UK’s trusted IT infrastructure partner since 2003
Servnet
FinanceToolsConfiguratorGet in Touch
Cyber Security

Microsoft Entra ID Conditional Access: Best Practices

Servnet Editorial · IT infrastructure analysis8 min read
Share

Enabling multifactor authentication across a corporate tenant is only the foundation of identity defence; without context-aware policy enforcement, attackers routinely exploit legacy protocol fallbacks and unmanaged endpoints. Designing an enterprise Conditional Access baseline requires balancing security posture against operational disruption and licence economics across Microsoft Entra ID tiers. This guide outlines an operational starter architecture—contrasting native Entra capabilities with alternatives like Okta and Ping Identity, detailing licence dependencies, and mapping technical controls against UK standards like Cyber Essentials Plus via structured identity and access management.

Baseline Conditional Access Policy Blueprint
Policy FocusTarget ScopeGrant EnforcementLegacy Auth BlockLegacy client appsAll users & cloud appsBlock accessDevice ComplianceDesktop & mobile appsAll users,corporate appsRequirecompliant deviceAdmin ProtectionDirectory admin rolesAll cloud applicationsRequire auth strengthHigh-Risk Sign-InHigh-risk detectionsAll users,all resourcesMFA andreauthenticationApproved Mobile AppMobile client platformsAll users,mobile accessRequire approved app
View the data behind this chart
Baseline Conditional Access Policy Blueprint
Policy FocusTarget ScopeGrant Enforcement
Legacy Auth BlockLegacy client appsAll users & cloud appsBlock access
Device ComplianceDesktop & mobile appsAll users, corporate appsRequire compliant device
Admin ProtectionDirectory admin rolesAll cloud applicationsRequire auth strength
High-Risk Sign-InHigh-risk detectionsAll users, all resourcesMFA and reauthentication
Approved Mobile AppMobile client platformsAll users, mobile accessRequire approved app

The MFA Baseline Fallacy: Why Default Configurations Fail

Many IT organisations operate under the assumption that completing a tenant-wide multifactor authentication enrolment campaign resolves their primary identity security risks. Multifactor authentication is an essential verification mechanism, but inside Microsoft Entra ID it functions merely as one grant control within a wider policy evaluation engine. While equivalent contextual access engines exist in platforms such as Okta (Authentication Policies) and Ping Identity, this guide focuses on Entra ID due to its deep native integration with Windows, Microsoft 365, and Azure infrastructure. Without structured Conditional Access rules governing under what conditions, from which platforms, and through which client applications authentication occurs, the identity perimeter remains porous.

The most critical operational exposure stems from legacy authentication protocols. Older connection mechanisms and legacy mail clients—including IMAP, POP3, and older ActiveSync configurations—bypass interactive authentication challenges entirely. In UK enterprise assessments, legacy protocol exploitation remains one of the primary vectors for business email compromise (BEC). Blocking legacy authentication across all users and cloud applications is mandatory for compliance with technical verification standards such as Cyber Essentials Plus and NCSC cloud security principles.

A secondary operational failure occurs when teams mistake policy creation for active enforcement. Relying on default tenant settings or prematurely assuming Microsoft-managed baseline policies provide full protection is a dangerous trap. Policies lingering in Report-only mode log theoretical outcomes without denying access or enforcing MFA. Transitioning from passive observation to an active, resilient zero-trust baseline requires rigorous sign-in log analysis and a staged enforcement roadmap.

  • •Legacy protocols generally cannot complete interactive MFA challenges; identify and then block legacy authentication where it is present, using a staged policy.
  • •Microsoft-managed policies are created in Report-only mode; for policies you create, choose Report-only initially where appropriate and verify the state before relying on enforcement.
  • •Authentication must be evaluated contextually across user state, client application, and device posture rather than treated as a static one-time prompt.
Illustration: Microsoft Entra ID Conditional Access: Best Practices

Policy Architecture: Scoping a Practical Starter Baseline

Start with a small set of separately scoped policies, grouping applications and users that share the same access requirements, then add controls as justified by risk and operational need. Avoid unnecessarily monolithic policies: group users, resources and conditions by clear access objectives so that policy behaviour remains understandable and testable. Many uncoordinated, narrowly scoped policies can make evaluation and troubleshooting harder; document scopes and review sign-in logs.

