Home / Blog / Cheat Sheets
IAM Fundamentals for SOC Analysts: Who, How and What Access
A plain guide to authentication, authorisation, accountability and the joiner, mover, leaver lifecycle, with two KQL checks you can run in Entra ID logs.
Most alerts you triage are really identity questions. Who signed in, how did they prove it, and what were they allowed to do next? This cheat sheet gives SOC analysts a shared vocabulary for Identity and Access Management (IAM), so you can read sign in and audit logs with the right questions in mind.
Four words that do most of the work
People mix these terms up, and the mix up costs time in an investigation. Keep them separate.
| Term | The question it answers | What it looks like in logs |
|---|---|---|
| Identity | Who or what is this? | A user principal name, a device, a service account or an application |
| Authentication | Can they prove it? | Password, MFA, passkey or certificate steps, with a success or failure result |
| Authorisation | What are they allowed to do? | Role assignments, group membership, access policy decisions |
| Accountability | Can we trace the action to one identity? | Audit events with an actor, a target and a time |
MFA stands for multi factor authentication: proving who you are with more than one type of evidence.
The NCSC treats these as parts of one system. As far as I could verify, its introduction to identity and access management covers identity management, authentication, authorisation and privileged user management. The Cyber Assessment Framework (CAF) groups verification, authentication and authorisation together under principle B2.
Authentication is not authorisation
A successful sign in only proves that someone held the right credentials. It says nothing about what they should be able to reach.
- A sign in with result code 0 in Entra ID means authentication worked.
- It does not mean the person is who they claim to be, because a stolen password plus an approved prompt also gives result code 0.
- Authorisation then decides what that signed in identity can open, change or delete.
This is why attackers like to attack identity. If they can pass the authentication step as someone else, authorisation hands them that person's access.
The identity lifecycle: joiner, mover, leaver
Every identity has a life. Access that is right on day one is often wrong a year later.
- Joiner: a new starter gets an account and a starting set of access. Look for accounts created outside the normal process.
- Mover: someone changes team or role. They should gain the new access and lose the old. In practice the old access often stays.
- Leaver: the account is disabled and access removed. Look for sign ins after the leaving date.
Leftover access from movers is the quiet problem. A person who has changed jobs three times can hold the combined rights of all three. If their account is compromised, the attacker inherits all of it.
Where this shows up in Entra ID logs
Microsoft Entra ID keeps four kinds of sign in log, as documented on Microsoft Learn:
- Interactive user sign ins: a person types a password or answers an MFA prompt. Table:
SigninLogs. - Non interactive user sign ins: an app acts for the user, for example refreshing a token. Table:
AADNonInteractiveUserSignInLogs. - Service principal sign ins: an application signs in with no user involved. Table:
AADServicePrincipalSignInLogs. - Managed identity sign ins: an Azure resource uses its built in identity. Table:
AADManagedIdentitySignInLogs.
Table names depend on how your logs are routed to Microsoft Sentinel, so check your workspace before you copy a query.
To see how people are authenticating, start with a summary of successful interactive sign ins:
SigninLogs
| where TimeGenerated > ago(24h)
| where ResultType == "0"
| summarize SignIns = count(), Apps = dcount(AppDisplayName), IPs = dcount(IPAddress) by UserPrincipalName, AuthenticationRequirement
| order by SignIns desc
- Line 2 keeps the last 24 hours.
- Line 3 keeps only successful sign ins.
- Line 4 counts sign ins, distinct apps and distinct IP addresses per user, split by whether one factor or several were needed.
- Line 5 puts the busiest accounts first.
Look for privileged or sensitive accounts showing singleFactorAuthentication. That is an authentication gap worth reporting.
For authorisation, the audit log tells you who granted what:
AuditLogs
| where TimeGenerated > ago(7d)
| where OperationName == "Add member to role"
| extend Actor = tostring(InitiatedBy.user.userPrincipalName)
| extend Target = tostring(TargetResources[0].userPrincipalName)
| project TimeGenerated, Actor, Target, Result
- Line 3 keeps role assignment events.
- Lines 4 and 5 pull out who made the change and who received the role.
- Line 6 shows a short, readable table.
Compare each row with your change records. A role granted with no ticket, by an account that does not normally do this, is the lead to follow.
Tip: Use the Correlation ID to group the steps of one sign in. It is supplied by the client, so treat it as a helpful pointer and not as proof.
Five questions for any identity alert
- Identity: is this a person, a service account or an application? The playbook differs for each.
- Authentication: which factors were used, and did anything unusual happen first, such as repeated failures?
- Authorisation: what could this identity reach, and was that access still appropriate?
- Accountability: can every action be tied to one named identity, or is a shared account hiding the trail?
- Lifecycle: is the owner still in the same role, or still employed?
In government settings, record the answers against your change control and incident process. NCSC 10 Steps guidance also advises logging and monitoring authentication and authorisation events for signs of compromise, so these records support that work.
What to try next
- Run the first query in a test tenant and list which accounts show single factor sign ins.
- Pick three recent role assignments in the audit log and match each to a change ticket.
- Read the NCSC introduction to identity and access management, then the CAF B2 page, and note one gap in your own environment.
- Skim NIST SP 800-63-4, which replaced the earlier 800-63 series in 2025, to see how authentication assurance is described.
Authentication tells you who showed up. Authorisation tells you what they can do. Good triage keeps the two apart, then checks whether the lifecycle left any access behind.
Written by Freya, Senior SOC Analyst, Government