7.1 Planning, Assignments & Controls
Key Takeaways
- Conditional Access is Microsoft Entra’s policy engine that evaluates assignments (who), target resources (what), and conditions (when/how) before applying grant and session controls.
- Modern designs prefer Conditional Access over Security defaults for granular MFA, device, location, and authentication-strength requirements, and they generally require Microsoft Entra ID P1 (or a suite that includes it).
- Grant controls include require multifactor authentication, authentication strength, compliant device, Microsoft Entra hybrid joined device, approved client app, app protection policy, password change, and terms of use—often combined with require all or require one of selected controls.
- Always exclude carefully managed break-glass (emergency access) accounts from CA policies that could lock out every other admin path, and monitor those accounts continuously.
- Report-only mode and the What If tool let you pilot impact and simulate sign-ins before enforcing policies that could block production access.
Why Conditional Access dominates SC-300 Domain 2
Conditional Access (CA) is the enterprise policy engine in Microsoft Entra ID that answers: Given this identity, this resource, and this context, should we allow, block, or allow only after additional requirements? Domain 2—Implement authentication and access management—is the heaviest SC-300 slice (about 25–30%). CA questions appear constantly because nearly every modern MFA, device, location, risk, guest, and admin-protection design ends in a Conditional Access policy.
You already met CA at a high level when comparing Security defaults versus Conditional Access MFA. This chapter is the deep dive: policy anatomy, grant controls, session controls, Continuous Access Evaluation, authentication context, protected actions, templates, and troubleshooting.
Zero Trust and the CA mental model
Zero Trust assumes breach, verifies explicitly, and uses least-privilege access. Conditional Access operationalizes that at sign-in (and, with Continuous Access Evaluation, during the session for supported critical events):
- Who is requesting access? (users, groups, directory roles, workload identities, guests)
- What are they trying to reach? (cloud apps, user actions, authentication contexts)
- Under what conditions? (location, device platform, client app, device state filters, sign-in risk, user risk, and more)
- What must be true to grant? (MFA, strength, compliant device, hybrid joined, app protection, and combinations)
- How constrained is the session? (sign-in frequency, browser persistence, app-enforced restrictions, Microsoft Defender for Cloud Apps control)
If any enforced policy blocks, access is denied. If multiple policies apply, all matching policies must be satisfied—a frequent exam trap when candidates assume only the “strongest single policy” runs.
Licensing and when CA is available
| Capability | Typical license signal (exam-level) |
|---|---|
| Conditional Access policies | Microsoft Entra ID P1 (or Microsoft 365 / Entra suites that include P1) |
| Risk-based CA (user/sign-in risk conditions) | Microsoft Entra ID P2 / Identity Protection |
| Security defaults (simple baseline MFA) | Available without P1; not a granular CA replacement |
SC-300 stems often assume the license required by the feature. If the scenario says “no P1,” prefer Security defaults for baseline MFA—not a full CA design. If the scenario needs exceptions, per-app rules, device grants, or authentication strengths, answer Conditional Access.
Plan Conditional Access for the organization
Planning questions rarely say “create one policy for everything.” They describe maturity, risk, and operational constraints.
| Planning question | Design implication |
|---|---|
| Who are the break-glass accounts? | Exclude from restrictive CA; monitor and protect separately |
| Which apps hold Tier-0 / regulated data? | Stricter grants (phishing-resistant strength, compliant device) |
| Hybrid or cloud-only endpoints? | Compliant device and/or Microsoft Entra hybrid joined |
| Guests and external collaboration? | Separate guest policies; do not blindly mirror member rules |
| Legacy authentication still present? | Block legacy auth with CA; modernize apps |
| Helpdesk readiness for MFA lockouts? | Report-only → pilot groups → enforce; TAP for recovery |
| Workload identities (service principals)? | Separate workload identity CA where licensed/supported |
Recommended rollout pattern for SC-300 design answers:
- Inventory apps, user populations, and emergency access accounts.
- Enable authentication methods and drive registration (previous chapter).
- Create baseline policies from templates or carefully designed custom policies (MFA for admins, MFA for all users, block legacy auth).
