2.2.4. Conditional Access and Microsoft Entra ID Protection
💡 First Principle: Conditional Access is an if-then engine: if this user, on this app, from this location or device, at this level of risk, then require these controls or block. ID Protection supplies one of the "if" signals — the probability that the user or the sign-in is compromised — and Conditional Access decides what to do about it.
Licensing and prerequisites. Conditional Access needs at least Microsoft Entra ID P1; risk conditions need P2. Security defaults — free, Microsoft-managed baseline protection that requires MFA registration, blocks legacy authentication and requires MFA for admins — can't coexist with Conditional Access policies that replace them. You must turn security defaults off (Entra ID > Overview > Properties > Manage security defaults) before building Conditional Access.
Anatomy of a policy:
- Assignments pick users or groups (with exclusions — always exclude break-glass accounts), target resources (cloud apps), and conditions: sign-in risk, user risk, device platforms, locations, client apps, device filters.
- Grant controls either block or require controls — MFA, an authentication strength, a compliant or hybrid-joined device, an approved app, a password change. "Require all the selected controls" is the default; "Require one of the selected controls" is the alternative.
- Session controls shape the session after access: sign-in frequency (hours, days or every time), persistent browser session (requires targeting all cloud apps), and app enforced restrictions (Exchange Online and SharePoint Online only).
- Report-only mode evaluates a policy and records the result in the sign-in logs without enforcing it — the safe way to roll out any new policy.
Named locations define trusted networks or countries — up to 195 per tenant, with up to 2,000 IP ranges each. They're separate from the legacy MFA "trusted IPs" list.
Risk-based policies now live in Conditional Access. ID Protection calculates user risk (the probability the account is compromised, such as leaked credentials) and sign-in risk (the probability this particular authentication isn't the real user). The legacy user-risk and sign-in-risk policies configured on the ID Protection pages retired on 1 October 2026; you now build them as Conditional Access policies, which adds report-only mode, multiple policies and finer conditions.
| Risk type | Typical grant controls in Conditional Access |
|---|---|
| User risk | Require risk remediation (adaptive: a secure password change for password users, session revocation for passwordless users), require password change, or block |
| Sign-in risk | Require MFA, require reauthentication (sign-in frequency every time), or block |
"Require risk remediation" addresses user risk, not sign-in risk, and it isn't supported for guest users. Risk policies can only prompt for MFA if the user has already registered a method — the ID Protection MFA registration policy makes new users register at first sign-in.
⚠️ Exam Trap: Two traps share this section. First, "keep security defaults enabled and add a Conditional Access policy for the exceptions" is impossible — security defaults must be off. Second, "configure the user risk policy in ID Protection" describes a retired location; build it in Conditional Access with a user-risk condition.
Reflection Question: Leadership wants users with leaked credentials to fix the problem themselves without calling the help desk, but wants risky sign-ins from unfamiliar locations challenged rather than blocked. Describe the two Conditional Access policies you'd build and the grant control each uses.