Clarification on Expected Behaviour for Data Quality Rules on Standalone Data Assets and Data Products

Question

Clarification on Expected Behaviour for Data Quality Rules on Standalone Data Assets and Data Products

Anitha Ganesan 0

Hi Microsoft Team,

We are seeking clarification on the intended behaviour of Microsoft Purview Data Quality when using standalone Data Assets and Data Products.

We have reviewed the following Microsoft documentation:

Microsoft Tech Community – Data Quality (DQ) for Standalone Data Assets at Microsoft Purview https://techcommunity.microsoft.com/blog/microsoft-security-blog/data-quality-dq-for-standalone-data-assets-at-microsoft-purview/4495338
Microsoft Learn – Configure a data quality scan for a data asset https://learn.microsoft.com/purview/unified-catalog-data-quality-scan-asset

From our understanding of the documentation, a standalone Data Asset can be profiled, have Data Quality (DQ) rules configured, and then be associated with one or more Data Products. The documentation also states that the associated Data Product can reuse the Data Quality rules and quality scores created for the standalone asset.

We would like to confirm whether this is the intended product behaviour.

During our UAT testing, we tested the following scenario:

Registered and scanned a Data Asset into the Data Map.
Added the Data Asset to a Governance Domain as a standalone Data Asset.
Ran profiling and configured Data Quality rules on the standalone Data Asset.
Executed the profiling and Data Quality scans successfully on the standalone Data Asset.
Associated the standalone Data Asset with a Data Product.

However, after associating the standalone Data Asset with the Data Product, we observed inconsistent behaviour, including:

Data Quality rules appeared as duplicate entries within the Data Product.
In some cases, duplicate rule entries appeared without associated quality scores.
Profiling statistics that were available on the standalone Data Asset were not visible after the asset was associated with the Data Product.
After rerunning Data Quality scans within the Data Product context, additional duplicate rule entries were intermittently observed, resulting in multiple instances of the same rule.

Could you please clarify the following?

Is the intended design that Data Quality rules, profiling results, and quality scores configured on a standalone Data Asset are automatically reused when that asset is associated with one or more Data Products (including when the same asset is associated with multiple Data Products)?
The behaviour we observed during UAT differed from our understanding of the documentation (duplicate rules, missing profiling statistics, missing quality scores, and additional duplicate rule entries after rerunning scans). Are these known issues, expected behaviour, or product limitations? If not, what is the recommended approach for this scenario?

If there are any recommended practices, known fixes, or documentation covering this scenario, we would appreciate your guidance.

Thank you for your assistance.

Ganesh Chelluri 190 Reputation points Microsoft External Staff Moderator

2026-07-16T23:24:51.52+00:00
HI @Anitha Ganesan ,

Q1 - Intended design? Yes. DQ rules, profiling, and scores are stored at the data asset level. When you associate the asset with one or more Data Products, the same rules/scores are meant to surface on each Data Product's DQ view no duplication, no re-authoring.

Q2 - What you observed? Not expected behavior. In every reproducible case, it traces to two documented Purview known issues:

Hierarchy auto-expansion the same asset gets associated more than once under the Data Product (especially via semantic model / dataset hierarchies).

"Missing duplication checks" on business concepts the platform doesn't block the duplicate association.

That combo produces exactly what you see: duplicate rule rows, phantom rows without scores, missing profile, and more duplicates on rerun (each rerun writes against every asset instance).

Fix to try before opening a case:

Data Product → Data assets → confirm each asset appears exactly once by FQN. Remove any duplicate.

Catalog management → Data products → asset details → Refresh (workaround for the "metadata don't autorefresh" known issue).

Reopen DQ view rules, profile, and last score should now match the standalone asset 1:1.

Run one fresh DQ scan from the Data Product context should update scores, not add rule rows.

If duplicates or missing profile persist after step 3, it's not documented as expected please raise a support case with governance domain, Data Product name, asset FQN, scan run ID, and UTC timestamp, and reference the two known issues above.Q1 - Intended design? Yes. DQ rules, profiling, and scores are stored at the data asset level. When you associate the asset with one or more Data Products, the same rules/scores are meant to surface on each Data Product's DQ view no duplication, no re-authoring.

