CertSafari
    Databricks Certified Data Analyst Associate· Lessons

    Domain 2 · Lesson 4/39

    Certified datasets in Unity Catalog: find, inspect, query and certify

    Use Unity Catalog to discover, query, and manage certified datasets.

    17 min read
    2.67% of exam
    8 sources
    Published 22 Oct 2026
    Docs as of 17 Oct 2026

    What you will be able to do

    • Explain what the certified and deprecated badges tell a data consumer and which tag produces them
    • Filter workspace search to certified or deprecated assets with the certificationStatus keyword and the Certification status filter
    • Choose between the Discover page and Catalog Explorer when looking for a trustworthy table
    • Inspect a certified table's sample data, details, comments and lineage before querying it
    • Certify or deprecate an asset in the UI or with SQL, and know the permissions this requires
    • Describe how tag-assignment automation certifies or deprecates many tables at once

    Key concept

    Certification status system tag — Certification in Unity Catalog is not a privilege, an owner, or a comment. It is the value of one Databricks-defined governed tag, system.certification_status, which can be set to certified or deprecated. Setting the tag shows the badge, and searching on the tag finds the asset.

    1.What the certified and deprecated badges mean

    A large catalog usually holds several tables that look alike. Picture orders, orders_v2, orders_backup and a personal scratch copy, all with plausible columns. Certification tells a consumer which one the organization stands behind. A certified asset meets internal standards for accuracy, completeness and trust. A deprecated asset is outdated or unreliable and should not be used in new work. You see both signals right next to the asset's name, so you can judge it before writing a single query.

    The two values of the certification status tag and what a consumer sees
    Tag valueIcon in the workspaceWhat it tells you
    certifiedCheck markThe asset met internal standards for accuracy, completeness, and trust
    deprecatedRestricted iconThe asset is outdated, no longer reliable, or should not be used in new workflows

    Both badges come from a single tag, system.certification_status, which Databricks defines. It is a system tag, a predefined type of governed tag. Its key and its two allowed values are fixed, and users cannot modify or delete them. In the tag picker, system tags carry a wrench icon. Because the tag is defined once for everyone, "certified" means the same thing in every catalog of the account. That is the point of using a system tag instead of letting each team invent its own gold or trusted label.

    The badge follows the asset into the tools where analysts work. The tag appears next to object names in the workspace and changes how data appears in notebooks and the SQL editor. It also feeds AI features: Genie One gives priority to assets the organization vouches for when it answers questions. Certification works on more than tables. You can apply it to catalogs, schemas, tables, views, volumes, functions, registered models, dashboards, Genie Agents, Databricks Apps and notebooks.

    Checkpoint 1 of 9· Check yourself

    A catalog holds orders, orders_v2, orders_backup and a scratch copy. The data team wants analysts to see a check-mark badge next to orders_v2 in every tool, and wants it to mean the same as "certified" in every other catalog. What should they do?

    Sources12

    2.Searching for certified data, and Discover versus Catalog Explorer

    Search filters on certification status with its own keyword, certificationStatus, so you don't have to open each table to check its badge. Pair it with type:table to get only certified tables. Use certificationStatus:deprecated on its own to list deprecated assets of every supported type, which is handy when you are cleaning up. If you prefer clicking to typing, the search filters include a Certification status menu with Certified and Deprecated options.

    Search query that returns only certified tablestext
    type:table certificationStatus:certified

    Close-sounding guesses fail. tag:certified is not the syntax, and "certified" is a value of the system tag, not a tag key you can search for under that name. Tag search also needs exact term matching. Three other limits matter in practice. Search on tags does not cover dashboards, Genie Agents or Databricks apps, so a certified dashboard will not appear under certificationStatus:certified even though it shows the check mark. A newly certified table can take a few minutes to show up in search results. And sorting a schema's tables by creation date tells you nothing about trust.

    Search is one entry point. The other question is which browser to start from. The Discover page (Discover in the sidebar) is a curated subset. Curators arrange assets into business domains and subdomains, and AI recommendations bring popular assets to the surface. Under any asset type, View all lists every asset of that type, with Certified assets and AI-suggested assets shown as groups. Catalog Explorer (Catalog in the sidebar) shows everything in your catalogs in the catalog hierarchy, including properties, permissions, sample data and lineage.

