Microsoft Purview Information Protection
Labels and classificationAuto Label Simulation Results
In brief
Updated Microsoft Purview documentation in Purview/auto-label-simulation-results.md.
What Purview admins need to know
Review the underlying documentation change to determine whether it affects tenant configuration or rollout plans.
Summaries are generated from the documentation change itself.
Documentation change
View on Microsoft LearnThe comparison below shows only the changed extract. Use the full-page view for complete context.
audience: Admin
ms.service: purview
ms.subservice: purview-information-protection
ms.date: 2026-10-14T00:11-06T00:00:00.0000000Z
ms.collection:
- purview-compliance
ms.topic: article
Field Description Label The sensitivity label that the policy applies to matched items. Status Simulation status: Simulation in progress, Simulation almost complete, Simulation complete, Simulation expiring soon, Expired, orPolicy can't be published, or Simulation failed. If a failure banner is displayed, review it for more information.Duration The length of time the simulation has been running in days and hours. Matched items The totalnumber of items that matched the policyconditionsconditions. A count marked (partial) includes only available workload results, not a complete total across all configured workloads. See Failed simulations and partial results.Sample items When displayed, the number of sampled items available for review. Samples are retained for 30 days. See Preserve results before editing or rerunning. Policy matches per rule Shows details about the policy rules and their matches: - Total Rules: the number of rules in the policy, not the number of rules with matches. - Total Matches: the number of items that matched rules. - Rule name: the name of the rule in the policy. - Matches: the number of items that matched this rule. - Location: The workload where matches occurred (Exchange, SharePoint, or OneDrive). Use the per-rule match counts to identify which rules matched and whether each rule produces the expected results. Matches by source A donut chart showing matches distributed across workloads, with an item count and percentage for each of the following workloads: - Exchange: Email messages that matched the policy. Matched Exchange items are based on sampled data and might not represent all mailboxes in the policy scope. - SharePoint: Files in SharePoint sites that matched. - OneDrive: Files in OneDrive accounts that matched.
The Matched items count is an estimate, not a count of successfully labeled files. Enforcement results can differ because files or classification results change, other active policies affect the outcome, or labeling fails. For details, see Simulation doesn't guarantee enforcement.
Failed simulations and partial results
In portal experiences that display simulation failure messages, Simulation couldn't be started means that the request to start a simulation job failed. Simulation failed means that a job encountered an error while running. Neither message means that the policy completed simulation with no matches.
Valid results from other workloads can remain available. For example, Exchange can show matches while SharePoint and OneDrive show Not available. A count such as 100 (partial) includes only the available results. Check Matches by source and the information icon beside a partial count, when displayed, to understand which results are included. Not available isn't a zero-match result.
Before selecting Restart simulation in the page header, preserve any available results you need for comparison. Restarting tries the simulation again; it doesn't guarantee a fix. Review the new run's status and workload results before deciding whether the policy is ready to turn on.
For the error messages, recovery steps, and information to collect if a failure persists, see Simulation failed or couldn't be started. Simulation failures are separate from failures to apply or remove labels after a policy is turned on.
Preserve results before editing or rerunning
Before you edit an auto-labeling policy or restart simulation, export any available results that you need for later comparison. Check the exported file before you make changes.
@@ -14,7 +14,7 @@ manager: laurawi audience: Admin ms.service: purview ms.subservice: purview-information-protection-ms.date: 2026-10-14T00:00:00.0000000Z+ms.date: 2026-11-06T00:00:00.0000000Z ms.collection: - purview-compliance ms.topic: article@@ -78,9 +78,9 @@ On the **Simulation overview** page, the following key metrics about the policy | Field | Description |
| --- | --- |
| **Label** | The sensitivity label that the policy applies to matched items. |
-| **Status** | Simulation status: **Simulation in progress**, **Simulation almost complete**, **Simulation complete**, **Simulation expiring soon**, **Expired**, or **Policy can't be published**. |
+| **Status** | Simulation status: **Simulation in progress**, **Simulation almost complete**, **Simulation complete**, **Simulation expiring soon**, **Expired**, **Policy can't be published**, or **Simulation failed**. If a failure banner is displayed, review it for more information. |
| **Duration** | The length of time the simulation has been running in days and hours. |
-| **Matched items** | The total number of items that matched the policy conditions across all workloads. |
+| **Matched items** | The number of items that matched the policy conditions. A count marked **(partial)** includes only available workload results, not a complete total across all configured workloads. See Failed simulations and partial results. |
| **Sample items** | When displayed, the number of sampled items available for review. Samples are retained for 30 days. See Preserve results before editing or rerunning. |
| **Policy matches per rule** | Shows details about the policy rules and their matches: - **Total Rules:** the number of rules in the policy, not the number of rules with matches. - **Total Matches:** the number of items that matched rules. - **Rule name:** the name of the rule in the policy. - **Matches:** the number of items that matched this rule. - **Location:** The workload where matches occurred (Exchange, SharePoint, or OneDrive). Use the per-rule match counts to identify which rules matched and whether each rule produces the expected results. |
| **Matches by source** | A donut chart showing matches distributed across workloads, with an item count and percentage for each of the following workloads: - **Exchange**: Email messages that matched the policy. Matched Exchange items are based on sampled data and might not represent all mailboxes in the policy scope. - **SharePoint**: Files in SharePoint sites that matched. - **OneDrive**: Files in OneDrive accounts that matched. |
@@ -92,6 +92,16 @@ Note
The **Matched items** count is an **estimate**, not a count of successfully labeled files. Enforcement results can differ because files or classification results change, other active policies affect the outcome, or labeling fails. For details, see [Simulation doesn't guarantee enforcement](auto-label-classification-simulation#simulation-doesnt-guarantee-enforcement).
+## Failed simulations and partial results
+
+In portal experiences that display simulation failure messages, **Simulation couldn't be started** means that the request to start a simulation job failed. **Simulation failed** means that a job encountered an error while running. Neither message means that the policy completed simulation with no matches.
+
+Valid results from other workloads can remain available. For example, Exchange can show matches while SharePoint and OneDrive show **Not available**. A count such as **100 (partial)** includes only the available results. Check **Matches by source** and the information icon beside a partial count, when displayed, to understand which results are included. **Not available** isn't a zero-match result.
+
+Before selecting **Restart simulation** in the page header, preserve any available results you need for comparison. Restarting tries the simulation again; it doesn't guarantee a fix. Review the new run's status and workload results before deciding whether the policy is ready to turn on.
+
+For the error messages, recovery steps, and information to collect if a failure persists, see [Simulation failed or couldn't be started](/en-us/troubleshoot/microsoft-365/purview/sensitivity-labels/troubleshoot-auto-labeling-policies#simulation-failed-or-couldnt-be-started). Simulation failures are separate from failures to apply or remove labels after a policy is turned on.
+
## Preserve results before editing or rerunning
Before you edit an auto-labeling policy or restart simulation, export any available results that you need for later comparison. Check the exported file before you make changes.