Q2 - What you observed? Not expected behavior. In every reproducible case, it traces to two documented Purview known issues:

Hierarchy auto-expansion the same asset gets associated more than once under the Data Product (especially via semantic model / dataset hierarchies).

"Missing duplication checks" on business concepts the platform doesn't block the duplicate association.

That combo produces exactly what you see: duplicate rule rows, phantom rows without scores, missing profile, and more duplicates on rerun (each rerun writes against every asset instance).

Fix to try before opening a case:

Data Product → Data assets → confirm each asset appears exactly once by FQN. Remove any duplicate.

Catalog management → Data products → asset details → Refresh (workaround for the "metadata don't autorefresh" known issue).

Reopen DQ view rules, profile, and last score should now match the standalone asset 1:1.

Run one fresh DQ scan from the Data Product context should update scores, not add rule rows.

If duplicates or missing profile persist after step 3, it's not documented as expected please raise a support case with governance domain, Data Product name, asset FQN, scan run ID, and UTC timestamp, and reference the two known issues above.
Ganesh Chelluri 190 Reputation points Microsoft External Staff Moderator

2026-07-18T00:17:57.17+00:00
Hi @Anitha Ganesan ,

Adding a short verification checklist so the deduplication result is deterministic across your governance domains and easy to hand off to whoever runs UAT next.

Verification checklist (do this per affected Data Product)

Confirm asset uniqueness under the Data Product Purview portal → Unified Catalog → Governance domains → your domain → Data products → open the Data Product → Data assets. Verify each impacted asset appears exactly ONCE by fully qualified name (FQN). Watch for near-duplicates coming from semantic-model / dataset hierarchy expansion (same table surfaced twice via different parent paths). Remove any duplicate association.

Force a metadata refresh (workaround for the documented autorefresh gap) Catalog management → Data products → open the asset detail page inside the product → Refresh. This pulls the latest owner / description / lineage / profile from Data Map, and in most tenants also reconciles the DQ view.

Validate the DQ view 1:1 against the standalone asset On the standalone asset side (Governance domain → Data assets): Note the exact rule count, rule names, and the last quality score. On the Data Product side: Open the same asset's DQ tab and confirm: Rule count matches exactly. Profile tab shows the same column-level stats. Overview shows the same last quality score. If any of these three differ AFTER steps 1–2, that's the signal it isn't just the known duplicate-association pattern.

Run one controlled DQ scan from the Data Product context Trigger a single fresh scan. Expected result: existing rules get updated scores; no new rule rows are added; the same asset is scored once. If new rule rows appear, capture the scan run ID and timestamp (UTC).

If steps 3 or 4 still show duplicates, missing profile, or missing scores after the deduplication + refresh Open a Microsoft support case with: Governance domain name Data Product name Data asset FQN Scan run ID from Monitoring UTC timestamp of the failing scan Screenshots of Rules tab before/after And reference the two Purview known issues that cover the surrounding gaps: a. Hierarchy auto-expansion under Data Products (semantic model / dataset paths). b. "Missing duplication checks" on business concepts (platform doesn't block duplicate association).

Direct answers restated

Q1. Intended design?

Yes. DQ rules, profiling results, and quality scores are stored at the data asset

level. When the asset is associated with one or more Data Products, the same

rules and scores are meant to surface on each Data Product's DQ view - no

duplication, no re-authoring, including multi-Data-Product associations.

Q2. What you saw during UAT?

Not the intended behavior. In every reproducible case reported so far it maps to

the two known-issue patterns above, plus the metadata autorefresh gap. The

checklist above resolves it in the vast majority of tenants. If it persists after

that, treat it as a fresh defect worth logging via support.

References

Data Quality overview: https://learn.microsoft.com/purview/unified-catalog-data-quality
Configure a DQ scan for a data asset: https://learn.microsoft.com/purview/unified-catalog-data-quality-scan-asset

Configure a DQ scan for a data product: https://learn.microsoft.com/purview/unified-catalog-data-quality-scan

Data governance known issues: https://learn.microsoft.com/purview/data-governance-known-issues

Tech Community – DQ for Standalone Data Assets: https://techcommunity.microsoft.com/blog/microsoft-security blog/data-quality-dq-for-standalone-data-assets-at-microsoft-purview/4495338

1 answer

Your answer

Answer 1

Sina Salam 31,296 Volunteer Moderator

Hello Anitha Ganesan,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that you will need clarification on Expected Behaviour for Data Quality Rules on Standalone Data Assets and Data Products.

