MFA and Identity Telemetry Reference: Test Sign-In, Challenge, and Token-Revocation Detections
MFAIdentity SecurityEntra IDOktaGoogle WorkspaceSIEM

MFA and Identity Telemetry Reference: Test Sign-In, Challenge, and Token-Revocation Detections

PPayloads Lab Editorial Team
2026-11-20
7 min read

A practical reference for validating MFA, sign-in, device, session, and service-account telemetry across major identity providers.

Identity detections are only as reliable as the telemetry behind them. This reference gives security teams a safe, repeatable way to test sign-ins, MFA challenges, device enrollment, session revocation, and service-account authentication across Microsoft Entra ID, Okta, and Google Workspace—without disrupting real users.

Overview

A useful MFA detection test does more than confirm that an alert appeared. It verifies the complete evidence chain: the identity provider records the event, the connector collects it, the SIEM maps important fields, the detection evaluates the event, and an analyst can investigate the resulting alert. A successful test should also show what happened immediately before and after the authentication decision.

Use a dedicated test account, an approved test device, and a controlled network location. Keep the exercise separate from production users wherever the identity platform permits it. The purpose is to validate telemetry and analytics, not to create a realistic account takeover or cause repeated authentication failures.

Provider schemas change. Field names, authentication-method labels, risk classifications, and retention behavior may differ between tenants and connectors. Treat the field names below as a practical starting point, then confirm them against the events arriving in your own environment. For broader normalization guidance, see the SIEM Field Mapping Guide.

What to track

1. Build a provider-aware event inventory

Track the event types that support both prevention and investigation. A single successful sign-in is rarely enough to explain an identity decision. Record the test action, expected provider event, normalized SIEM fields, detection outcome, and evidence location.

Test scenarioEvidence to expectImportant fields to normalize
Successful sign-inAuthentication accepted and session establishedUser, timestamp, source IP, device, application, result, authentication method
Failed sign-inAuthentication rejected before or during the sign-in flowUser, failure reason, status, source IP, client, application, correlation or event ID
MFA denialPrimary authentication may succeed, but the challenge is denied or not completedChallenge result, factor or method, requirement, policy, result detail
Repeated challengesSeveral prompts or challenge attempts within a defined windowCount, time window, user, source, device, challenge outcome
New-device enrollmentA device, authenticator, passkey, or factor is registeredActor, target account, device or factor type, registration time, source, status
Risky-location simulationA sign-in from an approved test location or controlled egress pointLocation, IP, risk label, policy decision, application, device state
Session revocationRefresh tokens or active sessions are invalidated according to the platform's behaviorActor, target, action, timestamp, session or token context, result
Service-account authenticationNon-human identity authenticates through its expected protocol and sourceAccount type, workload or application, credential type, source, interactive flag

2. Compare the three provider views

Microsoft Entra ID commonly separates sign-in activity from audit activity and may expose distinct fields for status, conditional-access evaluation, authentication details, device information, and risk. Okta commonly records system-log events for authentication, factor challenges, enrollment, and policy decisions, with event types and outcome fields that should be preserved during ingestion. Google Workspace audit data may distinguish login activity, administrator actions, and token or application events. The exact names and availability depend on the service, edition, configuration, and connector.

For each provider, capture both the raw event and a normalized record. A portable identity schema should answer these questions:

  • Who attempted the action, and which account or resource was targeted?
  • When did the provider receive and record it?
  • Where did it originate, and what device or client was involved?
  • Which application, protocol, or workload requested access?
  • What authentication method and factor were used?
  • Was the result success, failure, challenge, denial, or policy block?
  • Which risk, policy, or conditional-access context influenced the decision?

3. Use a safe test-account checklist

Before testing, document the account owner, business purpose, provider, permitted applications, test device, approved source network, and rollback contact. Exclude the account from automated lockout escalation only through an approved, temporary control that cannot weaken real-user protections. Do not use a privileged administrator account for routine telemetry tests.

Prepare a test matrix with one row per scenario. Include the expected result, the expected raw event, the normalized fields, the alert name, and the analyst action. Never test repeated MFA prompts against an unsuspecting person. Use a test account and stop after the minimum number of events needed to validate the analytic.

For adjacent telemetry work, the MITRE ATT&CK Detection Coverage Worksheet can help map each safe test to an analytic and an evidence gap.

Cadence and checkpoints

A quarterly review is a practical baseline for identity telemetry, with additional checks after major authentication or SIEM changes. A monthly lightweight check is useful for high-value sign-in paths, especially during an MFA rollout or SSO migration.