    Choosing where to look for a trustworthy table
    AspectDiscover pageCatalog Explorer
    Opened fromDiscover in the sidebarCatalog in the sidebar
    What it showsA curated subset of your dataEverything in your catalogs
    Organized byDomains and subdomains, or asset typeThe catalog hierarchy
    Use it whenYou want the assets your organization recommendsYou need the complete picture, or need to inspect and configure a specific object

    In practice, a business user who wants "the revenue table finance trusts" should start in Discover, inside the Finance domain. An analyst who already knows the table is somewhere in main.sales and needs its column types should go straight to Catalog Explorer.

    Checkpoint 2 of 9· Match them up

    Match each starting point to the job it fits best when looking for trustworthy data

    Tap a term, then the definition that fits it.

    Checkpoint 3 of 9· Check yourself

    Yesterday morning, Finance certified both its revenue table and its revenue dashboard. Both now show the check mark. Today, a search for certificationStatus:certified returns the table but not the dashboard. What explains this?

    Checkpoint 4 of 9· Exam question

    A data analyst at a retail company wants to search Catalog Explorer for only the tables that have been certified as trustworthy for analysis, without manually checking each table's status. Which search approach lets them filter the results to just certified assets?

    Sources1234

    3.Inspecting a certified table before you query it

    A check mark tells you someone vouched for the table. It does not tell you whether the table fits your question. Before you query, open it in Catalog Explorer and its object tabs and spend a minute on three tabs:

    • Sample Data shows real rows, so you can spot unexpected NULLs or odd encodings early. It needs an active compute resource.
    • Details shows table details, including table properties.
    • Comments on the table and its columns describe what the data means. You can read them in Catalog Explorer or with DESCRIBE.

    Lineage is the next check. Unity Catalog records lineage automatically for queries that run on Databricks and shows it in Catalog Explorer; Data lineage covers the graph in depth. For a certified dataset, trace it upstream before you trust it. A certified table built from a source that has since been deprecated deserves a second look. The Lineage tab lists related tables, and its details panel can be filtered by notebooks, jobs, pipelines and queries.

    Lineage also answers the reverse question that comes up when you manage a dataset: who reads this table? Before you deprecate or drop a table, the Lineage tab shows what consumes it. Other tabs can look like they answer that question, but they don't. The History tab (DESCRIBE HISTORY) lists changes made to the table, not who read it. The Permissions tab shows who *may* read it, and a SELECT grant does not mean any job actually uses it. Tags do not list consumers either.

    Once the table checks out, query it from a notebook or the SQL editor. The badge appears there as well. The SQL equivalent of the Sample Data tab is a simple bounded read:

    SQL that reads a sample of rows, like the Sample Data tabsql
    SELECT * FROM table_name LIMIT 1000;

    Checkpoint 5 of 9· Check yourself

    Before deprecating main.sales.orders_backup, a data steward needs to find out which jobs and queries still read it. Where should they look?

    Checkpoint 6 of 9· Exam question

    Before dropping a Delta table that several downstream dashboards may depend on, a data engineer wants to confirm exactly which dashboards and jobs consume it so the change doesn't break anything. What should they do in Unity Catalog?

    Sources5637

    4.Certifying or deprecating an asset in the UI or SQL

    Because certification is a tag, certifying a table is a tagging operation. Everything about Tags and governed tags applies: who may tag, which statements set a tag, and how to undo it. Two permission checks have to pass. First, you need the ASSIGN permission on the system.certification_status governed tag itself. Second, you must own the object or hold APPLY TAG on it, USE SCHEMA on its parent schema and USE CATALOG on its parent catalog.

    In the UI, open a supported object, click the kebab menu, choose Assign certification, pick Certified, Deprecated or None, and click Save. The status can be changed or removed at any time. To do the same in a script, set the tag directly:

    Certifying one table and deprecating another with SET TAGsql
    -- Apply certified tag key
    SET TAG ON TABLE main.sales.transactions `system.certification_status` = `certified`;
    
    -- Apply deprecated tag key
    SET TAG ON TABLE main.sales.old_summary `system.certification_status` = `deprecated`;

    SET TAG / UNSET TAG need Databricks Runtime 16.1 or above. The older form, ALTER <object> ... SET TAGS / UNSET TAGS, works on Databricks Runtime 13.3 and above. Either one sets the same tag. A nightly job can therefore certify a table with ALTER TABLE gold.sales.orders SET TAGS ('system.certification_status' = 'certified') and get the same check mark as the UI.