In a normal circumstances standalone asset’s Data Quality rules, profiling results, and score should not be treated as a live synchronized result set across every Data Product. - https://learn.microsoft.com/en-us/purview/unified-catalog-data-quality-scan-asset The duplicate rules, missing scores, and missing profiling statistics are not a confirmed expected final state. They are most likely caused by stale Data Product asset metadata, duplicate asset association, hierarchy expansion when adding assets, or a backend product issue. - https://learn.microsoft.com/en-us/purview/data-governance-known-issues

To resolve this, I will suggest you remove duplicate asset associations, refresh the asset in the Data Product, reimport schema if the source schema changed, run profiling in the Data Product context if profiling results are required there, and then run the Data Quality scan locally from that Data Product. Profiling and Data Quality scans are context-based operations in Purview Health Management, so the Data Product view should be validated from the Data Product’s own Data Quality page, not only from the standalone asset page. - https://learn.microsoft.com/en-us/purview/unified-catalog-data-quality-profiling. If duplicate rule rows or blank scores still appear after the above process, this should be treated as an unresolved product issue, not as expected behaviour. At that point, the correct next step for you is to open a Microsoft Support case via Azure Portal or contact priority customer support (PCS) with details. Microsoft Support or the Purview Product Group must then validate the backend metadata state. Use the associated links for more reading and steps.

I hope this is helpful. Please! Do not hesitate to let me know if you have any other questions, steps or clarifications.

Please do not close the thread by upvoting and accepting the answer if any part of it is helpful.

Anitha Ganesan 0 Reputation points

2026-07-23T08:40:55.1166667+00:00

Hi @Sina Salam

Thanks for the comment. I have one clarification regarding DQ rule synchronization.

We validated that when a standalone Data Asset is first associated with a Data Product, the existing DQ rules are correctly reflected after a metadata refresh. However, the profiling statistics are not, and I will open a support case for that issue.

However, we also observed that if a DQ rule is later added, updated, or deleted on the standalone Data Asset, those changes are not automatically reflected in the associated Data Products.

Can you please confirm whether this is the expected behaviour, or whether subsequent DQ rule changes should automatically synchronize to the associated Data Products?
Sina Salam 31,296 Reputation points Volunteer Moderator

2026-07-23T13:34:51.07+00:00

Hello @Anitha Ganesan ,

Thanks for the clarification.

Based on the current Microsoft Purview Data Quality behaviour, DQ rule changes made later on a standalone Data Asset should not be expected to automatically synchronize to Data Products that were already associated with that asset.

When a standalone Data Asset is first associated with a Data Product, its existing Data Quality information, such as the latest score and related rule context, may be reflected in the Data Product after refresh. However, after association, the Data Product should be treated as a separate Data Quality evaluation context. This is because the same asset can be used by multiple Data Products, and each Data Product may have different business expectations for data quality. Microsoft documentation also states that a Data Quality scan must be run locally in the target Data Product to evaluate and update the quality score in that context. https://learn.microsoft.com/en-us/purview/unified-catalog-data-quality-scan-asset

Therefore, if a DQ rule is later added, changed, or deleted on the standalone Data Asset, that change should not be assumed to automatically update the already-associated Data Products. The recommended approach is to refresh the asset metadata in the Data Product, review the Rules tab in that Data Product context, make any required rule changes there, and then rerun the Data Quality scan locally from the Data Product.

The missing profiling statistics after association should still be raised with Microsoft Support if it continues after metadata refresh and local profiling or DQ scans. Likewise, duplicate rule entries, blank scores, or repeated rule instances after rerunning scans should be treated as inconsistent behaviour and investigated by Microsoft Support. When opening the case, include the governance domain, Data Product name, asset fully qualified name, scan run ID, timestamps, screenshots, and details of the steps used to reproduce the issue.

