Theory · 7 min · Triage, severity and the handover
Severity and priority
Set a severity you can defend, from impact and urgency rather than from the alert title.
Severity is how bad it is or could be: the impact if the activity is malicious. Priority is what to work on first, which also depends on urgency (is it still happening?) and on what else is in the queue.
The rule’s default severity is a starting point. The same "suspicious PowerShell" alert is low on a lab VM and high on a domain controller. The entity changes the answer.
Evidence of progress raises severity: credentials used elsewhere, data leaving, more hosts involved, security tools being switched off. Containment that already happened (an EDR block) can lower urgency, but check it really stopped the activity.
When unsure, say what you know and what you do not. "Medium, rising to high if the session token was replayed; checking SigninLogs now" is a useful note.
Your turn: decide on a scenario. Two alerts arrive with the same rule, "Encoded PowerShell command". One is on a developer’s test VM, the other on the domain controller DC01, run by an account that never logs on there. Which should get the higher severity?
Do the interactive check Free, in the browser.
Sources
- Manage incidents in the Microsoft Defender portal: classification Microsoft Learn
- SP 800-61 Rev. 3: Incident Response Recommendations (CSF 2.0 Community Profile), April 2025 NIST