By Matt O'Donnell, Co-founder, Kenseme and Macula Systems
On 12 November, Microsoft announced that it is retiring Purview Data Governance and making the OneLake catalog in Microsoft Fabric the foundation for governing data across the estate.
If you run a governance program on Purview, I'd guess your first thought wasn't about the technology. It was about the work: the domains, the glossary, the owners and data products your team spent the last year or two getting right, and the ways you've shaped Purview to fit how your program runs. What happens to all of that? What's the plan, and where is governance in Fabric headed? And what do you tell your stakeholders and leadership?
The good news is that a year is enough time to do this well, though it's wise to get started soon. I've spent the last few years working with Purview customers, so here's how we're thinking about it at Macula.
I'll keep this section to what Microsoft has said, and update it as they publish more.
Change like this is never convenient, but the direction makes sense to us. Governance sits right next to the data in Fabric, where people already find and use it. There's one less catalog to run, and the cost is part of the capacity you already pay for. For most teams that should mean a simpler experience, with governance built into the platform people already work in. It's also why we've built Kenseme to publish into Fabric.
Governance is mostly people work. Somebody had to agree what "active customer" means, decide who owns the finance domain, and chase the descriptions for a few hundred tables. That effort is the most valuable part of a Purview estate, and a platform change doesn't erase it. It can be read, kept and carried across, so you shouldn't have to start over.
A year sounds like a lot until you count the steps: find out what you have, decide what's still worth keeping, clean it up, plan who moves what, move it, and check nothing fell through. Large estates with several business units need most of that year.
With our current customers, we're starting now and planning it like this:
There's also a lot worth doing along the way that counts wherever you land: keeping your current workflows and governance operations running, maintaining metadata, naming owners, tying data to the business terms people actually use, and general cleanup once you've decided what stays and what goes.
Purview deserves credit, and this change doesn't take that away. It gave our Microsoft customers one catalog for their governed estate. Its Data Map scans widely and at low cost. Data domains and data products gave business stakeholders the means to build, govern and publish trusted data. Many organizations bought Purview first for protection and compliance, which stays in Purview. The work your team has done in Purview Data Governance is real, and it's worth keeping.
With the announcement, the OneLake catalog becomes the home for much of what you've come to rely on in Purview Data Governance. It isn't a one-for-one replacement, though, and the move is more than a technical exercise. Some things will work differently, and that can reach into how your program runs day to day.
Before the technical work starts, it's important to take a careful look at how you manage your Purview estate today and what that should look like in the OneLake catalog. That keeps the move in line with your governance practices as they actually run. It means deciding first what's still used, what's governed and what's worth carrying, and who in your organization owns each part of it.
This is the question I hear most from leadership teams right now, and a count of terms written and assets scanned doesn't really answer it. We find it easier to break it into three simpler questions: Is it written down? Can people trust it? Can an AI agent use it safely?
Those turn into a ladder you can measure, as a share of your published data products:
In the estates we've worked on, a lot more is described than trust-ready, and a lot more is trust-ready than AI-ready. That gap is where the work is, and it's the work that gives you the controls and the governance model AI needs. Looked at domain by domain, the ladder gives leadership something to track through the move, and gives stewards a short list of where to start.
Kenseme is a Macula Systems company. Macula has worked with Purview customers since before Unified Catalog. In 2024 we put Macula Purview Automate (MPA) on the Azure Marketplace, for free, to help teams adopt and migrate to Purview. It's been downloaded more than 1,000 times and used by many of our customers. We've run Collibra to Purview migrations, helped teams set up how their governance program runs day to day, and worked alongside the Purview product group on migration scenarios and core product features.
What we learned in that work is that a move like this is a process first and a technology second. It goes well when you know what you have, decide what matters, agree who owns it, and then move it. That's how Macula runs these programs.
MPA changed a lot along the way. Customers kept bringing us new situations, and we kept improving it to handle them. Its Purview work is now part of Kenseme, built to help you through exactly this kind of change: assessing what you have and moving it with care.
We've put together a path for Purview customers through this change, and it starts with what you already have. You connect Kenseme to Purview, which takes a few minutes, and get a free snapshot of what's connected, what you've got and how much of it is governed. From there, Kenseme measures the readiness ladder across your estate and shows the gaps behind each step, then helps you plan the move and carry your domains, terms, data products and owners into the OneLake catalog.
Start the plan now, take the snapshot before the year is out, and you'll go into 2027 knowing what you have and what comes next. If you'd like a snapshot of your estate, or just want a first conversation about what the change means for your program, get in touch or write to hello@maculasys.com and we'll find a time. We're here to help.
Matt O'Donnell is a co-founder of Macula Systems and Kenseme. He has helped many organizations set up and run their data governance programs, and get their data ready for AI.
This post also appears on the Kenseme blog.