Final clarification about your question: Initial association can reuse or reflect existing Data Quality information, but it is not a live synchronization mechanism. Each associated Data Product should be managed and scanned as its own Data Quality context.

Cheers
Anitha Ganesan 0 Reputation points

2026-07-23T14:37:26.8633333+00:00

Hi @Sina Salam ,

Thanks for the clarification.

I have one additional question regarding the recommended approach.

I understand that if different Data Products have different Data Quality rules or business expectations, each Data Product should be managed and scanned independently.

In our scenario, however, the same Data Asset may be associated with multiple Data Products, and the intention maybe to use the same set of DQ rules and business expectations across all of them to avoid duplicate maintenance.

In this scenario, is it still recommended to run profiling and Data Quality scans locally within each associated Data Product, or can the profiling results and DQ scores from the standalone Data Asset be reused?

We are trying to understand whether local scans are still required in this scenario, as they would introduce additional effort and potentially additional scan and profiling costs.
Sina Salam 31,296 Reputation points Volunteer Moderator

2026-07-23T17:53:01.2466667+00:00
Hello @Anitha Ganesan ,

Thank you once again for your clarification.

The only evidence based I can give you is:

Current Microsoft documentation does not document support for reusing standalone asset profiling results or Data Quality scores across multiple Data Products. Until Microsoft documents such a capability, the supported and safest assumption is that each Data Product should be scanned independently if it requires its own Data Quality results.

https://learn.microsoft.com/en-us/purview/unified-catalog-data-quality-scan-asset

https://learn.microsoft.com/en-us/purview/unified-catalog-data-quality-profiling

However, if you have any workaround to implement your scenario and it works for you, that will be fine.

Cheers
Anitha Ganesan 0 Reputation points

2026-07-28T15:29:58.64+00:00
Hi @Ganesh Chelluri / @Sina Salam

Thank you for your responses and clarification. We performed some additional validation and observed the following behaviour.
We are using standalone Data Assets so that the same Data Quality rules and business expectations can be reused across multiple Data Products, rather than maintaining the same DQ rules separately for each Data Product.
During the initial setup, we configured DQ rules on a standalone Data Asset and associated it with a Data Product. After refreshing the Data Product → Data Assets, the Data Product correctly reflected the latest DQ configuration from the standalone Data Asset.
Later, after adding a new DQ rule to the standalone Data Asset, we understood that changes made after the initial association are not synchronized automatically to the Data Product. As a workaround, we removed the standalone Data Asset from the Data Product and re-associated it so that the latest DQ configuration would be reflected.
However, instead of refreshing the existing DQ rules, each re-association retained the existing DQ rules and appended the current set of DQ rules from the standalone Data Asset, resulting in duplicate DQ rules.

Observed behaviour:

Initial association: 2 DQ rules (both with DQ scores)

Added a third DQ rule to the standalone Data Asset

First re-association: 5 DQ rules (3 with DQ scores, 2 without DQ scores)

Second re-association: 8 DQ rules (3 with DQ scores, 5 without DQ scores)

Third re-association: 11 DQ rules (3 with DQ scores, 8 without DQ scores)

Could you please clarify whether this behaviour is related to the known issue regarding hierarchy expansion during Data Product asset association, or if it should be treated as a separate product issue?
Ganesh Chelluri 190 Reputation points Microsoft External Staff Moderator

2026-07-30T23:38:28.46+00:00

Hi @Anitha Ganesan ,

Based on the current Microsoft documentation, the behavior you described is not documented as expected behavior.

Repeatedly removing and re-associating the same Data Asset should not continuously append duplicate DQ rule entries. The observed growth in rule count after each re-association suggests duplicate metadata associations rather than a supported synchronization mechanism.

This appears more consistent with the documented Purview known issues related to:

• Automatic hierarchy expansion when adding assets to Data Products.

• Missing duplication checks during creation/association workflows.

If the asset appears only once in the Data Product and the issue is still reproducible, I would recommend treating this as a potential product defect and opening a Microsoft Support case with the Governance Domain, Data Product name, Asset FQN, scan run IDs, timestamps, and screenshots showing the rule count after each re-association.