Skip to main content
The Ops Playbook

Retire SMS and Voice MFA in Entra ID Without Locking Users Out

Disabling SMS MFA in Entra ID? Find phone-dependent users with Graph, run an Authenticator registration campaign, then switch SMS and voice off with a rollback.

Retire SMS and Voice MFA in Entra ID Without Locking Users Out
KBY Technologies12 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 retire SMS and voice call MFA in Microsoft Entra ID, turn the SMS and Voice call methods off in the Authentication methods policy, but only after the people who rely on them have registered Microsoft Authenticator. The order is the whole job: find the users whose strongest registered method is a phone, nudge them to Authenticator with a registration campaign, back up the SMS and voice method configuration, exclude a pilot group first, then set each method’s state to disabled with the backup ready to restore. One constraint catches tenants out: a user enabled for a method in any policy can still use it, so the change only fully takes effect once the tenant’s authentication methods migration is set to Migration Complete.

#Current method: phone factors nobody owns

In many tenants SMS and voice were first enabled through the legacy MFA and self-service password reset (SSPR) settings, then carried into the Authentication methods policy, because Microsoft’s migration guide recommends enabling any method that either legacy policy enabled. Microsoft now recommends that users move away from text messages and voice calls for MFA and names Microsoft Authenticator as the alternative. Its documentation also shows why phone factors generate tickets: Microsoft doesn’t guarantee that prompts come from the same number, may add or remove short codes as it adjusts routing, can throttle repeated attempts with messages such as You’ve hit our limit on text verification codes, and lists blocked caller ID, voicemail forwarding and poor signal among the known failure causes.

The usual attempt at fixing this is a single switch: someone disables SMS in the portal, and the next morning the service desk takes calls from users whose only registered method was their phone number. The problem is not the switch: nobody measured who depended on it, offered a replacement or kept a record of the previous policy. If you already run self-service resets, the same population will show up there too, which is why self-service MFA reset for lockout tickets belongs in the same review.

#Improved workflow: measure, migrate users, then switch off in stages

Each step changes one control and produces evidence.

  1. Inventory (read-only). Use the authentication methods registration report to list users whose system-preferred method is a phone method. That list is the population at risk, and its size is your baseline.
  2. Give them a replacement. Make sure Microsoft Authenticator is enabled for those users, then run an Authenticator registration campaign so they are prompted to set it up during normal MFA sign-in.
  3. Back up and check scope. Export the SMS and voice method configurations to files and confirm the migration state, because legacy settings are still respected until migration is complete.
  4. Pilot by exclusion. Add a small pilot group to the SMS and voice excludeTargets. Exclusion is a narrow, reversible change that touches only that group.
  5. Switch off. When the pilot is clean and the inventory has shrunk to an agreed exception list, set each method’s state to disabled, or limit it to a named exception group.

The trade-off is time: a campaign with snoozes takes days or weeks to move a population, and the registration report can lag by up to 36 hours. In exchange you never remove a method from someone who has nothing else.

#Implementation

You need Microsoft Graph PowerShell, at least the Authentication Policy Administrator role to change the policy, and a reporting role such as Reports Reader, Security Reader or Global Reader to read registration details. The Activity dashboard in the admin center requires a Microsoft Entra ID P1 or P2 licence. Rehearse in a test tenant or with test accounts first.

#
Stage 1: find phone-dependent users (read-only)

The userRegistrationDetails report exposes systemPreferredAuthenticationMethods, the most secure methods among those a user has registered, with possible values including sms, voiceMobile, voiceAlternateMobile and voiceOffice. If a user’s system-preferred method is one of those, the phone is the best factor they have.

1Connect-MgGraph -Scopes "AuditLog.Read.All", "Policy.Read.AuthenticationMethod"
2$phone = 'sms', 'voiceMobile', 'voiceAlternateMobile', 'voiceOffice'
3Get-MgReportAuthenticationMethodUserRegistrationDetail -All |
4  Where-Object { @($_.SystemPreferredAuthenticationMethods | Where-Object { $_ -in $phone }).Count -gt 0 } |
5  Select-Object UserPrincipalName, IsAdmin, IsSsprCapable,
6    @{n='SystemPreferred';e={$_.SystemPreferredAuthenticationMethods -join ';'}},
7    @{n='UserPreferred';e={$_.UserPreferredMethodForSecondaryAuthentication}},
8    @{n='MethodsRegistered';e={$_.MethodsRegistered -join ';'}} |
9  Export-Csv -Path .\phone-dependent-before.csv -NoTypeInformation

Expected evidence: a CSV with one row per phone-dependent user. Record the row count and the number of rows where IsAdmin is True. In the admin center, Entra ID > Authentication methods > Activity shows the same picture: the Registration tab lists users registered by method, and the Usage tab shows sign-ins by authentication method. Note the SMS and voice sign-in counts as your second baseline.

