Skip to main content
The Ops Playbook

Require Phishing-Resistant MFA for Entra ID Admins With Authentication Strengths

Enforce phishing-resistant MFA for Entra ID admins: check registration with Graph, test an authentication strength in report-only mode, then roll back safely.

Require Phishing-Resistant MFA for Entra ID Admins With Authentication Strengths
KBY Technologies11 min readTier L115 min

This playbook covers

Advertising disclosure: this article includes partner links. KBY may earn a commission if you sign up or buy through one, at no extra cost to you. How we make money.

Share

To require phishing-resistant MFA for Microsoft Entra ID administrators, create a Conditional Access policy that includes the privileged directory roles, excludes your emergency access accounts, targets All resources and grants access only with the built-in Phishing-resistant MFA strength. That strength accepts three methods: Windows Hello for Business or platform credential, a FIDO2 security key (passkey), and multifactor certificate-based authentication. The key constraint is order: every admin must have one of those methods registered before the policy is enforced, otherwise Microsoft warns you risk locking yourself out of the tenant. So the safe sequence is register, prove registration, run the policy in report-only mode, read the results, then switch it on with a one-command rollback ready.

#Current method: MFA that an attacker can relay

Most tenants already require MFA for admins, often through the Require multifactor authentication grant control. That control is satisfied by the same combinations as the built-in MFA strength, which includes a password plus a text message, voice call, push notification or one-time passcode. Those methods prove the user has a second factor; they do not bind the sign-in to the site the user is actually on. Microsoft describes the phishing-resistant strength as the methods that require an interaction between the authentication method and the sign-in surface, which is the property a relayed sign-in lacks.

Accounts holding privileged roles are frequent targets, so this gap matters most for them. The friction in the usual approach is operational as much as technical:

  • No registration evidence. Teams switch on a stronger policy without a list showing which admins hold a phishing-resistant method, then discover the gaps through locked-out admins.
  • Enforcement without rehearsal. A policy goes straight to On, so the first evidence of its effect is a support call.
  • Break-glass accounts caught in scope. An emergency access account inside an enforced policy may be unusable during the exact outage it exists for.
  • No recorded way back. When something breaks, the rollback is improvised in the portal under pressure.

#Improved workflow: prove readiness, rehearse, then enforce

The improved workflow treats the policy as a change with evidence at each stage. It has five steps, and each exists for a reason:

  1. Inventory registration. Read each admin’s registered methods from the authentication methods registration report. Input: the directory. Output: a CSV of admins and their methods. This is the gate that prevents lockout.
  2. Close the gaps. Admins without a phishing-resistant method register one, typically by using a Temporary Access Pass to enrol a passkey. Output: an updated CSV with no gaps.
  3. Back up and create in report-only mode. Export the current Conditional Access policies, then create the new policy in report-only mode. Report-only policies are evaluated at sign-in but not enforced, so nobody is blocked while you watch.
  4. Read the results. Use policy impact, the sign-in logs and the What If tool to see which admin sign-ins would succeed and which would need user action.
  5. Enforce with a rollback. Move the policy from report-only to On, and keep the command that sets it back to report-only ready.

Roles and permissions follow least privilege. Creating the policy needs at least the Conditional Access Administrator role. Reading the registration report needs the AuditLog.Read.All permission and a role such as Reports Reader, Security Reader, Security Administrator or Global Reader. Updating a policy through Microsoft Graph needs Policy.Read.All and Policy.ReadWrite.ConditionalAccess. Conditional Access itself needs a Microsoft Entra ID P1 licence.

The trade-off you are accepting is convenience: an admin who signs in with a password and a push notification will be asked for a phishing-resistant method before they can continue, because authentication strengths are evaluated after initial authentication. That friction is the point of the control.

#Implementation

Prerequisites: a P1 licence, at least two cloud-only emergency access accounts that use passkeys or certificate-based authentication, the Microsoft Graph PowerShell SDK, and a validation tenant where you can rehearse first. If you use external authentication methods, stop here: Microsoft states they are currently incompatible with authentication strengths, and you should use the Require multifactor authentication grant control instead.

