User Activity Tracking with KQL

Microsoft Sentinel & Defender XDR - SOC Analyst Playbook

🎯 Overview

This playbook provides KQL queries and investigation workflows for account usage analysis across:

Microsoft Sentinel (Azure Sentinel)
Microsoft Defender XDR (Defender for Endpoint, Identity, Cloud Apps)
Azure Active Directory (Entra ID)
Windows Security Events

📊 Data Sources Quick Reference

Available Tables

Table Name

Source

Description

Retention

SecurityEvent

Windows Events

Security.evtx via Log Analytics

90 days default

IdentityLogonEvents

Defender for Identity

Domain authentication

30 days

DeviceLogonEvents

Defender for Endpoint

Local logons

30 days

DeviceEvents

Defender for Endpoint

General device activity

30 days

SigninLogs

Azure AD/Entra ID

Cloud authentication

30 days

AADNonInteractiveUserSignInLogs

Azure AD

Service principal/app auth

30 days

AuditLogs

Azure AD

Account changes

30 days

IdentityDirectoryEvents

Defender for Identity

AD changes

30 days

CloudAppEvents

Defender for Cloud Apps

SaaS activity

30 days

BehaviorAnalytics

Sentinel UEBA

User behavior anomalies

14 days

OfficeActivity

Office 365

Exchange/SharePoint/Teams

90 days

Event ID to Table Mapping

Windows Event ID

Sentinel Table

Defender XDR Table

4624 (Logon)

SecurityEvent

DeviceLogonEvents

4625 (Failed Logon)

SecurityEvent

DeviceLogonEvents

4648 (Explicit Creds)

SecurityEvent

DeviceEvents

4672 (Admin Logon)

SecurityEvent

DeviceLogonEvents

4776 (NTLM Auth)

SecurityEvent

IdentityLogonEvents

4768 (Kerberos TGT)

SecurityEvent

IdentityLogonEvents

4769 (Kerberos Service)

SecurityEvent

IdentityLogonEvents

4771 (Kerberos Failed)

SecurityEvent

IdentityLogonEvents

🔍 Phase 1: Initial Triage Queries

1.1 Quick Account Profile (Sentinel)

// Get comprehensive account overview
let TargetAccount = "john.doe@contoso.com"; // or "DOMAIN\\username"
let TimeRange = 7d;
// Check if account exists and get details
IdentityInfo
| where TimeGenerated > ago(TimeRange)
| where AccountUPN == TargetAccount or AccountName contains TargetAccount
| summarize 
    arg_max(TimeGenerated, *),
    GroupMemberships = make_set(GroupMembership),
    Tags = make_set(Tags)
| project 
    AccountName,
    AccountUPN,
    AccountDomain,
    Department,
    JobTitle,
    Manager,
    IsAccountEnabled,
    GroupMemberships,
    Tags,
    LastSeen = TimeGenerated

1.2 Quick Account Profile (Defender XDR)

// Get account activity summary
let TargetAccount = "john.doe";
let TimeRange = 7d;
DeviceLogonEvents
| where Timestamp > ago(TimeRange)
| where AccountName == TargetAccount
| summarize 
    TotalLogons = count(),
    SuccessfulLogons = countif(ActionType == "LogonSuccess"),
    FailedLogons = countif(ActionType == "LogonFailed"),
    UniqueDevices = dcount(DeviceName),
    UniqueIPs = dcount(RemoteIP),
    LogonTypes = make_set(LogonType),
    FirstSeen = min(Timestamp),
    LastSeen = max(Timestamp)
| extend FailureRate = round(FailedLogons * 100.0 / TotalLogons, 2)

1.3 Is This Account Under Attack? (Quick Check)

// Rapid compromise indicators
let TargetAccount = "john.doe@contoso.com";
let TimeRange = 24h;
union 
    // Failed logons
    (SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4625
    | where TargetUserName contains TargetAccount
    | summarize FailedLogons = count() by IpAddress
    | where FailedLogons > 10
    | extend Indicator = "High Failed Logons", Severity = "High"),
    // Impossible travel
    (SigninLogs
    | where TimeGenerated > ago(TimeRange)
    | where UserPrincipalName == TargetAccount
    | where ResultType == "0"
    | extend Country = tostring(LocationDetails.countryOrRegion)
    | order by TimeGenerated asc
    | serialize
    | extend PrevCountry = prev(Country), TimeDiff = datetime_diff('minute', TimeGenerated, prev(TimeGenerated))
    | where Country != PrevCountry and TimeDiff < 60
    | summarize ImpossibleTravel = count()
    | where ImpossibleTravel > 0
    | extend Indicator = "Impossible Travel", Severity = "Critical"),
    // Suspicious logon type changes
    (DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where AccountName contains TargetAccount
    | summarize LogonTypes = make_set(LogonType)
    | where array_length(LogonTypes) > 3
    | extend Indicator = "Multiple Logon Types", Severity = "Medium")
| project Indicator, Severity