    Three statements look as if they should certify a table, and none of them does. GRANT gives privileges such as SELECT or MODIFY, and no CERTIFIED privilege exists. COMMENT ON TABLE ... IS 'certified' only writes free-text description. SET OWNER makes someone the owner, and owning a table only *permits* them to tag it; a table does not inherit a status from its owner.

    Certifying a catalog or schema labels that container and nothing else. Tags pass down the hierarchy only when Unity Catalog evaluates ABAC policies, so a certified catalog does not put a check mark on its tables. To mark every table in a schema, tag each table, or use the automation described next.

    Checkpoint 7 of 9· Fill the gap

    Which tag key completes this script so that main.sales.transactions shows the certified check mark and main.sales.old_summary shows the deprecated icon?

    -- Apply certified tag key
    SET TAG ON TABLE main.sales.transactions  ?  = `certified`;
    
    -- Apply deprecated tag key
    SET TAG ON TABLE main.sales.old_summary `system.certification_status` = `deprecated`;

    Checkpoint 8 of 9· Exam question

    A Unity Catalog admin is scripting the certification of newly onboarded Gold-layer tables as part of a nightly job, and wants to avoid the manual UI workflow. Which SQL statement correctly marks a table as certified?

    Sources128

    5.Certifying and deprecating many tables at once

    Tagging one object at a time does not scale to a catalog with thousands of tables. For many objects, Databricks recommends automating tag assignment (in beta). You define conditions, and the automation sets system.certification_status on every Unity Catalog table or volume that matches. Conditions can use query usage, asset age, owner, description, name or existing tags.

    The standard example is cleanup. An automation finds every table in the temp catalog with no read or write queries in the last 30 days, sets the tag to deprecated, and emails the asset owners. Nobody has to remember to retire stale scratch tables, and owners get warned before anyone removes anything. To set one up, describe your criteria to Genie, preview the matches with a dry run, and then enable the automation to run on a schedule.

    Checkpoint 9 of 9· Put it in order

    To deprecate every table in the temp catalog that had no read or write queries in the last 30 days, without tagging tables one by one, put these steps in order

    1. 1.Preview the matching tables with a dry run
    2. 2.The automation sets system.certification_status to deprecated on each matching table and emails the asset owners
    3. 3.Describe the criteria to Genie: tables in the temp catalog with no read or write queries in the last 30 days
    4. 4.Enable the automation so it runs on a schedule

    Sources12

    Practise it for real

    Certify one table, deprecate another, and confirm that both show up correctly in search

    1. 1.In the SQL editor, run SET TAG ON TABLE on a table you own, setting system.certification_status = certified.

      Why: Certification is a tag, so a tagging statement is all it takes.

      You should see: The statement succeeds, and the table shows a check mark in Catalog Explorer. If it fails with a permission error, you are missing ASSIGN on the governed tag.

    2. 2.Set the same tag to deprecated on a second table you own.

      Why: Shows the other lifecycle value and its icon.

      You should see: The second table shows the restricted icon next to its name.

    3. 3.In workspace search, run type:table certificationStatus:certified, then certificationStatus:deprecated.

      Why: Confirms that search filters on the tag value.

      You should see: Each table appears under its own status. Allow a few minutes for the change to reach search.

    4. 4.Open the first table's kebab menu, choose Assign certification, select None and save.

      Why: Status can be changed or removed at any time.

      You should see: The check mark disappears, and after a short delay the table no longer appears in the certified search.

    Stuck? Get a nudge

    If search still shows the old status, wait a few minutes. Certification updates take a little while to reach search results.

    Sources

    (8)

    Every claim above is drawn from one of these pages, quoted as it was written on the date shown.

