In this post we are going to cover Unified Catalog, governance domains and data products in Microsoft Purview: how Purview turns a long list of scanned tables into something the business can actually use. These are 7 real Purview interview questions, asked the way they come up on a real project, and the last one is a scenario where the CFO asks you to make the monthly revenue report trustworthy.
Unified Catalog is the business side of Purview. The Data Map knows the tables and columns; Unified Catalog adds who owns the data, what the words mean, and which data is ready to use. Interviewers ask about it because it is where governance stops being a technical inventory and starts being something leadership can see and pay for.
Part 4 of the Real Time Microsoft Purview Interview Questions and Answers series on TechBrothersIT.
- The short answer
- Question 1: The business says the catalog is a list of ten thousand tables nobody understands. What is Unified Catalog, and how is it different from the Data Map?
- Question 2: You are setting up governance for Sales, Finance and HR. What is a governance domain, and how do you structure them?
- Question 3: The analysts want one place to find certified sales data. What is a data product, and what does it take to publish one?
- Question 4: Finance and Sales define revenue differently. How do glossary terms help, and where do they live now?
- Question 5: Leadership wants to track whether governance is working. What are OKRs and critical data elements in Unified Catalog?
- Question 6: Your company still has hundreds of terms in the classic business glossary. What do you do with them?
- Scenario 7: The CFO asks you to make the monthly revenue report trustworthy using Purview. What do you set up, in order?
The short answer
The Data Map is the technical inventory; Unified Catalog is the business layer on top: governance domains, data products, glossary terms, OKRs and critical data elements. A data product needs a published domain, an asset and an access policy to publish.
Keep that in mind. Every question below is one piece of it.
Question 1: The business says the catalog is a list of ten thousand tables nobody understands. What is Unified Catalog, and how is it different from the Data Map?
The Data Map is the technical layer. Unified Catalog is the business layer on top of it.
Let's look at the layers from the bottom up:
- Data Map. Scans fill it with assets, schemas, classifications and lineage. That's where the ten thousand tables come from.
- Unified Catalog. It runs on the Data Map, and adds governance domains, data products, glossary terms, OKRs and critical data elements, plus data quality and health management.
- Business users. They browse data products, not raw tables.
So the fix for ten thousand tables is curation. Business users browse a few dozen data products, each with an owner, a description and a use case, instead of raw tables.
Two facts worth adding. First, Unified Catalog needs the enterprise version with pay-as-you-go billing, or you stay on the classic Data Catalog. Then, billing counts governed assets: the ones you link to data products, critical data elements, glossary terms or quality rules.
Assets that are only scanned aren't billed. You pay for what you actually govern.
Question 2: You are setting up governance for Sales, Finance and HR. What is a governance domain, and how do you structure them?
A governance domain is a business boundary. It holds the data products, glossary terms, OKRs and critical data elements for one area, and its owners set the rules for access and use.
Let's set it up. For Sales, Finance and HR, I create one domain each, of type functional unit. There are five types: functional unit, line of business, data domain, regulatory and project. But they're only labels. They don't behave differently.
Then nesting. You can nest domains, for example Finance, with subdomains for Accounts Payable and Accounts Receivable. The limit is 200 domains and 5 levels deep, but keep it shallow.
Now two things people miss. A domain must be published before you can publish anything inside it. And governance domains are not Data Map domains:
| Governance domain | Data Map domain | |
|---|---|---|
| PURPOSE | business boundary | sources, scans, permissions |
| HOLDS | data products, terms, OKRs, CDEs | collections, sources, credentials |
| LIMIT | 200 domains, 5 levels | default + 4 custom |
Data Map domains organize sources, scans and permissions, with a default domain plus up to four custom ones. Governance domains organize the business side.
If a data product won't publish, check the domain before anything else.
Question 3: The analysts want one place to find certified sales data. What is a data product, and what does it take to publish one?
A data product is a group of data assets, like tables, files and Power BI reports, with a defined use case, that you share with other people. It belongs to one governance domain, but people can discover it across domains.
Let's create one. You give it a name, a description, a type, like dataset or semantic model, an audience, owners and a use case. For the certified part, there's a mark as endorsed option.
To publish it, three things must be true:
- The domain is published.
- At least one asset is added.
- An access policy is configured.
Then the lifecycle. Products go from draft to published, and later expired, and you can only edit them in draft.