- Put new policies in report-only first; review insights and sign-in logs.
- Pilot with IT / security groups; refine exclusions.
- Enforce for production; monitor false positives and guest impact.
- Layer device, location, risk, authentication strength, and session controls as maturity grows.
Policy components: assignments
Every Conditional Access policy has assignments and access controls. Assignments answer who and what—and under which conditions.
Users, groups, roles, and workload identities
| Assignment target | Use on SC-300 |
|---|---|
| All users | Broad workforce baselines (still exclude break-glass) |
| Select users and groups | Pilots, departments, privileged cohorts |
| Directory roles | Target Global Administrator, Privileged Role Administrator, and other roles without maintaining a static group |
| Guests / external users | Guest-specific policies or include guests only where intended |
| Workload identities | Service principals / apps as identities—separate from human-user policies when the feature set applies |
| Exclude | Break-glass accounts, service accounts that cannot MFA, carefully justified exceptions |
Exclude is as important as include. Classic failure: policy includes All users, targets All cloud apps, requires MFA, and accidentally includes emergency accounts that have no second factor registered—tenant lockout risk.
Cloud apps, user actions, and authentication context
| Target type | Meaning |
|---|---|
| Cloud apps | Microsoft 365 apps, Azure Management, third-party enterprise apps registered/integrated with Entra |
| All cloud apps | Broad net; powerful but easy to over-scope |
| User actions | Special actions such as Register security information or Register or join devices |
| Authentication context | Labels you define for step-up protection on sensitive resources/actions (section 7.3) |
Exam cue: “Require MFA when users register security info” → target the Register security information user action, not only Exchange Online. “Protect Azure portal admin access” → include Microsoft Azure Management (or equivalent Azure management app targeting in CA), not only Office 365.
Conditions
Conditions refine when the policy applies.
| Condition | Typical SC-300 use |
|---|---|
| Locations (named locations / countries / trusted IPs) | Require MFA outside corporate egress; block high-risk countries |
| Device platforms | iOS/Android/Windows/macOS-specific rules |
| Client apps | Browser, mobile apps/desktop clients, legacy authentication clients |
| Device state / device filters | Filter on device attributes for advanced targeting |
| Sign-in risk (P2) | Require MFA or block on medium/high sign-in risk |
| User risk (P2) | Require password change and/or block on high user risk |
| Insider risk (where available/licensed) | Advanced scenarios beyond baseline associate depth |
Named locations replace old “trusted IPs that skip MFA” thinking with explicit include/exclude logic. Marking a location as trusted can interact with risk calculations and policy design—do not treat “trusted = safe forever.”
Legacy authentication (older protocols that cannot do interactive MFA cleanly) should typically be blocked with a dedicated CA policy targeting legacy client apps. Leaving legacy open undermines MFA for many mail clients and scripts.
Grant controls
Grant controls decide whether access is blocked or allowed only if requirements are met.
| Grant control | What it requires | Common exam scenario |
|---|---|---|
| Block access | Deny when conditions match | Block legacy auth; block unknown countries |
| Require multifactor authentication | Any allowed MFA method per methods policy | Baseline workforce MFA |
| Require authentication strength | Specific strength (for example phishing-resistant MFA) | Admins, Tier-0, regulated apps |
| Require device to be marked as compliant | Intune (or equivalent MDM) compliance | Corporate data only on healthy managed devices |
| Require Microsoft Entra hybrid joined device | Device hybrid joined to on-premises AD and Entra | Hybrid Windows fleets without full Intune compliance yet |
| Require approved client app | App on Microsoft’s approved client list | Older mobile data protection patterns |
| Require app protection policy | Intune App Protection (MAM) on the client | BYOD with protected Outlook/Teams, not full device enrollment |
| Require password change | User must change password (often with risk) | High user risk remediation |
| Require terms of use | User must accept ToU | Legal acceptance before app access |
You can select multiple grant controls and choose:
- Require all the selected controls — every chosen control must pass (stricter).
- Require one of the selected controls — any one is enough (flexible).
Classic design: for Windows hybrid users, require one of compliant or hybrid joined so either managed path works. For executives on a sensitive app, require all of phishing-resistant authentication strength and compliant device.