    1. 1.
      “certified: Indicates that a data asset has met internal standards for accuracy, completeness, and trust.”
      ↩︎ What the certified and deprecated badges mean
      “deprecated: Warns that a data asset is outdated, no longer reliable, or should not be used in new workflows.”
      ↩︎ What the certified and deprecated badges mean
      “The tag is displayed next to object names in the workspace, and influences how data appears in notebooks and the SQL editor.”
      ↩︎ What the certified and deprecated badges mean
      “This helps Genie One improve its answers by prioritizing the assets your organization vouches for.”
      ↩︎ What the certified and deprecated badges mean
      “In the search filters, you can also select Certified or Deprecated from the Certification status filter menu.”
      ↩︎ Searching for certified data, and Discover versus Catalog Explorer
      “It might take a few minutes for certification or deprecation updates to appear in search results.”
      ↩︎ Searching for certified data, and Discover versus Catalog Explorer
      “You must have the ASSIGN permission on the system.certification_status governed tag to apply it to objects.”
      ↩︎ Certifying or deprecating an asset in the UI or SQL
      “Click the kebab menu and select Assign certification.”
      ↩︎ Certifying or deprecating an asset in the UI or SQL
      “An automation assigns the system.certification_status tag to every Unity Catalog table or volume that matches conditions you define”
      ↩︎ Certifying and deprecating many tables at once
      “It matches every table in the temp catalog with no read or write queries in the last 30 days”
      ↩︎ Certifying and deprecating many tables at once
      “It is a system-governed tag with two tag values: certified and deprecated.”
      ↩︎ Key concept
      “In the Search field, use the certificationStatus keyword to query objects by their certification status.”
      ↩︎ Exam trap 1
      “You must have the ASSIGN permission on the system.certification_status governed tag to apply it to objects.”
      ↩︎ Exam trap 2
      “Search using tags is not supported on dashboards, Genie Agents, or Databricks apps.”
      ↩︎ Exam trap 4
      “The tag key is system.certification_status and there are two tag values:”
      ↩︎ Checkpoint
      “In the Search field, use the certificationStatus keyword to query objects by their certification status.”
      ↩︎ Prediction
      “Search using tags is not supported on dashboards, Genie Agents, or Databricks apps.”
      ↩︎ Checkpoint
    2. 2.
      “Users cannot modify or delete system tag keys or values.”
      ↩︎ What the certified and deprecated badges mean
      “In Databricks Runtime 13.3 and above, use the ALTER <object> SQL command with SET TAGS or UNSET TAGS to manage tags on securable objects.”
      ↩︎ Certifying or deprecating an asset in the UI or SQL
      “In Databricks Runtime 16.1 and above, use SET TAG and UNSET TAG to manage tags on securable objects.”
      ↩︎ Certifying or deprecating an asset in the UI or SQL
      “Implicit tag inheritance occurs when evaluating ABAC policies only. Tag inheritance doesn't apply generally.”
      ↩︎ Certifying or deprecating an asset in the UI or SQL
      “preview the results with a dry run, then enable the automation to tag safely on a schedule”
      ↩︎ Certifying and deprecating many tables at once
      “Implicit tag inheritance occurs when evaluating ABAC policies only. Tag inheritance doesn't apply generally.”
      ↩︎ Exam trap 3
      “Tell Genie your criteria for certifying or deprecating assets, preview the results with a dry run, then enable the automation”
      ↩︎ Checkpoint
    3. 3.
      “Point users at the Discover page when they want the assets your organization recommends, and at Catalog Explorer when they need the complete picture”
      ↩︎ Searching for certified data, and Discover versus Catalog Explorer
      “Unity Catalog captures lineage automatically for queries run on Databricks and displays it in Catalog Explorer.”
      ↩︎ Inspecting a certified table before you query it
      “so users can trace where data comes from and how it's used before they trust it”
      ↩︎ Checkpoint
    4. 4.
      “Click View all under any asset type to browse all assets of that type, including Certified assets and Assets suggested by AI.”
      ↩︎ Searching for certified data, and Discover versus Catalog Explorer
    5. 5.
      “Select the Sample Data tab to view sample data. You must have access to active compute to sample data.”
      ↩︎ Inspecting a certified table before you query it
      “Select the Details tab to review table details, including table properties.”
      ↩︎ Inspecting a certified table before you query it
    6. 6.
      “You can view comments in Catalog Explorer or using the DESCRIBE command for an object.”
      ↩︎ Inspecting a certified table before you query it
    7. 7.
      “You can filter by notebooks, jobs, pipelines, and queries.”
      ↩︎ Inspecting a certified table before you query it
      “Select the Lineage tab. The lineage panel appears and displays related tables.”
      ↩︎ Exam trap 5
    8. 8.
      “Access control in Unity Catalog works by granting privileges, such as SELECT, MODIFY, or USE SCHEMA, on these securable objects.”
      ↩︎ Certifying or deprecating an asset in the UI or SQL

    Spotted a mistake, or was something unclear? Tell us.