Share via

Purview Exchange Mailbox (at-rest) Data Discovery and Classification

AzGeek 0 Reputation points
2026-02-26T04:15:42.93+00:00

Hello,

I understand that currently Purview Information Protection (auto-labeling policies with SIT's as criteria) and Purview eDiscovery and Content Search may have limitations to scanning and classifying content in Exchange online emailboxes at-rest.

This use case is for adherance with comlipliance regulations to ensure data is not retained for longer than necessary and so that sensitive data within older emails can be known.

Can we confirm whether there is any workaround/alternative solution to scan and classify emails based on any sensitive data held in the email?

Are there any other Purview Modules which can be leveraged for scanning and classifying stored emails?

Microsoft Security | Microsoft Purview
{count} votes

2 answers

Sort by: Most helpful
  1. Pilladi Padma Sai Manisha 5,240 Reputation points Microsoft External Staff Moderator
    2026-02-26T12:48:27.0666667+00:00

    Hi AzGeek

    Thank you for reaching out microsoft Q&A!. You are correct that today there are platform limitations in Microsoft Purview when it comes to automatically scanning and classifying Exchange Online mailbox content at-rest using Sensitive Information Types (SITs).

    Current capability At present, Purview Information Protection auto-labeling policies don’t retroactively apply sensitivity labels to emails already stored in Exchange Online mailboxes. Auto-labeling primarily applies to SharePoint/OneDrive content and to emails during send/receive or when they are modified. Similarly, eDiscovery and Content Search help locate data but don’t perform continuous classification or automatic labeling of mailbox items.

    Recommended alternatives / supported approach Although there isn’t a direct workaround to auto-classify historical emails, you can achieve the compliance objective by combining several Purview capabilities:

    Microsoft Purview Data Loss Prevention (DLP): Use DLP policies with SITs to detect sensitive information in Exchange and generate alerts or enforcement actions. This provides ongoing visibility even though labels aren’t applied retroactively.

    eDiscovery or Content Search: Run targeted searches to identify older emails that may contain sensitive content and perform review or remediation workflows as required.

    Retention policies and retention labels: Implement lifecycle management to ensure emails are retained only for the required regulatory period and automatically deleted afterward.

    Exchange mail flow rules / future auto-labeling: Use these for forward-looking protection so that new or modified emails are classified going forward.

    Other Purview modules Modules such as Data Classification Analytics or Insider Risk Management provide reporting and activity insights but don’t automatically classify existing mailbox content at rest.

    there isn’t currently a Purview feature that fully scans and auto-labels historical Exchange Online emails. The recommended design is to use DLP for detection, eDiscovery for discovery and remediation, and retention policies for lifecycle governance while applying auto-labeling to new email activity moving forward.

    Please let us know if you’d like guidance designing a policy approach aligned to your compliance requirements, and we’ll be happy to assist further.

    As for other Purview Modules, here are some related resources:


  2. Vasil Michev 125.4K Reputation points MVP Volunteer Moderator
    2026-02-26T07:13:33.7633333+00:00

    We've been asking about this for years, still not supported, afaik. If your primary goal is to retain data, you should be able to come up with retention policies with adequate conditions. Worst case scenario, use a blanket policy and enable manual disposition, so that items are reviewed at the end of the retention period and disposed of as needed.


Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.