The phased rollout
On this page
- Microsoft's phased Conditional Access rollout
- Gate conditions — before deploying any policy
- Phase 1 — Foundation (weeks 1–2)
- Phase 2 — Core authentication (weeks 2–3)
- Phase 3 — Advanced protection (weeks 3–4)
- Report-only guidance
- Evaluating impact
- Testing — the step almost everyone skips
- Rollback options
- Troubleshooting information to collect
Microsoft's phased Conditional Access rollout
Source: Plan your Microsoft Entra Conditional Access deployment
· doc date 2026-06-01 · retrieved 2026-07-28
Microsoft publishes a three-phase, roughly four-week rollout. This is the schedule the
agent should offer when a user asks for a rollout plan, adjusted for their size and change
capability — Microsoft explicitly says timelines can be adjusted.
Gate conditions — before deploying any policy
Microsoft's own "Important" callout:
- Verify emergency access accounts are excluded from all policies
- Test policies with a pilot group before organisation-wide rollout
- Ensure users have registered required authentication methods
- Communicate changes to affected users with supporting documentation
Point 3 is the one that causes the most damage when skipped: enforcing MFA on users who
have never registered a method blocks them rather than challenging them.
Phase 1 — Foundation (weeks 1–2)
Establish baseline controls and prepare for MFA enforcement. **Prerequisite: ensure users
can register for MFA before enabling enforcement policies.**
| Policy | Scenario | Licence |
|---|---|---|
| Block legacy authentication | All users | Entra ID P1 |
| Secure the MFA registration (My Security Info) page | All users | Entra ID P1 |
| Privileged built-in roles enforce phishing-resistant methods | Privileged users | Entra ID P1 |
Note the ordering: phishing-resistant MFA for admins is Phase 1, not a late-stage
hardening step. Microsoft puts it before all-user MFA.
Phase 2 — Core authentication (weeks 2–3)
Enforce MFA for all users and guests; protect mobile with app protection policies.
**Key impact: users will be required to use MFA for all sign-ins and use approved apps with
app protection on mobile.** Communication plan must already be executed and support ready.
| Policy | Scenario | Licence |
|---|---|---|
| All user sign-in activity uses strong authentication methods | All users | Entra ID P1 |
| Guest access is protected by strong authentication methods | Guest access | Entra ID P1 |
| Require approved client apps or app protection policy | Mobile users | Entra ID P1 |
| Require MFA for device join and device registration (user action) | All users | Entra ID P1 |
Phase 3 — Advanced protection (weeks 3–4)
| Policy | Scenario | Licence |
|---|---|---|
| Restrict high-risk sign-ins | All users | Entra ID P2 |
| Restrict access to high-risk users | All users | Entra ID P2 |
| User sign-in activity uses token protection | All users | Entra ID P1 |
| Restrict device code flow | All users | Entra ID P1 |
| Authentication transfer is blocked | All users | Entra ID P1 |
| Conditional Access policies for Privileged Access Workstations (PAW) | Privileged users | Entra ID P1 |
Microsoft's note on PAW: it requires significant infrastructure planning and should only be
implemented after a PAW deployment strategy exists and secure devices are provisioned. The
agent should not casually recommend PAW policies.
Report-only guidance
Enable each policy in report-only mode for at least one week before enforcement.
Review sign-in logs and communicate changes to users before moving to the next phase.
Policies created from Conditional Access policy templates default to report-only.
Evaluating impact
- Use policy impact or report-only mode to confirm settings before and after changes.
- Microsoft's caveat: *a simulated run gives a good idea of how a policy affects sign-in,
but it doesn't replace an actual test run in a properly configured development
environment.* The agent should repeat this rather than overselling What If.
Testing — the step almost everyone skips
Ensure you test the exclusion criteria of a policy.
Microsoft's reasoning: you might exclude a user or group from a policy requiring MFA, but
the combination of other policies can still require MFA for those users. So test that
excluded users actually experience the exclusion. An exclusion in one policy is not an
exemption from the estate.
This directly applies to break-glass validation: excluding the account from Policy A proves
nothing if Policy B still catches it.
Rollback options
Microsoft lists three, in order:
- Disable the policy. It stops applying at sign-in; re-enable when ready.
- Exclude a user or group. With an explicit caution: *use exclusions sparingly, only
where the user is trusted. Add users back to the policy or group as soon as possible.*
- Delete it if disabled and no longer needed.
Soft delete — a genuinely useful recovery fact
If a Conditional Access policy or named location is deleted, it can be restored
within a 30-day soft-delete period.
Restoring requires the Conditional Access Administrator role. The agent should offer
this whenever someone reports an accidentally deleted policy — most administrators do not
know it exists.
Troubleshooting information to collect
Microsoft's list for a user reporting a CA problem:
- User principal name
- User display name
- Operating system name
- Time stamp (approximate is fine)
- Target application
- Client application type (browser or client)
- Correlation ID (unique to the sign-in)
If the user sees a More details link on the error page, they can collect most of this
themselves. Including request ID, time and date is what lets Microsoft support locate the
event if a case is opened.