Security

Conditional access on a budget: the five rules worth having

Key takeaways

  • Conditional Access needs Microsoft Entra ID P1, which Microsoft 365 Business Premium customers can also use, and risk-based policies need P2
  • Create two or more cloud-only emergency access accounts with passkeys and exclude them from blocking policies before you switch anything on
  • The five rules are MFA for all users, block legacy authentication, phishing-resistant MFA for admins, protect security info registration and block countries you never work from
  • Start every policy in report-only mode and read the sign-in logs before turning it on
  • Conditional Access runs after the password has been checked, so it limits what a stolen password can do rather than stopping the guess

Conditional Access is the Microsoft Entra ID feature that decides, at each sign-in, whether to let someone in, ask for more proof or block them. If you already pay for Entra ID P1, five policies and an afternoon of careful setup cover the basics for a small business.

What Conditional Access is, and what it costs

Microsoft describes Conditional Access policies, at their simplest, as if-then statements: if a user wants to access a resource, then they must complete an action. The "if" side can include the user or group, the IP address or country, the device and the app. The "then" side is either a block, or access granted on conditions such as multifactor authentication (MFA, a second proof of identity beyond the password), a device marked as compliant or an app protection policy.

The licence is the budget question. Conditional Access requires Microsoft Entra ID P1, and Microsoft says customers with Microsoft 365 Business Premium licences can also use it. Risk-based policies, which react to sign-ins or users that Microsoft scores as risky, need Entra ID Protection, which is a P2 feature. Every rule in this guide works on P1.

If you have no P1 licences, use security defaults instead. These are free, preconfigured settings that require everyone to register for MFA, require administrators to use MFA, block legacy authentication and block device code flow, among other things. Microsoft says security defaults are probably not right for organisations with P1 or P2 licences, because security defaults don't allow the customisation that many organisations require.

One detail worth knowing before you commit: if the licences lapse, Microsoft says existing policies are not automatically disabled or deleted, but you can no longer update them.

Before you switch anything on

Two pieces of groundwork stop Conditional Access locking you out of your own tenant.

Create emergency access accounts. An emergency access account, often called a break-glass account, is an administrator account kept for the day normal admin sign-ins fail. Microsoft recommends two or more, cloud-only and using the .onmicrosoft.com domain, so they do not depend on a synchronised or federated identity. Protect them with a passkey (FIDO2), which Microsoft recommends, or with certificate-based authentication, and use a different method from your everyday admin accounts. Put them in a dedicated group, exclude that group from every policy that blocks or restricts sign-in, alert on every sign-in, and check they still work at least every 90 days.

Use report-only mode. Report-only mode records what a policy would have done without enforcing it. Microsoft's templates create each policy in report-only mode by default, and CISA's baseline for Entra ID says it "recommends initially setting a policy to Report-only when it is created." Leave each new policy in report-only until you have confirmed its effect, then turn it on.

Three smaller points catch people out. Policies created from Microsoft's templates exclude only the person who created them, so you must add the emergency access group yourself. Service principals, the identities that apps and scripts sign in with, are not blocked by policies scoped to users, so none of these rules cover automation. And you have to turn security defaults off to use your own policies. Microsoft says organisations replacing security defaults with Conditional Access must disable security defaults and should then enable their policies immediately. Build all five in report-only first, so the gap is minutes rather than days.

The five rules

1. Require MFA for everyone, on everything

This is the baseline. Microsoft recommends that all organisations create a policy that targets all users and all resources, without app exclusions, and requires MFA. Use the built-in "Multifactor authentication strength". An authentication strength is a named set of sign-in methods that a policy will accept. Exclude the emergency access group and, if you sync from on-premises Active Directory with Microsoft Entra Connect or Cloud Sync, the Directory Synchronization Accounts role.

Microsoft's page for this policy quotes Alex Weinert, its Director of Identity Security: based on Microsoft's studies, an account is more than 99.9% less likely to be compromised if it uses MFA. Microsoft also documents how to skip MFA on a trusted office network. To do so, add a named location and exclude trusted networks from the policy.

2. Block legacy authentication

Legacy authentication covers sign-ins from clients that do not use modern authentication, such as Office 2010, and from older mail protocols such as IMAP, SMTP and POP3. These clients cannot do MFA, so an attacker with a password can use them to walk round rule 1. Microsoft says more than 97 percent of credential stuffing attacks and more than 99 percent of password spray attacks use legacy authentication.