Stop condition: if any administrator appears in the CSV, resolve those accounts by hand before going further, and make sure every account you would use to roll back has a non-phone method.

The report does not include disabled or soft-deleted users, so an account re-enabled later is not covered by this snapshot.

Stylish adult man using his smartphone for voice commands in an outdoor urban setting.
Photo by Theo Decker on Pexels

#
Stage 2: enable Authenticator and run a registration campaign

  1. In Entra ID > Authentication methods > Policies, confirm Microsoft Authenticator is enabled for the target users with Authentication mode set to Any or Push. With Passwordless, users aren’t eligible for the nudge.
  2. Put the users from the CSV in a security group, for example MFA-Phone-Dependent.
  3. Browse to Entra ID > Authentication methods > Registration campaign and select Edit.
  4. Set State to Enabled, Authentication method to Microsoft Authenticator, choose Days allowed to snooze (0 to 14) and decide on Limited number of snoozes. When it is enabled, users can skip three times before they must register.
  5. Include the group and select Save.

Users are prompted after they complete MFA during an interactive sign-in, and only if they don’t already have Authenticator push set up. Conditional Access policies that govern security information registration apply first, so a policy restricting registration to the internal network suppresses the prompt elsewhere. Microsoft managed is the alternative state: for Authenticator it targets users who perform MFA by SMS or voice, allows one day of snooze and makes registration required after three snoozes, but you can’t change those settings.

Rollback: set the campaign State to Disabled to stop the prompts.

#
Stage 3: back up the method configuration and confirm the migration state

1Connect-MgGraph -Scopes "Policy.ReadWrite.AuthenticationMethod"
2foreach ($id in 'sms', 'voice') {
3  Get-MgPolicyAuthenticationMethodPolicyAuthenticationMethodConfiguration -AuthenticationMethodConfigurationId $id |
4    ConvertTo-Json -Depth 10 | Out-File -FilePath ".\$id-config-before.json" -NoClobber
5}

Expected evidence: two files showing state, includeTargets (for SMS, each target’s isUsableForSignIn), excludeTargets and, for voice, isOfficePhoneAllowed. ConvertTo-Json defaults to a depth of 2, which can cut off the nested target details, so keep -Depth 10. Also note what the Enable and Target and Configure tabs show for each method.

Next, open the Authentication methods policy and select Manage migration. Microsoft documents three states:

Authentication methods migration states and which settings apply
StateAuthentication methods policyLegacy MFA and SSPR settings
Pre-migrationUsed only for authenticationRespected
Migration in ProgressUsed for authentication and SSPRRespected
Migration CompleteUsed for authentication and SSPRIgnored

Until the state is Migration Complete, a user enabled for Text message to phone or Call to phone in the legacy MFA policy, or for Mobile phone or Office phone in the legacy SSPR policy, can still register and use SMS or voice whatever the Authentication methods policy says. Before you move to Migration Complete, check that every method you still need is enabled in the Authentication methods policy. Security questions stay in the legacy SSPR policy and keep working. You can move back to Migration in Progress at any time.

#
Stage 4: pilot by exclusion

Create a pilot group of users who appear in the CSV and have since registered Authenticator, plus IT staff. Run the change with -WhatIf first, then without it. The @odata.type line is required in the body.

1$params = @{
2  "@odata.type"  = "#microsoft.graph.smsAuthenticationMethodConfiguration"
3  excludeTargets = @(
4    @{ id = $pilotGroupId; targetType = "group" }
5  )
6}
7Update-MgPolicyAuthenticationMethodPolicyAuthenticationMethodConfiguration -AuthenticationMethodConfigurationId sms -BodyParameter $params

Repeat for voice with "#microsoft.graph.voiceAuthenticationMethodConfiguration" and -AuthenticationMethodConfigurationId voice. If the backup already lists exclusions, include them in the array as well, so the request sets the full list you intend. A successful update returns no output (Graph returns 204 No Content); rerun the Stage 3 Get to confirm the pilot group is listed in excludeTargets.

Stop condition: if a pilot user cannot complete MFA or SSPR, roll the exclusion back before investigating.

