Detecting Credential Stuffing / Brute Force in Azure AD Sign-In Logs
Updated
Why this matters
Azure AD / Entra ID is the front door to almost everything in a modern Microsoft 365 environment — mail, files, Teams, and usually a dozen federated SaaS apps besides. A successful brute-force or credential-stuffing hit against it isn’t “one account compromised,” it’s “one identity that can now reach everything that identity was trusted with.”
Indicators to look for in Azure AD sign-in logs
- Repeated
status.errorCodevalues for invalid credentials against the sameuserPrincipalName, especially from a single IP or a small rotating pool - A burst of failures across many different
userPrincipalNamevalues from the same source — this is password spraying, the credential-stuffing variant that targets breadth over depth - A
riskLevelAggregatedofmediumorhighthat wasn’t acted on before a later successful sign-in - Sign-in attempts using
userAgentstrings associated with known authentication-attack tooling rather than real browsers - IP addresses or ASNs already flagged by external threat intelligence as credential-stuffing infrastructure
How LogTriage detects this
LogTriage enriches every sign-in’s source IP against ASN reputation, AbuseIPDB, OTX, and ThreatFox in addition to the rule-based credential-stuffing pattern detector. For Azure AD specifically, Microsoft’s own riskLevelAggregated field is folded directly into the event’s risk factors — so when Entra ID already suspected something and nothing was done about it, that gap is visible in the report instead of buried in a console nobody checked that day.
Detection / evidence checklist
- Determine whether this is targeted (one account, many passwords) or spray (many accounts, few passwords)
- Check if any of the targeted accounts have weak or reused passwords — that’s your real exposure
- Confirm smart lockout / Conditional Access sign-in risk policies are actually enabled, not just available
- Force MFA enrollment for any account without it, prioritizing accounts with elevated privileges
- Block the source IP/ASN range at the Conditional Access named-locations level if the activity is ongoing
Frequently Asked Questions
- What's the difference between a brute force attack and password spraying in Azure AD logs?
- Brute force targets one account with many password attempts — visible as many failure events against the same userPrincipalName. Password spraying targets many accounts with a small number of common passwords — the same source IP cycling through a large list of userPrincipalNames, each with only one or two attempts. Spraying is specifically designed to stay under per-account lockout thresholds.
- Does Microsoft Entra ID's own risk detection make LogTriage redundant?
- They're complementary. Entra ID's riskLevelAggregated and sign-in risk signals are incorporated directly into LogTriage's risk assessment — if Microsoft already suspected something and it wasn't acted on, that gap is visible in the report. LogTriage adds external threat intelligence (AbuseIPDB, OTX, ThreatFox) that Entra doesn't include, and produces a structured incident report rather than just flagged sign-in events.
- Can I detect attacks using legacy authentication protocols from sign-in logs?
- Yes. The clientAppUsed field in Azure AD sign-in logs identifies the authentication protocol. Legacy protocols like IMAP, POP3, and older Exchange ActiveSync clients appear in this field and can't support modern Conditional Access MFA requirements — making them a common bypass path. LogTriage flags legacy protocol usage as a distinct risk factor.
- Which export format does LogTriage expect for Azure AD sign-in logs?
- Both formats work: the Graph API export ({"value": [...]}) and the bare array format from the Entra portal export. LogTriage auto-detects both automatically.
Related Resources
Reference
- Entra ID (Azure AD) Sign-In Log Fields: field reference → every field explained, with real sample records
- Entra ID (AADSTS) sign-in error codes →
- Paste one record and see what each field means → free, runs in your browser
- Look up any error code in these events → free, runs in your browser
See this detection run on a real report
Try the live demo with a pre-loaded malicious log set — no signup required — or upload your own log file and get a full AI-reviewed threat report in minutes.