Monthly smoke test

  1. Run one successful sign-in and one controlled failed sign-in.
  2. Confirm that the provider events arrive within the expected ingestion window.
  3. Check that user, source, device, application, result, and authentication-method fields are searchable.
  4. Verify that duplicate events do not create misleading alert counts.
  5. Record any missing or renamed fields in the test log.

Quarterly control review

  1. Repeat MFA denial, repeated-challenge, new-device enrollment, session-revocation, and service-account scenarios.
  2. Test each provider and each important log connector, not just the most commonly used tenant.
  3. Compare raw provider evidence with the SIEM's normalized representation.
  4. Review detection latency, alert grouping, severity, enrichment, and retention.
  5. Confirm that expected test activity is closed or labeled so it does not distort operational metrics.

Expected-versus-observed evidence template

Keep a compact record for every run:

Scenario: MFA denial
Provider and tenant: [name]
Test account and device: [identifier]
Expected provider event: [event type or documented description]
Observed event ID or correlation ID: [value]
Expected normalized fields: user, source.ip, device, auth.method, result, policy
Observed normalized fields: [values or missing]
Detection: [rule name, alert, or no alert]
Latency: [provider time to SIEM time]
Disposition: pass / partial / fail
Follow-up owner and date: [value]

Sample detection logic

Vendor-neutral logic for repeated MFA challenges can be expressed as:

GROUP BY user, source_ip
WHERE event_type IN (mfa_challenge, authentication_challenge)
  AND outcome IN (prompted, denied, failed)
COUNT events >= threshold
WITHIN time_window
EXCLUDE approved test runs
THEN alert with user, source_ip, device, application, and outcomes

A conceptual Microsoft Sentinel KQL pattern might look like this, subject to local table and field mappings:

SigninLogs
| where TimeGenerated > ago(15m)
| where ResultType != 0
| summarize Attempts=count(), Outcomes=make_set(ResultDescription)
    by UserPrincipalName, IPAddress, AppDisplayName
| where Attempts >= 5

A comparable Splunk pattern is:

index=identity (event_type=mfa_challenge OR event_type=authentication)
| where outcome IN (denied, failed, prompted)
| bin _time span=15m
| stats count dc(device_id) as devices values(outcome) as outcomes
  by user, src_ip, _time
| where count >= 5

These examples illustrate structure rather than a guaranteed production rule. Validate the actual event types, result values, time fields, and exclusions in your environment before enabling an alert.

How to interpret changes

A missing alert does not automatically mean the identity control failed. First determine whether the provider generated an event. If it did, check connector filters, ingestion delay, parsing, field mapping, and the detection's time window. If no provider event exists, review whether the test actually reached the expected authentication path or whether the platform records the action in a different audit stream.

Interpret common outcomes as follows:

  • Provider event present, SIEM event absent: investigate collection, permissions, API scope, connector health, and retention.
  • SIEM event present, key fields empty: inspect parser changes, schema drift, and normalization logic.
  • Fields present, no detection: compare result values, timestamps, aggregation thresholds, exclusions, and identity joins.
  • Detection fires repeatedly for one test: review duplicate ingestion, retries, event fan-out, and alert grouping before changing the threshold.
  • Alert fires but lacks context: add device, application, policy, factor, source, and correlation identifiers to the investigation view.

Separate telemetry completeness from analytic quality. A successful authentication event may be fully collected while still being insufficient to explain why MFA was or was not required. Track both coverage and usability: event arrival, field population, detection behavior, analyst context, and retention.

For regression discipline, pair this identity review with a documented rule-testing workflow such as the Detection Rule Regression Testing guide. The same principle applies to identity analytics: preserve test inputs, expected outcomes, observed evidence, and changes between runs.

When to revisit

Revisit this reference monthly for a smoke test and quarterly for a full identity-control review. Run an additional test whenever an identity provider releases a schema change, a SIEM connector is replaced, a conditional-access or authentication policy changes, a new factor is introduced, an SSO application is onboarded, or log retention and licensing settings are modified.

Start with the test matrix rather than an ad hoc sign-in. Select one scenario for each important control path, run it with an approved account, and save the provider event ID alongside the SIEM record. Mark each checkpoint as pass, partial, or fail. For every partial or failed result, assign an owner and a review date.

At the end of the review, update three items: the provider-to-SIEM field map, the detection test cases, and the analyst investigation guide. If a field or event label changed, record the old and new representation so future regression tests can distinguish a real behavioral change from a parser change. This simple record turns MFA detection testing into a repeatable SOC validation lab rather than a one-time exercise.

Related Topics

#MFA#Identity Security#Entra ID#Okta#Google Workspace#SIEM
P

Payloads Lab Editorial Team

Security Editorial Team

Senior editor and content strategist. Writing about technology, design, and the future of digital media. Follow along for deep dives into the industry's moving parts.