#
Stage 5: switch the methods off

When the rerun inventory shows only users on your agreed exception list, set the state to disabled:

1$params = @{
2  "@odata.type" = "#microsoft.graph.smsAuthenticationMethodConfiguration"
3  state         = "disabled"
4}
5Update-MgPolicyAuthenticationMethodPolicyAuthenticationMethodConfiguration -AuthenticationMethodConfigurationId sms -BodyParameter $params

Repeat for voice. If some users genuinely need a phone factor, keep the method enabled and change its target on the Enable and Target tab from all users to a single exception group instead. Microsoft gives this pattern as an example: Authenticator for all users, SMS and voice limited to one small group.

#Guardrails

  • Change one method and one scope at a time, and keep the JSON backups next to the change record.
  • From Migration in Progress onwards the Authentication methods policy governs SSPR as well as sign-in, so disabling a method there also removes it from password reset. Watch IsSsprCapable as well as MFA.
  • Consolidate targets into one group per method. Microsoft warns that registration can fail when many groups are included, and that the policy may not save once it exceeds 20 KB.
  • Users who lose their phone during the campaign still need a recovery route; Temporary Access Passes for lost-phone MFA resets covers one.
  • Authentication Policy Administrator is the least privileged role for the method policy and campaign changes here; don’t make them as Global Administrator out of habit. Creating the groups needs a groups management role.
Senior man sitting indoors using smartphone and laptop, smiling with technology.
Photo by Helena Lopes on Pexels

#Validation

  1. Rerun the Stage 3 Get for both methods. Pass: state and excludeTargets match the intended change.
  2. Sign in as a pilot user. Pass: MFA completes with Authenticator and SMS or voice is not offered.
  3. Run SSPR as a pilot user. Pass: the reset completes with the remaining methods.
  4. Once the report refreshes (allow up to 36 hours), rerun the Stage 1 inventory. Pass: no pilot user appears, and the overall count is lower than the baseline.
  5. Check Activity > Usage. Pass: SMS and voice sign-ins come only from users outside the pilot or on the exception list.

#Common mistakes and how to diagnose them

  • Users still receive texts after SMS is off. Likely cause: migration isn’t complete and a legacy setting such as Mobile phone for SSPR or Text message to phone for MFA still enables them. Check Manage migration. Microsoft gives the same example for voice: users outside the targeted group can still sign in with a voice call because a legacy policy enables it.
  • Nobody is prompted by the campaign. Check that Authenticator’s mode isn’t Passwordless, that users are in scope of the Authenticator method, that they sign in interactively with MFA, and that a registration Conditional Access policy isn’t suppressing the prompt.
  • A user drops off the inventory but still can’t sign in. The report can lag by up to 36 hours and omits disabled users. Check the user’s methods in the admin center, where usable methods are listed before non-usable ones.
  • The policy won’t save. Too many targeted groups; consolidate them into one group per method.
  • Cross-tenant synchronization skips users. SMS Use for sign-in is enabled by default, and users enabled for SMS sign-in are skipped by cross-tenant synchronization. Turn it off for those users if you keep SMS for an exception group.

#Recovery and rollback

  1. For a pilot problem, re-send the excludeTargets list recorded in the backup file, or remove the pilot group from the method’s exclusions in the admin center.
  2. For a problem after Stage 5, send the same body with state = "enabled" for the affected method.
  3. If the migration state was changed, move it back to Migration in Progress so legacy settings apply again.
  4. Disable the registration campaign if the prompts themselves are causing trouble.
  5. Verify recovery by rerunning the Get and comparing it with the backup file, then by a successful sign-in and SSPR test from an affected user.

Users who changed phones mid-rollout may need the route in passwordless new-phone MFA recovery rather than a rollback.

#Measurable outcome

Use two baselines from Stage 1: the number of phone-dependent users in the CSV, and SMS and voice sign-ins on the Usage tab. The success signal is the inventory falling to the agreed exception list, and SMS and voice sign-ins coming only from that list. Review weekly while the campaign runs, allowing for the 36-hour reporting delay. The decision rule for each stage is simple: expand the exclusion, or switch a method off, only when the previous stage’s group shows no phone-dependent users and no lost SSPR capability. Track MFA-related tickets alongside these numbers so you can tell whether the change reduced work or moved it.

#SMS and voice MFA retirement checklist

  • Phone-dependent inventory exported, counted and checked for administrators.
  • Authenticator enabled in Any or Push mode for the target users.
  • Registration campaign enabled for the phone-dependent group, with snooze settings recorded.
  • SMS and voice configurations backed up to JSON.
  • Migration state confirmed; legacy phone settings no longer apply.
  • Pilot exclusion applied, validated for MFA and SSPR, and rollback rehearsed.
  • Methods disabled or limited to one exception group; exception owners named.
  • Next decision recorded: review the exception group at each weekly check and remove members once they register Authenticator.

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 Retire SMS and Voice MFA in Entra ID Without Locking Users Out. 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.