Understand data protection and governance tasks for Microsoft 365 and Copilot
Information Protection, Classification, and Lifecycle
Classify and persistently protect sensitive content, then retain or delete it according to governance requirements.
What you need to know
- Sensitivity labels classify content and can apply encryption, access restrictions, or visual markings; the label persists in supported content metadata.
- Sensitive information types use patterns; trainable classifiers use examples and context for less structured content.
- Retention policies and retention labels govern how long content is kept and what happens afterward. Retention is distinct from sensitivity.
Objects and administrative surfaces
- Sensitivity labels, label policies, classifiers, Data Explorer, Activity explorer, and retention — Microsoft Purview portal.
- Files, emails, meetings, groups, and sites support different labeling scopes and settings.
How it works
- Labels must be published through a label policy before targeted users can apply them.
- A document can have both a sensitivity label and a retention label because protection and lifecycle solve different problems.
When to use it
- Use a sensitivity label for classification and persistent protection; use retention for keep/delete requirements.
- Use a sensitive information type for patterned values and a trainable classifier for contextual document categories.
Security and governance implications
- Limit Data Explorer content access because viewer roles can expose sensitive item content.
- Test automatic labeling and retention before broad rollout.
Troubleshooting signals
- Confirm label publication scope, client support, permissions, and replication time.
- Confirm whether the requirement concerns access protection or lifecycle before changing the policy.
Exam traps
- A retention label does not encrypt a file.
- A site sensitivity label can configure the container without automatically labeling each file inside it.
Key takeaways
- Sensitivity protects; retention governs lifecycle; classification identifies.
- Patterned data → sensitive information type; contextual category → trainable classifier.
Related objectives
- D2.1.a
- D2.1.b
- D2.1.c