The policy targets all users and all resources, sets the client apps condition to "Exchange ActiveSync clients" and "Other clients", and blocks access. Before you enforce it, filter the sign-in logs by the Client App column, on both the interactive and non-interactive tabs, to find what still uses these protocols. In a small office, check scanners and multifunction printers that send email first.

CISA's Entra ID baseline is blunt about this one: "Legacy authentication SHALL be blocked."

3. Require phishing-resistant MFA for administrators

Phishing-resistant MFA uses methods that a fake sign-in page cannot relay, such as a passkey (FIDO2) or certificate-based authentication. The NCSC's MFA guidance describes attacks on weaker methods, such as one-time password interception and bombarding a user with approval prompts until they approve the attacker's sign-in, and points organisations towards phishing-resistant options. It is also clear that any MFA is better than none: "any implementation of MFA is still superior to reliance on passwords alone."

Target the policy at directory roles rather than named people, and choose the built-in "Phishing-resistant MFA strength". Microsoft recommends requiring it for at least a list of privileged roles. The policy only applies to built-in roles, not custom roles or roles scoped to an administrative unit. CISA's baseline says phishing-resistant MFA "SHALL be required for highly privileged roles."

Register the keys first. Microsoft warns that enabling this policy before administrators have registered suitable methods risks locking you out of the tenant. A Temporary Access Pass, a time-limited credential an administrator issues, lets someone register a passkey without already having one.

4. Protect security info registration

Security info registration is the moment someone adds or changes their MFA methods. If it is unprotected, anyone holding a password for an account that has not yet set up MFA can register their own phone as its second factor. This rule targets the "Register security information" user action. Microsoft's template applies it to all users, excludes guests and the emergency access group, and requires an authentication strength whenever the user is outside a trusted location (an IP range you have marked as your own).

New starters cannot pass MFA before they have any method, so issue each one a Temporary Access Pass; Microsoft says it satisfies Conditional Access requirements for MFA. Temporary Access Pass does not work for guest users, which is why the template excludes them.

5. Block countries you never work from

The location condition lets you control access by network location, and Microsoft says it is commonly used to block countries or regions where an organisation knows its traffic should not come from. Create a named location listing the countries to block, then a policy for all users and all resources that blocks access from it, with the usual emergency access exclusion.

Be honest about the limits. An attacker can rent a server in your own country, so this rule cuts noise rather than stopping a determined attacker. Also remember that Microsoft says Conditional Access is enforced after first-factor authentication is completed, so this rule stops the session after the password check, not the password guessing itself.

What to add when the budget grows

Device compliance. Requiring a compliant device means only machines that meet your Microsoft Intune compliance policy can reach company data. Microsoft warns that without a compliance policy created in Intune, the Conditional Access policy "won't function as intended", so this rule depends on Intune being in place. For organisations not ready to require it for everyone, Microsoft suggests alternatives such as a compliant device, hybrid joined device or MFA for all users. CISA's baseline says managed devices "SHOULD be required for authentication."

If you run Google Workspace. Google's equivalent is Context-Aware Access, which builds policies from IP, device, geographic origin and custom attributes. Google lists it for Frontline Standard and Plus, Enterprise Standard and Plus, Education Standard and Plus, Enterprise Essentials Plus and Cloud Identity Premium.

What to do next week

  1. Monday: create two cloud-only emergency access accounts, register a passkey on each, put them in a dedicated group and set an alert on their sign-ins.
  2. Tuesday: ask every administrator to register a phishing-resistant method, using a Temporary Access Pass where needed.
  3. Wednesday: filter the sign-in logs for legacy authentication and fix or replace what you find.
  4. Thursday: Create the five policies, from templates or custom, in report-only mode and add the emergency access group to each exclusion list.
  5. The following week: review the report-only results, turn off security defaults and switch the five policies on straight away. Put the 90-day emergency access test in the calendar.

Frequently asked questions

Do I need Entra ID P1 for Conditional Access?

Yes. Conditional Access requires Microsoft Entra ID P1, and Microsoft 365 Business Premium customers can also use it. Risk-based policies need P2. Without a licence, security defaults give free baseline protection, including MFA and blocking legacy authentication, but cannot be customised.

Can I use security defaults and Conditional Access together?

No. Microsoft says organisations that replace security defaults with Conditional Access policies must disable security defaults, then enable their policies immediately. Build your policies in report-only mode first so the gap between the two is short.

Should break-glass accounts be excluded from Conditional Access?

Yes, from policies that block or restrict sign-in. Microsoft recommends two or more cloud-only emergency access accounts protected by a passkey (FIDO2) or certificate, excluded through a dedicated group, monitored with alerts and tested at least every 90 days.

More from Future Technology HQ