the calculation entirely.

Should your data governance strategy depend on the roadmap of the one vendor hosting your data warehouse?

You have 12 months to answer that deliberately, not by default.

What’s Actually Changing

Microsoft’s retirement affects:

• Unified Catalog, including business glossaries, data products, and data quality capabilities
• Data Map and its metadata management
• Governance workflows
• Classic Azure Purview Data Governance

Microsoft Purview Data Security and Compliance remain supported, including sensitivity labels and data loss prevention.

The bigger shift is architectural. Governance is moving into Fabric’s storage and compute environment, simplifying metadata management for Fabric users. However, organizations using Snowflake, Databricks, AWS, or on-premises systems may need additional integrations, which could limit governance coverage.

Cost is another consideration. While OneLake Catalog may be included with Fabric, governance workloads can consume shared compute capacity, potentially increasing costs.

For multicloud organizations, the tradeoff is clear: Fabric-native convenience versus the flexibility of independent governance.

The Real Question: How Much of Your Data Actually Lives in Microsoft?

Most large enterprises do not run on one platform. Different business units adopt different tools. Acquisitions bring in new ones. Regulations dictate where some data has to sit. And AI agents now need trusted context pulled from all of it at once, not just the slice that happens to live in Fabric.

This is exactly the gap: enterprises with complex, multicloud footprints that need deep cross-platform lineage, mature business glossaries, or active stewardship workflows will still need a third-party governance layer, because OneLake Catalog’s near-term reality is Fabric-centric discovery, not enterprise-wide governance. The shift also trades platform neutrality for ecosystem lock-in: multicloud enterprises are left choosing between centralizing more workloads in Fabric or keeping an independent governance layer to avoid that dependency.

Your governance architecture should reflect where your data lives today and where the business is taking it next, not where one vendor would prefer it to live.

Five Reasons to Choose Independent Governance Over Another Platform Catalog

1. Govern Data Wherever It Actually Lives

A finance team on Snowflake, an engineering team on Databricks, and a business team in Power BI should work from the same business definitions and the same trusted metadata, not three versions stitched together after the fact. An independent governance platform gives every team the same vocabulary, regardless of which engine holds the data.

2. Stop Betting Governance on One Vendor’s Roadmap

Purview customers are now replanning their catalogs, glossaries, workflows, and data quality programs for the second time in this platform’s life. Gartner’s advice to these customers is not to let sunk costs stall progress: audit what you have, decide what must be preserved, and evaluate independent alternatives early if OneLake Catalog cannot meet your requirements now, not after the support window narrows. An independent governance layer means the next roadmap change is someone else’s transition, not yours.

3. Preserve the Business Context You Already Built

A catalog is not a list of tables. It is years of defined business terms, assigned ownership, documented lineage, and certified data products, exactly the manually curated context that automated scanning cannot simply regenerate. Before committing to a replacement, get specific about how that context (glossaries, ownership, lineage, data products, and policies) will be exported, mapped, validated, and kept current in the new environment.

4. Connect Data Quality to Governance, Not Just to Fabric

Purview customers using its data quality capabilities should not assume existing rules, scores, and monitors transfer automatically. Decide now which controls must survive the transition, which should improve, and how monitoring continues while you migrate, especially in hybrid environments where data quality problems usually originate upstream, outside Fabric entirely.

5. Build the AI Foundation on Context, Not on One Platform

AI agents fail when they get access to data without the context to use it correctly: what a term means, which dataset is approved, who owns it, whether it can be trusted. Actian customers report eliminating up to 60% of AI failures once production data carries real context, not just a schema.

That context rarely lives on one platform. An AI agent answering a revenue question might need financial data from one system, customer data from another, and definitions a separate team owns. An independent metadata and governance layer makes that context available consistently across every AI tool and data environment you run, not just the one Microsoft hosts.