A disciplined architecture groups policies logically by persona, risk tier, and target resource rather than attempting an unmanageable catch-all rule. In practice, production failures occur when organisations conflate protocol deprecation with user-access restrictions. By isolating legacy client applications into an independent, tenant-wide block rule with explicit exclusions only for emergency recovery accounts, security teams can sever insecure authentication vectors cleanly without interfering with day-to-day workforce workflows.

The foundational starter set should then branch into policies governing standard user access, elevated administrator controls, compliant device enforcement, and automated response to anomalous sign-ins. For organisations tracking compliance against frameworks such as Cyber Essentials or internal governance audits, establishing explicit boundaries between baseline authentication and privileged administrative access is essential. Each policy must have an unambiguous objective, clear scoping parameters, and measurable grant criteria.

  • •Target legacy client applications with a dedicated, tenant-wide block rule.
  • •Separate standard workforce baseline controls from privileged administrative requirements.
  • •Establish discrete policies for device compliance, application management, and high-risk sign-in conditions.

Device Health and Application Posture Controls

Restricting access based on device health constitutes the second layer of a robust identity architecture. Even when user credentials and MFA factors are verified, permitting unmanaged or non-compliant endpoints to access sensitive corporate repositories introduces severe data-exfiltration and malware risks. Choose the appropriate combination of MFA, compliance and hybrid-join controls for the target population; do not require hybrid join for users or devices that cannot meet that condition.

Organisations must decide whether to require all selected controls or allow access if any single condition is met. In environments where endpoints are managed via Microsoft Intune, device compliance serves as an indispensable gatekeeper against unmanaged device sprawl. For UK organisations pursuing Cyber Essentials Plus, demonstrating that access to corporate data is restricted to managed, compliant endpoints is a core audit hurdle. For estates transitioning from on-premises Active Directory, permitting hybrid Entra joined devices provides a secure operational bridge while full Intune enrolment matures.

Mobile access and unmanaged BYOD scenarios demand nuanced controls rather than blunt network blocks. Leveraging app-protection policies alongside approved client application requirements allows staff to access corporate email and collaborative data securely on personal iOS and Android devices. This mobile application management (MAM) pattern prevents local data exfiltration and unmanaged copy-pasting into personal apps, satisfying corporate data-loss prevention standards without the friction and privacy concerns of full MDM enrolment.

Phishing-Resistant Authentication Strengths for Admins

Privileged administrators represent the highest-value target within any tenant, necessitating controls that strictly prohibit easily intercepted MFA methods such as SMS or voice OTP. Production experience indicates that credential replay and adversary-in-the-middle (AiTM) phishing attacks routinely circumvent basic push notifications. Enforcing dedicated authentication strengths—mandating FIDO2 security keys, certificate-based authentication, or Windows Hello for Business—ensures administrative sessions are hardened against modern phishing infrastructure.

Consider a dedicated policy for privileged directory roles and sensitive resources, selecting scope and applications appropriate to the tenant. Select an authentication strength that excludes weaker methods when phishing-resistant authentication is required, and scope it to the relevant administrator roles and resources. This structured approach directly complements broader initiatives across a Zero Trust framework.

Furthermore, administrative access must be tied directly to dynamic threat intelligence. Sign-in risk policies evaluate anomalous characteristics in real time—such as unfamiliar sign-in properties, atypical travel, or leaked credentials—automatically triggering reauthentication or session revocation. When combined with strict session lifetime controls and continuous access evaluation (CAE), risk-driven policies drastically shrink the dwell time of compromised administrative sessions.

Emergency Access Runbook: Hardening Break-Glass Accounts

The most critical vulnerability introduced during aggressive Conditional Access enforcement is the risk of total administrative lockout. A misconfigured enforced policy can cause lockout; maintain independently protected, tested emergency accounts and exclude them from relevant enforced restrictive policies. Microsoft recommends two or more emergency-access accounts to reduce the risk of accidental administrator lockout.

Enterprise resilience requires establishing two cloud-only break-glass accounts permanently assigned the Global Administrator role, provisioned with complex, split passwords stored in a secure physical safe or enterprise vault. These accounts must never be assigned to named individuals, must bypass federated identity providers, and must be placed in a dedicated security group that is excluded from all blocking Conditional Access policies. A common failure in UK tenant audits is leaving break-glass accounts protected by third-party MFA providers: if that external provider experiences an outage, administrators become permanently locked out.