#
Stage 1: inventory admin registration (read-only)

1Connect-MgGraph -Scopes 'AuditLog.Read.All','Policy.Read.All'
2Get-MgReportAuthenticationMethodUserRegistrationDetail -All |
3  Where-Object { $_.IsAdmin } |
4  Select-Object UserPrincipalName, IsPasswordlessCapable, @{n='MethodsRegistered';e={$_.MethodsRegistered -join ';'}} |
5  Export-Csv -Path .\admin-methods-before.csv -NoTypeInformation

Expected evidence: one row per user whose isAdmin property is true, with the methods registered, for example passKeyDeviceBound. Do not rely on IsPasswordlessCapable alone: it also counts Microsoft Authenticator passwordless sign-in, which satisfies the Passwordless MFA strength but not the Phishing-resistant MFA strength. Read the methods column. Stop condition: any admin who must keep access has no phishing-resistant method.

High angle view of a businessman working at a desk with a laptop, tablet, and documents.
Photo by RDNE Stock project on Pexels

#
Stage 2: close registration gaps

For each gap, an admin registers a passkey (FIDO2) or Windows Hello for Business. Passkeys are available in all Microsoft Entra ID editions, including Free. Where an admin cannot sign in to register, a Privileged Authentication Administrator can issue a Temporary Access Pass; with a one-time pass, the passkey must be registered within 10 minutes of sign-in. The patterns in issuing access passes so users can re-enrol without a reset ticket apply directly here. Rerun Stage 1 and keep the new CSV as evidence.

#
Stage 3: back up policies and create the policy in report-only mode

1Get-MgIdentityConditionalAccessPolicy -All |
2  ConvertTo-Json -Depth 10 |
3  Out-File .\ca-policies-before.json

The -Depth value matters: the default is 2, which would flatten the nested conditions and grant controls you need for a faithful record. Then, in the Microsoft Entra admin center:

  1. Browse to Entra ID > Conditional Access > Policies and select New policy, with a name that follows your naming standard.
  2. Under Assignments, select Users or workload identities. Under Include, select Directory roles and choose at least the roles Microsoft recommends: Global Administrator, Application Administrator, Authentication Administrator, Billing Administrator, Cloud Application Administrator, Conditional Access Administrator, Exchange Administrator, Helpdesk Administrator, Password Administrator, Privileged Authentication Administrator, Privileged Role Administrator, Security Administrator, SharePoint Administrator and User Administrator.
  3. Under Exclude, select Users and groups and choose your emergency access accounts, ideally through a dedicated group.
  4. Under Target resources > Resources > Include, select All resources.
  5. Under Access controls > Grant, select Grant access, then Require authentication strength, choose Phishing-resistant MFA strength and select Select.
  6. Set Enable policy to Report-only and select Create.

Expected evidence: the policy appears in the list in report-only state, and Get-MgIdentityConditionalAccessPolicy returns it with State : enabledForReportingButNotEnforced. Record its Id.

#
Stage 4: read the report-only results

Let real admin sign-ins accumulate. Policy impact shows potential effects over the past 24 hours, 7 days or 1 month, and each sign-in log entry shows the policy result on its Conditional Access and Report-only tabs.

Report-only results for the admin policy and what each means
ResultMeaning for this policyAction
Report-only: SuccessThe sign-in already satisfied the phishing-resistant strength.None.
Report-only: User action requiredThe admin would be asked for a phishing-resistant method once enforced.Confirm the admin has one registered; if not, return to Stage 2.
Report-only: FailureA required non-interactive control was not met.Investigate before enforcing.
Report-only: Not appliedThe user was out of scope, for example an excluded emergency account.Confirm only intended accounts appear here.

Use the What If tool (Conditional Access > Policies > What If) to simulate an emergency access account and an ordinary admin; policies in report-only mode are included in its evaluation. It requires identity, target resource, device platform and client app, and it does not test service dependencies.

#
Stage 5: enforce

1Connect-MgGraph -Scopes 'Policy.Read.All','Policy.ReadWrite.ConditionalAccess'
2Update-MgIdentityConditionalAccessPolicy -ConditionalAccessPolicyId $policyId -State enabled

Run this only when Stage 1 shows no gaps and Stage 4 shows no unexplained results. A successful update returns nothing (Graph answers 204 No Content), so confirm with Get-MgIdentityConditionalAccessPolicy -ConditionalAccessPolicyId $policyId and check that State reads enabled. The admin center equivalent is moving the Enable policy toggle from Report-only to On.