Five Actions Purview Customers Should Take Now

1. Audit your dependencies. Inventory catalogs, glossaries, lineage, data quality rules, workflows, and classifications. Separate what can be rediscovered automatically from what represents manually curated business knowledge.
2. Decouple security and compliance now. Those capabilities are unaffected and fully supported; isolate them from the governance assets that are retiring so nothing gets caught in the transition by accident.
3. Map your real data footprint. How much genuinely sits in Fabric versus Snowflake, Databricks, AWS, Google Cloud, Oracle, SAP, and on-premises systems, and where is that footprint headed next?
4. Evaluate OneLake Catalog against independent alternatives on current capability, not roadmap promises. Benchmark multicloud scanning, cross-platform lineage, business glossaries, and policy enforcement as they work today.
5. Model the full cost, not just the license. There is a real risk here: governance scanning inside Fabric draws on shared Fabric Capacity Units, which can compete with the Power BI refreshes and ETL pipelines already running on that capacity. Price that in before you commit.

Where Actian Fits

The right replacement for Purview should give organizations the flexibility to govern data across environments, preserve business context, and support AI requirements without forcing them to consolidate technology investments. 

The Actian Data Intelligence Platform isn’t another solution that locks you in. It is a composable data management platform, positioned as the context platform for trusted AI, built on three commitments: 

1. Discover: one federated knowledge graph spanning every cloud, legacy system, and third-party catalog, and can act as a catalog of catalogs, so context is enterprise-wide rather than per-tool. 
2. Trust: governance expressed as executable data contracts on open standards (ODCS and ODPS) and enforced at runtime, with observability measuring quality continuously rather than at design time. 
3. Activate: governed context served to analysts, applications, and AI agents through open protocols including MCP and A2A. 

Each capability ships modular, so you add pieces as you need them instead of committing to a full suite on day one. That’s the same principle this blog opened with: governance should adapt to your architecture, not the other way around.

In practice, that looks like: 

• Modular by design: start with one capability, catalog or observability, and layer in governance, data quality, or AI readiness as requirements grow. No need to buy a full platform to meet a Purview deadline.
• A flexible metamodel: live connectors harvest metadata across distributed domains without centralizing the data itself.  BPCE used that same flexibility to connect 30 individual platforms natively, replacing the third-party middleware a more rigid catalog required.
• Bidirectional metadata exchange, not a rip-and-replace: a federated semantic overlay syncs governance, lineage, and policy across Snowflake, Databricks, Fabric, on-premises, or SaaS, without rebuilding your stack around Actian first.
• Governance embedded in the data product: data products operationalize data contracts, lineage, and trust signals at runtime, so governance ships inside the product from day one.
• Governed context for AI: the same modular metadata and business definitions become available to AI agents through a knowledge graph built for fast AI context. 

With Actian, Generali Group governs more than 11,000 datasets and 300,000 fields across 50-plus countries on the same platform. France Travail cataloged 20,000 datasets and 30,000 glossary definitions, and now sees 1,000 weekly active users with 77% satisfaction. None of that required standardizing on a single cloud.

Deployment doesn’t require a nine-to-18-month services engagement either: Actian customers typically go live in about six weeks using 75-plus automated connectors.

Organizations can keep using Microsoft technologies wherever they make sense, while running a governance strategy built to work across the rest of the stack too.

12 Months to Migrate. A Chance to Choose Differently. 

Microsoft’s decision forces a technology decision inside the next year, whether you make it deliberately or by default. For organizations fully inside Fabric, OneLake Catalog may be the simplest path. For the much larger group running data across multiple clouds, business units, and platforms, this is the moment to choose governance that does not quietly re-create the same dependency under a different name.

Before you commit to another platform-specific catalog, ask the question that actually matters: does this architecture give you the freedom to choose what comes next?

Talk to a Data Intelligence Expert to map your Purview migration, or explore the Actian Data Intelligence Platform to see how it governs data across every environment you run. 

About Author

About Thaise Skogstad

Thaise Skogstad is a Product Marketing Director at Actian, leading product marketing for the Data Intelligence Platform. She brings experience in product marketing, data management, analytics, and AI from roles at Great Expectations, data.world, Anaconda, and 3M. She focuses on helping enterprises leverage data and analytics to drive business impact.