Authentication strength vs plain MFA
Require multifactor authentication accepts any method the tenant allows that satisfies MFA. Require authentication strength points at a built-in or custom strength definition (for example Phishing-resistant MFA covering FIDO2, CBA, Windows Hello for Business in appropriate configurations). When a stem says “phishing-resistant,” do not answer plain MFA alone—answer authentication strength (plus methods enablement/registration from chapter 6).
Approved client app vs app protection policy
| Control | Idea | Prefer when |
|---|---|---|
| Approved client app | Only Microsoft-approved clients may access | Older designs; limited modern emphasis |
| App protection policy | Intune MAM policies protect org data in the app | BYOD mobile access without full MDM enrollment |
SC-300 increasingly favors app protection policy (and device compliance) over relying only on approved client app for modern mobile scenarios.
Report-only mode
Report-only is a policy state that evaluates the policy and logs what would have happened without enforcing block/grant challenges in the full enforcement sense used for production impact analysis. Use it to:
- Estimate how many users would be prompted or blocked.
- Find apps and populations missing MFA registration or compliant devices.
- Validate guest impact before go-live.
Workflow: create policy → Report-only → review Conditional Access insights / sign-in logs → fix exclusions and registration gaps → On for a pilot group → expand.
Report-only is not a permanent security control. Leaving “require MFA” in report-only forever means attackers still get password-only access.
What If tool (concepts)
The What If tool simulates a sign-in: you pick a user, app, IP/location, device platform, client app, and other signals. Entra returns which policies would apply and what grants would be required. Use What If when:
- A user reports unexpected MFA or blocks.
- You are validating that break-glass exclusions work.
- You need to explain multi-policy stacking (“why two MFA policies hit this app”).
What If is a simulation, not a substitute for reading sign-in logs after a real failure. On the exam, if the stem says “predict impact before enabling,” think report-only + What If; if it says “user failed yesterday at 14:02,” think sign-in logs / policy details.
Break-glass (emergency access) exclusions
Break-glass accounts are emergency admin accounts used when normal identity systems fail (MFA outage, misconfigured CA, IdP issues). Microsoft guidance emphasizes:
- At least two cloud-only emergency accounts (not dependent on a single on-premises sync path).
- Excluded from Conditional Access policies that could prevent their use.
- Strong credentials (long complex passwords and/or FIDO2), stored securely, not used daily.
- Continuous monitoring and alerting on any sign-in.
- Documented emergency procedures.
Exam wrong answers: “Never exclude anyone from CA,” or “Put break-glass in the same MFA policy as Global Administrators with SMS only.” Right answers balance lockout prevention with detection.
Putting assignments and grants together
Example policy set (conceptual, not the only valid design):
| Policy intent | Include | Target | Conditions | Grant |
|---|---|---|---|---|
| Admin MFA baseline | Directory roles: admins | All cloud apps | — | Auth strength phishing-resistant |
| User MFA baseline | All users; exclude break-glass | All cloud apps | — | Require MFA |
| Block legacy auth | All users | All cloud apps | Client apps: legacy | Block |
| Sensitive finance app | Finance group | Payroll app | — | MFA + compliant device (require all) |
| Register security info | All users | User action: register security info | — | Require MFA |
When you read an SC-300 stem, extract who, what resource, which condition, and which grant. If the answer choice only configures authentication methods policy without CA, it enables factors but may not enforce them. If it only toggles Security defaults, it cannot express per-app phishing-resistant device grants.
An enterprise with Microsoft Entra ID P1 must require multifactor authentication for all users accessing Microsoft 365, exclude two emergency access accounts, and later add per-app device requirements. Which approach best fits a modern SC-300 design?
A Conditional Access policy must force phishing-resistant methods (such as FIDO2 or certificate-based authentication) for Global Administrators. Which grant control best expresses that requirement?
Before enforcing a new “require compliant device for SharePoint” policy, the identity team wants to estimate how many users would be blocked without changing production access. What should they use first?
Why should break-glass emergency access accounts typically be excluded from restrictive Conditional Access policies?