The data product owner can't add any tables. That role isn't enough on its own. She also needs Data Map read permission, Data reader on the collection, to find the assets.
Question 4: Finance and Sales define revenue differently. How do glossary terms help, and where do they live now?
A glossary term gives a business word one agreed definition, with an owner. Owners are required. Terms also carry acronyms, synonyms, related terms, experts and resources.
Terms now live inside governance domains in Unified Catalog, not in one flat classic glossary. So let's split revenue in two:
- Finance owns Net Revenue: revenue after returns and discounts.
- Sales owns Booked Revenue: the value of signed orders.
Two clear terms, instead of one word that means two things.
Then you link them. Related terms and synonyms can cross domains, so each term points to the other. Parent and child terms stay in the same domain. Terms go from draft to published, and everyone can read published terms.
You can also link terms to data products, assets and columns. That part is in preview, and needs asset curation turned on. And a term can carry an access policy, which data products that use the term inherit.
Question 5: Leadership wants to track whether governance is working. What are OKRs and critical data elements in Unified Catalog?
OKRs are objectives and key results: trackable business goals tied to governance domains and data products. An objective has a target date, and each key result has a goal, a progress value and a status: on track, behind or at risk.
Let's write one. The objective: trusted finance reporting by year end. The key results:
- Every finance data product has an owner.
- Quality scores above 90 percent.
Data stewards create them, and a health control called business OKRs alignment checks that they're linked to data products.
A critical data element (CDE) is a logical grouping of the important columns across tables. CustID in one system and CID in another both map to one element, Customer ID. It has a data type, can link to glossary terms, and its data products are found automatically from the mapped columns.
Critical data elements are still in preview. An access policy on one applies to every associated data product.
The mapped assets become governed assets, billed per day. So map the columns that matter, not everything.
Question 6: Your company still has hundreds of terms in the classic business glossary. What do you do with them?
The classic glossary still works. It's under catalog management, classic types, glossaries, and it's tied to Data Map collections, not governance domains. But newer features, like linking terms to data products, need Unified Catalog terms.
If the migration tool is available in your tenant, it's under classic types, enable asset curation, and it's in preview. You pick one of three options:
| Option | What happens |
|---|---|
| Sample migration | up to 5 glossaries |
| Full migration | all glossaries and their terms |
| Switch directly | no migration; classic terms deactivated |
Then here is what happens, in order:
- Migrated terms land as drafts in a system staging domain, which can take up to 48 hours.
- Data stewards move each term to its owning domain and publish it.
- Links to assets and columns come along, but workflows don't.
Before you start, clean up: remove duplicates and unused terms, and agree which domain owns each term. To start the migration, you need governance domain creator and data curator.
There is no undo, and afterwards classic terms are read-only. That's why the cleanup comes first.
Scenario 7: The CFO asks you to make the monthly revenue report trustworthy using Purview. What do you set up, in order?
Let's build it in six steps.
- Step 1: scan the sources into the Data Map: the ERP database, the lakehouse and the Power BI workspace, so the assets and their lineage are there.
- Step 2: create and publish a Finance governance domain, with an owner and data stewards.
- Step 3: publish the Net Revenue glossary term, so everyone uses one definition, and a critical data element for the revenue amount columns.
- Step 4: build a Monthly Revenue data product with the revenue tables and the Power BI semantic model. Add an access policy, mark it endorsed, and publish it.
- Step 5: add data quality on the assets in that data product. Add rules like no blank amounts and unique invoice numbers, on a schedule with alerts. The score rolls up to the data product, so the CFO sees one number.
- Step 6: track it with an OKR, and show lineage from the report back to the source.
Then be clear about cost: those assets become governed assets, and quality scans use data governance processing units.
Quality scores roll up to the data product, so build the product in step four before the quality rules in step five.
Key points to remember
- Data Map = technical; Unified Catalog = business layer
- Governance domains: publish first; not Data Map domains
- Data product: published domain, an asset, an access policy
- Glossary terms live in domains; linking is preview
- OKRs track goals; CDEs (preview) group key columns
Part 5: Lineage Across ADF, Fabric and Databricks
Part of the Real Time Microsoft Purview Interview Questions and Answers series on TechBrothersIT. Product names, features and their status (GA or preview) are current as of October 2026. Microsoft changes Purview every month, so check the Microsoft Learn page for anything you quote in an interview.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.