Microsoft Purview Data Quality: Scan and Monitor Data Health

If you’re working with data governance, you’ve likely asked: How can I measure and improve data quality at scale? With Microsoft Purview, data stewards can now run data quality scans to assess, track, and manage the health of data assets across the organization.

In this tutorial, we’ll walk through how to set up data quality scans, create rules, monitor results, and even schedule alerts all within the unified data catalog of Microsoft Purview.

For hands-on Power BI training, visit Power BI Training by Data Bear.

Why Data Quality Scans Matter

High-quality data is the foundation of trustworthy analytics. With Microsoft Purview’s health management features, you can:

• Identify blank or missing values
• Validate relationships between fact and dimension tables
• Detect outliers
• Track data quality over time
• Trigger alerts based on rule failures

Step 1: Establish a Data Connection for Quality Scans

Before running a scan, you must connect Purview to the actual data source, not just its metadata.

To do this:

1. Navigate to Health Management > Data Quality in Microsoft Purview.
2. Create a new connection, specifying the workspace and lakehouse IDs.
3. Use Managed Identity as the credential.
4. Grant that identity contributor access to your Fabric workspace.

This gives Purview secure access to scan the actual data for quality issues.

Step 2: Profile Your Dataset

Choose the data asset you want to assess. In the example, a lakehouse table called gold_factvideostats is selected.

1. Click Profile Data from the data quality overview.
2. Select all or specific columns to include in the profile.
3. Run the profiling job to generate basic statistics and quality metrics.

Step 3: Create Data Quality Rules

Now that the dataset is profiled, add validation rules to detect issues.

Use Built-In Rules

Purview provides predefined rules such as:

• No empty or blank fields
• Valid date values
• Table lookups (e.g., foreign key validation)

For example:

• Ensure capture_date is present for every video
• Confirm channel_id in the fact table exists in the related dimension table

Create Custom Rules

You can also define custom rules using expression language. This allows more advanced or domain-specific validations.

Step 4: Run the Quality Scan

Once your rules are set:

• Toggle on/off specific rules
• Execute the scan
• Monitor progress in Manage > Monitoring

Results are returned with a quality score (e.g., 91.8%), highlighting issues like:

• Blank values
• Outliers
• Failed rule validations

Step 5: Investigate and Assign Issues

After running scans:

• Review flagged issues under the Actions section
• Assign items to team members for resolution
• Track issue progress over time

You can also explore profiling results, view schema-level metrics, and see which columns scored poorly.

Step 6: Set Up Alerts and Schedules

To ensure ongoing monitoring:

1. Schedule scans daily, weekly, or monthly
2. Create alerts for rule failures or scan errors
3. Define notification recipients (e.g., data product owners)

Example:

• Alert Patrick (data owner) if capture_date is ever blank
• Alert on any scan failure tied to the videostats table

Step 7: Use Reports and Data Observability

Purview includes prebuilt reports for:

• Term of use compliance
• Glossary publication gaps
• Overall domain health

And with data observability (in preview), you can:

• Visualize asset lineage across domains
• Track data quality trends at a domain level

This gives data stewards, product owners, and consumers a comprehensive overview of data readiness.

Why This Matters for Data Consumers

With data quality visibility:

• Consumers can determine if data is fit for purpose
• Stewards can prioritize cleanup tasks
• Organizations can build trust in analytics

Whether you’re monitoring critical datasets or implementing governance best practices, data quality scans in Microsoft Purview make it actionable and scalable.

Final Thoughts

Microsoft Purview’s data quality capabilities empower organizations to treat data as a managed product. By profiling, validating, and monitoring data health, teams can ensure that their data is accurate, complete, and ready for business use.

Start improving your data quality posture today. And for end-to-end Power BI skills, check out Power BI Training by Data Bear.