Runbook: Conditional Access Policy Rollback
Restore access safely with temporary exclusions and staged re-enforcement.
⚠️ Business Consequence: Why Emergency Rollback Matters
- Financial Impact: Org-wide authentication lockout = $20K–$100K per hour (entire workforce unable to work)
- Compliance Exposure: Emergency bypass of change control = audit findings, policy violations
- Operational Risk: Break-glass account failure = no recovery path, complete system lockout
- Reputation Impact: Executive lockout during board meetings = business credibility damage
Average rollback time: 5–15 minutes emergency disable — prevents P1 escalation.
🚀 Before You Start
15-30 minutes (emergency restore) or 45-60 min (full re-enable)
Entra ID Conditional Access admin with break-glass access
Creates temporary security gap during emergency rollback. Mitigated by monitoring + re-enable.
Break-glass account, Azure portal, Conditional Access Administrator role
⚠️ Break-glass account unavailable? Talk to an Exchange Security Specialist immediately for P1 incident escalation.
⚠️ Runbook Summary
- Severity: P1 - Critical lockout scenario
- Total time: 15-30 minutes (emergency restore), 45-60 min (complete phased re-enable)
- Risk level: Medium (temporary security gap during rollback)
- Requires: Break-glass account access, Azure AD admin role
Pre-Rollback Checklist (5 minutes)
Complete these validation steps before proceeding with rollback:
- Confirm root cause: Verify lockout is caused by specific CA policy via Azure AD Sign-In Logs (error code 53003 or 53000)
- Break-glass validation: Ensure break-glass account can still authenticate and has Conditional Access Administrator role
- Identify affected policy: Note policy name, GUID, and exact conditions causing lockout
- Document current state: Export policy configuration before making changes (for audit trail)
- Assess impact: Determine number of affected users and criticality (all users vs. subset)
3-Phase Rollback Procedure
Choose the appropriate phase based on urgency and risk tolerance:
Phase 1: Emergency Disable (5-10 min) - Use for Complete Lockout
- Access Azure AD: Sign in with break-glass account → Azure Active Directory → Security → Conditional Access
- Locate policy: Find the policy causing lockout (match name/GUID from sign-in logs)
- Disable policy: Set "Enable policy" toggle to Off
- Document change: Add note to policy description: "Disabled [date] [time] due to lockout incident [incident-ID]"
- Test immediately: Have 2-3 affected users attempt sign-in within 2 minutes
⚠️ Risk: All protections from this policy are now disabled. Proceed to Phase 2 immediately.
Phase 2: Add Emergency Exclusions (10-15 min) - Use for Partial Lockout
- Create exclusion group: Azure AD → Groups → New group → "CA-Emergency-Exclusion-[date]"
- Add affected users: Add locked-out users to exclusion group (or affected security group)
- Modify policy: Edit CA policy → Conditions → Users → Exclude → Select emergency group
- Set expiration: Add calendar reminder to remove exclusion within 48 hours
- Validate: Test with 2 pilot users before announcing
✓ Benefit: Policy remains enforced for most users; only affected users excluded.
Phase 3: Staged Re-Enablement (30-45 min) - After Root Cause Fixed
- Adjust policy conditions: Modify grant controls or conditions that caused lockout (e.g., change "Require compliant device" to "Require MFA only")
- Create pilot group: "CA-Policy-Pilot" with 5-10 test users
- Enable for pilot: Edit policy → Users → Include pilot group → Enable policy
- Monitor sign-ins: Watch Azure AD Sign-In Logs for 15 minutes; verify no failures
- Expand gradually: Add groups incrementally (IT dept → Finance → All users) over 24-48 hours
- Remove exclusions: Once fully rolled out, delete emergency exclusion group
Validation & Success Criteria
Confirm rollback success using these checkpoints:
Immediate Validation (Within 5 min of rollback)
- User sign-in success: At least 3 affected users can sign in successfully to Outlook, OWA, and Teams
- No MFA loops: Users not prompted repeatedly for authentication
- Device access restored: Both compliant and non-compliant devices can authenticate (if policy was disabled)
Azure AD Sign-In Log Verification
- Status: Sign-ins show "Success" (green checkmark) instead of "Failure" (red X)
- Conditional Access: Shows "Not Applied" (if disabled) or "Success" (if exclusion used)
- Error codes cleared: No more 53003 (policy blocked) or 53000 (device not compliant) errors
Service Health Monitoring (24-48 hours)
- Support ticket volume: No new lockout complaints after rollback
- Policy compliance: Security team reviews adjusted policy for compliance gaps
- Exclusion cleanup: Emergency exclusions removed after permanent fix validated