Operational governance over these emergency credentials must be absolute. Security teams must configure automated monitoring—via Microsoft Sentinel or standard Entra audit alert rules—to trigger immediate high-priority alerts whenever an emergency account signs in. Furthermore, testing these credentials under a structured change window at least quarterly confirms that credentials remain functional, exclusion groups remain intact, and alerting pipelines are operating as intended.

  • •Maintain two or more cloud-only emergency accounts permanently assigned the Global Administrator role.
  • •Place the emergency accounts in a dedicated group and exclude that group from enforced Conditional Access policies that block or restrict sign-in.
  • •Conduct documented sign-in verification tests at least quarterly to validate account operability.
Phased Policy Rollout Lifecycle
W0W2W4W6W7What If Impact Analysis1wSuggested two-weekobservation2wPilot User Testing1wPhased Production Rollout2wTotal: 7 weeks end-to-end
View the data behind this chart
Phased Policy Rollout Lifecycle
PhaseStarts (week)Duration (weeks)
What If Impact Analysis11
Suggested two-week observation22
Pilot User Testing41
Phased Production Rollout52

Rollout Protocol: Report-Only Validation and Impact Analysis

Enforcing Conditional Access policies across an active production estate without rigorous telemetry analysis almost guarantees unexpected outages and service desk surges. Running newly authored policies in Report-only mode for a minimum of one to two weeks provides the empirical data needed to uncover shadow IT, unmanaged mobile devices, and legacy automated scripts before enforcement begins.

Before flipping policies to active enforcement, administrators should run impact simulations using the Conditional Access What If tool. By testing edge cases—such as external auditors accessing SharePoint from non-standard operating systems or executives travelling abroad—teams can predict policy evaluation paths and identify unintentional blocks without exposing live users to access friction.

A structured production cutover follows a phased cadence: evaluate initial telemetry in Report-only mode, deploy to a designated IT pilot user group, and only then roll out tenant-wide with clear end-user communications. Documenting expected authentication behaviours reduces helpdesk ticket volume and avoids the common pitfall of emergency policy rollbacks.

Operational Pitfalls: Policy Collisions, B2B Guests, and Exceptions

As an identity estate matures, administrators frequently encounter unexpected access denials caused by policy interactions. Review all applicable policies: a block control denies access, while grant and session requirements from applicable policies can collectively affect the resulting access decision. Review the sign-in logs and policy details rather than inferring the result from policy names alone.

A recurring operational fault occurs when device compliance policies are applied indiscriminately across all users, inadvertently capturing B2B guest accounts and external contractors. Test guest and external-user scenarios separately, including the tenant’s cross-tenant access, authentication-trust and device-compliance configuration. Use appropriate guest policies, authentication strength and cross-tenant access settings rather than assuming resource-tenant device compliance will or will not work.

Finally, service accounts and automated workloads must be catalogued and managed intentionally. While human identities transition toward interactive MFA and phishing-resistant passkeys, non-interactive service principals and automation accounts must not be left unmonitored. Inventory non-interactive identities separately. Apply Conditional Access for workload identities to service principals where supported, replace script credentials with managed identities where possible, and handle any remaining legacy accounts through a documented, narrowly scoped exception with compensating controls.

Licensing Architecture: Entra ID P1 vs P2 Control Mapping

A frequent barrier in enterprise rollouts is budgeting for the specific licence tiers required by each Conditional Access control. Microsoft Entra ID P1 (included in Microsoft 365 Business Premium at ~£18.10/user/month and M365 E3 at ~£33.80/user/month, or £4.90/user/month standalone) covers the operational baseline: blocking legacy authentication, trusted IP location fencing, and requiring device compliance via Microsoft Intune.

By contrast, advanced dynamic capabilities require Microsoft Entra ID P2 (included in Microsoft 365 E5 at ~£49.60/user/month or £7.40/user/month standalone). P2 is mandatory for real-time user risk and sign-in risk policies, automated self-remediation, and custom authentication strengths. Workload identity protection requires a separate add-on (~£2.50/workload identity/month) under Microsoft 365 licensing. Aligning control design with these licence tiers prevents unexpected procurement delays during deployment.