#Guardrails

  • Keep emergency access accounts out of scope. Microsoft recommends at least two cloud-only accounts on the *.onmicrosoft.com domain, protected by passkeys or certificate-based authentication that differ from your normal admin methods, with permanent active Global Administrator assignments in PIM.
  • Know what role targeting does not cover. Conditional Access enforces policies for built-in roles only, not for administrative unit-scoped or custom roles. Cover those admins with a group-based policy.
  • Service principals are not affected. Calls made by service principals are not blocked by user-scoped policies; use Conditional Access for workload identities for those.
  • Do not combine grant controls. Require multifactor authentication and Require authentication strength cannot be used together in the same policy.
  • Change one policy at a time, with the JSON backup taken immediately before.
Wooden tiles spelling 'phishing' highlight cybersecurity themes.
Photo by Markus Winkler on Pexels

#Validation

  1. Sign in as a test admin with a passkey. The sign-in log’s Authentication Details tab shows the authentication strength name in the Requirement column, and the Conditional Access tab shows the policy applied.
  2. Sign in as a test admin with password and push. The admin should be prompted for a phishing-resistant method and blocked if none is registered.
  3. Sign in with each emergency access account. The policy should show as not applied.
  4. Rerun the Stage 1 inventory and confirm no new admin lacks a method.

The pass condition is all four results as stated, recorded with timestamps alongside the two CSVs.

#Common mistakes and how to diagnose them

  • Symptom: an expected method is not offered. Cause: the method is not registered for the user or not enabled in the Authentication methods policy. Diagnosis: check Authentication methods > Policies and the user’s registered methods. Correction: enable the method for the admin group or register it.
  • Symptom: an admin signed in with a password is never offered Windows Hello for Business. Cause: if primary sign-in used another method, the user is not prompted for Windows Hello for Business. Correction: restart the session, choose Sign-in options and pick a method the strength accepts.
  • Symptom: a custom-role admin bypasses the policy. Cause: role targeting ignores custom and administrative unit-scoped roles. Correction: add a group-based policy for them.
  • Symptom: an admin can’t register a passkey during sign-in. Cause: some methods can’t be registered in combined registration at sign-in. Correction: register with a Temporary Access Pass first, as in the passwordless route through new-phone MFA recovery.

#Recovery and rollback

If admins are blocked unexpectedly, sign in with an emergency access account and set the policy back to report-only:

1Update-MgIdentityConditionalAccessPolicy -ConditionalAccessPolicyId $policyId -State enabledForReportingButNotEnforced

Use -State disabled to stop evaluation entirely. Report-only keeps collecting evidence, so it is the better first rollback. Verify recovery by confirming State with Get-MgIdentityConditionalAccessPolicy, then by a successful sign-in from an affected admin. The JSON backup records every policy’s prior configuration if you need to compare. Review every emergency account sign-in afterwards: Microsoft recommends alerting on any use of those accounts and a post-mortem for each. If a user is locked out because no method is registered, the remedy is registration, not an exclusion; see self-service reset routes that cut MFA lockout tickets.

#Measurable outcome

Use measures you can read from your own tenant rather than borrowed figures:

  • Baseline: the count of admins in the Stage 1 CSV without a phishing-resistant method.
  • Success signal: zero admins without one, and the policy enforced with no Report-only: User action required results left unexplained.
  • Measurement method: the registration report, policy impact over 7 days, and sign-in logs filtered to admin accounts.
  • Review cadence: rerun the inventory monthly and whenever a role is assigned; test emergency access accounts at least every 90 days, as Microsoft recommends.
  • Decision threshold: any unexplained block after enforcement returns the policy to report-only until the cause is fixed.

#Phishing-resistant admin MFA readiness checklist

  • P1 licence confirmed, and no external authentication methods in use.
  • Two cloud-only emergency access accounts with passkeys or certificate-based authentication, excluded through a dedicated group.
  • Stage 1 CSV shows every admin with a phishing-resistant method.
  • JSON backup of all Conditional Access policies saved before the change.
  • Policy created in report-only mode, results reviewed for at least one normal working week.
  • What If run for an emergency account and an ordinary admin.
  • Rollback command and policy Id recorded in the change ticket.
  • Next safe decision: after a stable enforced period, consider extending authentication strengths to other sensitive resources.

About this page

Credited to

Dates

Published

Sources

This page lists no external sources.

Corrections

No corrections are listed for this page. Some review sweeps corrected pages without naming each one; those are in the corrections log.

Reader Interaction

Comments

Add a thoughtful note on Require Phishing-Resistant MFA for Entra ID Admins With Authentication Strengths. Comments are checked for spam and held for moderation before appearing.

Loading comments...

Discover more

Was this useful?

Reports go to the site owner and are used to correct pages. Corrections are listed at /corrections.

Operate smarter, with fewer recurring tickets.

Receive new operational playbooks, incident-prevention guidance, automation scripts and recovery runbooks.