Microsoft is planning a new Microsoft Purview capability called Classifier Simulation Mode, the first component of a broader Classifier Health Monitoring platform. According to the Microsoft 365 Roadmap entry, the feature will let organizations test and analyze custom Purview classifiers against production data before publishing them for live use.
The roadmap item, ID 523201, was updated on July 14, 2026 and remains marked In development. Microsoft lists public preview for October 2026 and general availability for November 2026, initially for worldwide standard multi-tenant tenants through the Purview web experience. As with all roadmap dates, those targets can change.
Custom classifiers are used in Purview Information Protection to identify organization-specific sensitive content, commonly through regular expressions, keyword logic, or other detection criteria. They can underpin sensitivity labeling, data loss prevention, and related compliance controls.
The problem is that a classifier can look sound in a test case but behave badly at tenant scale. A broad pattern may match far more content than intended, producing excessive detections and false positives. Poorly constructed regular expressions can also add scanning latency. Those outcomes can make downstream policy actions noisy and difficult to trust.
Microsoft says Simulation Mode will allow administrators to validate classifier quality and performance on production data before committing the classifier to a published state. In practical terms, that should give compliance teams a chance to inspect likely match volume, assess whether the logic is too broad, and catch performance concerns before a classifier begins influencing live labeling or protection workflows.
Simulation Mode is intended to reduce that gap by putting the validation stage closer to real tenant content without immediately applying a live classifier. Microsoft frames the feature as a way to improve classifier precision and health while avoiding high-match “noisy” rules and slow scans.
The roadmap does not yet specify the simulation report format, retention behavior, licensing requirements, permissions model, or whether simulations will cover every custom-classifier type and policy integration at launch. It also does not say whether simulation activity will be exposed through Purview audit or reporting tools.
For now, the feature is a roadmap commitment rather than a shipping control, with preview currently scheduled for October 2026 and general availability targeted for November 2026.
The roadmap item, ID 523201, was updated on July 14, 2026 and remains marked In development. Microsoft lists public preview for October 2026 and general availability for November 2026, initially for worldwide standard multi-tenant tenants through the Purview web experience. As with all roadmap dates, those targets can change.
A pre-production check for custom classifiers
Custom classifiers are used in Purview Information Protection to identify organization-specific sensitive content, commonly through regular expressions, keyword logic, or other detection criteria. They can underpin sensitivity labeling, data loss prevention, and related compliance controls.The problem is that a classifier can look sound in a test case but behave badly at tenant scale. A broad pattern may match far more content than intended, producing excessive detections and false positives. Poorly constructed regular expressions can also add scanning latency. Those outcomes can make downstream policy actions noisy and difficult to trust.
Microsoft says Simulation Mode will allow administrators to validate classifier quality and performance on production data before committing the classifier to a published state. In practical terms, that should give compliance teams a chance to inspect likely match volume, assess whether the logic is too broad, and catch performance concerns before a classifier begins influencing live labeling or protection workflows.
Why it matters to Purview admins
Purview deployments often depend on classifiers that reflect internal formats: customer identifiers, engineering project codes, case numbers, records templates, or locally defined confidential data. Those patterns tend to need iteration. Today, making that iteration safely can mean limited test data, staged policies, or accepting some uncertainty when a classifier is promoted.Simulation Mode is intended to reduce that gap by putting the validation stage closer to real tenant content without immediately applying a live classifier. Microsoft frames the feature as a way to improve classifier precision and health while avoiding high-match “noisy” rules and slow scans.
The roadmap does not yet specify the simulation report format, retention behavior, licensing requirements, permissions model, or whether simulations will cover every custom-classifier type and policy integration at launch. It also does not say whether simulation activity will be exposed through Purview audit or reporting tools.
What to do now
There is no tenant action required while the item remains in development. Purview administrators should, however, identify custom classifiers that generate frequent false positives or are based on complex regex expressions, since those are the most obvious candidates for early testing once preview arrives.For now, the feature is a roadmap commitment rather than a shipping control, with preview currently scheduled for October 2026 and general availability targeted for November 2026.
References
- Primary source: Microsoft 365 Roadmap
Published: 2026-07-14T22:41:38.6349466Z
Microsoft 365 Roadmap | Microsoft 365
The Microsoft 365 Roadmap lists updates that are currently planned for applicable subscribers. Check here for more information on the status of new features and updates.www.microsoft.com