  • •Entra ID P1 (or Business Premium / E3): Legacy authentication blocks, trusted locations, Intune device compliance, and approved MAM client apps.
  • •Entra ID P2 (or E5 / E5 Security): User-risk and sign-in-risk policies, automated self-remediation flows, and custom phishing-resistant authentication strengths.
  • •Workload Identity Premium: Conditional Access evaluation, anomaly detection, and access controls for service principals and machine accounts.

Sources

Every figure in this article traces to the sources below.

  • •Microsoft Learn — Block legacy authentication
  • •Microsoft Learn — Plan a Conditional Access deployment
  • •Microsoft Learn — Manage emergency-access accounts
  • •Microsoft Learn — Require compliant or hybrid joined devices
  • •Microsoft Learn — Protect data with Conditional Access and Intune
  • •Microsoft Learn — Microsoft-managed Conditional Access policies
Conditional Access Defence Layers
4Emergency Access ResiliencyTwo cloud-only Global Admins in dedicated group, tested quarterly3Privileged Role ControlsPhishing-resistant authentication strengths on administrator roles2Device & Application PostureRequire compliant or hybrid-joined devices and approved client apps1Perimeter & Protocol HygieneBlock legacy client protocols across all users and cloud apps
View the data behind this chart
Conditional Access Defence Layers
LayerDetail
Emergency Access ResiliencyTwo cloud-only Global Admins in dedicated group, tested quarterly
Privileged Role ControlsPhishing-resistant authentication strengths on administrator roles
Device & Application PostureRequire compliant or hybrid-joined devices and approved client apps
Perimeter & Protocol HygieneBlock legacy client protocols across all users and cloud apps
Share
Key takeaways
  • ✓Licensing governs architecture: Baseline legacy auth blocks and Intune device compliance run on Entra ID P1 / Business Premium, whereas real-time risk policies and authentication strengths require Entra ID P2.
  • ✓Cyber Essentials Plus alignment: UK organisations must enforce MFA on all cloud services and restrict legacy protocols to satisfy CE Plus technical control requirements.
  • ✓Alternative IdP parity: While this framework focuses on Entra ID, equivalent contextual policies exist in Okta and Ping Identity, though native Intune signal integration offers cost advantages in pure Microsoft estates.
  • ✓Operational fail-safes: Maintain segregated, cloud-only break-glass credentials tested quarterly to safeguard against catastrophic administrative lockouts during enforcement.
  • ✓Validation over assumption: Staged Report-only telemetry and What If impact simulations prevent unexpected disruption to third-party integrations and B2B federated guests.
Frequently asked

FAQs — Microsoft Entra ID Conditional Access

How long should a Conditional Access policy remain in Report-only mode?

Microsoft recommends enabling each Conditional Access policy in Report-only mode for at least one week. This observation window allows administrators to review sign-in logs, evaluate the volume of legacy clients or unmanaged devices affected, communicate upcoming changes, and verify policy behaviour before enforcing controls.

Why do emergency-access break-glass accounts need explicit exclusions?

Because enforced Conditional Access policies evaluate all matching rules, an accidental misconfiguration could block all administrative access. Microsoft recommends two or more cloud-only emergency accounts in a dedicated group excluded from blocking policies, ensuring administrators can access the tenant if normal authentication paths fail.

What is the primary operational risk of legacy authentication?

Legacy authentication protocols cannot present interactive multifactor authentication challenges. Consequently, any account targeted through legacy endpoints can be compromised via basic credential attacks, bypassing modern MFA prompts entirely unless a dedicated policy explicitly blocks legacy client applications.

How does the Conditional Access What If tool assist deployment?

The What If tool simulates Conditional Access policy evaluation for specified users, target resources, client apps, and device platforms. It estimates which policies apply and which grant controls will be triggered without executing actual authentication or disrupting live production users.

Can device compliance and MFA be required simultaneously in one policy?

Yes. Microsoft's compliant-device policy supports combining Require multifactor authentication, Require device to be marked as compliant, and Require Microsoft Entra hybrid joined device as grant controls, allowing administrators to require all selected controls or any single satisfied condition.

Related

Got a question this article didn't answer?

One conversation with an engineer who's done this before. No sales script.

Talk to Servnet →

Talk to a UK specialist

Get expert advice or a no-obligation quote — servers, storage, networking, maintenance, finance and cloud. We reply the same working day.

or call 0800 987 4111