🔐 Phase 2: Authentication Analysis

2.1 Complete Authentication Timeline (Sentinel)

// Unified authentication timeline across all sources
let TargetAccount = "john.doe@contoso.com";
let TimeRange = 7d;
union 
    // Windows Security Events
    (SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID in (4624, 4625, 4648, 4672, 4776, 4768, 4769, 4771)
    | where TargetUserName contains TargetAccount or Account contains TargetAccount
    | extend 
        AuthType = case(
            EventID == 4624, "Successful Logon",
            EventID == 4625, "Failed Logon",
            EventID == 4648, "Explicit Credentials",
            EventID == 4672, "Admin Logon",
            EventID == 4776, "NTLM Auth",
            EventID == 4768, "Kerberos TGT",
            EventID == 4769, "Kerberos Service",
            EventID == 4771, "Kerberos Failed",
            "Other"
        ),
        LogonTypeName = case(
            LogonType == "2", "Interactive",
            LogonType == "3", "Network",
            LogonType == "4", "Batch",
            LogonType == "5", "Service",
            LogonType == "7", "Unlock",
            LogonType == "8", "NetworkCleartext",
            LogonType == "9", "NewCredentials",
            LogonType == "10", "RDP",
            LogonType == "11", "CachedInteractive",
            strcat("Type ", LogonType)
        )
    | project 
        TimeGenerated,
        Source = "WindowsEvent",
        AuthType,
        Status = case(EventID in (4625, 4771), "Failed", "Success"),
        Account = TargetUserName,
        Computer,
        IpAddress,
        LogonTypeName,
        EventID),
    // Azure AD Sign-ins
    (SigninLogs
    | where TimeGenerated > ago(TimeRange)
    | where UserPrincipalName == TargetAccount
    | extend 
        AuthType = "Azure AD Sign-in",
        Status = case(ResultType == "0", "Success", "Failed")
    | project 
        TimeGenerated,
        Source = "AzureAD",
        AuthType,
        Status,
        Account = UserPrincipalName,
        Computer = DeviceDetail.displayName,
        IpAddress = IPAddress,
        LogonTypeName = AppDisplayName,
        EventID = ResultType),
    // Defender for Identity
    (IdentityLogonEvents
    | where Timestamp > ago(TimeRange)
    | where AccountUpn == TargetAccount or AccountName contains TargetAccount
    | extend 
        AuthType = Protocol,
        Status = case(ActionType == "LogonSuccess", "Success", "Failed")
    | project 
        TimeGenerated = Timestamp,
        Source = "DefenderIdentity",
        AuthType,
        Status,
        Account = AccountName,
        Computer = DestinationDeviceName,
        IpAddress = IPAddress,
        LogonTypeName = LogonType,
        EventID = ActionType),
    // Defender for Endpoint
    (DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where AccountName contains TargetAccount
    | extend 
        AuthType = "Local Logon",
        Status = case(ActionType == "LogonSuccess", "Success", "Failed")
    | project 
        TimeGenerated = Timestamp,
        Source = "DefenderEndpoint",
        AuthType,
        Status,
        Account = AccountName,
        Computer = DeviceName,
        IpAddress = RemoteIP,
        LogonTypeName = LogonType,
        EventID = ActionType)
| order by TimeGenerated desc
| extend 
    Hour = hourofday(TimeGenerated),
    DayOfWeek = dayofweek(TimeGenerated)

2.2 Failed Logon Analysis - Brute Force Detection

// Detect brute force and password spray attacks
let TimeRange = 24h;
let FailedThreshold = 10;
SecurityEvent
| where TimeGenerated > ago(TimeRange)
| where EventID == 4625 // Failed logons
| extend 
    FailureReason = case(
        Status == "0xC000006D", "Bad Username",
        Status == "0xC000006E", "Account Restriction",
        Status == "0xC000006F", "Time Restriction",
        Status == "0xC0000070", "Workstation Restriction",
        Status == "0xC0000071", "Password Expired",
        Status == "0xC0000072", "Account Disabled",
        Status == "0xC000006A", "Bad Password",
        Status == "0xC0000234", "Account Locked",
        Status == "0xC0000193", "Account Expired",
        Status == "0xC0000064", "Account Does Not Exist",
        Status
    )
| summarize 
    FailedAttempts = count(),
    TargetAccounts = make_set(TargetUserName),
    UniqueAccounts = dcount(TargetUserName),
    FailureReasons = make_set(FailureReason),
    StartTime = min(TimeGenerated),
    EndTime = max(TimeGenerated),
    Computers = make_set(Computer)
    by IpAddress
| where FailedAttempts >= FailedThreshold
| extend 
    Duration = datetime_diff('minute', EndTime, StartTime),
    AttackType = case(
        UniqueAccounts > 10 and FailedAttempts / UniqueAccounts < 5, "Password Spray",
        UniqueAccounts < 3 and FailedAttempts > 50, "Brute Force",
        "Suspicious Activity"
    ),
    Severity = case(
        FailedAttempts > 100, "Critical",
        FailedAttempts > 50, "High",
        "Medium"
    )
| order by FailedAttempts desc
| project 
    IpAddress,
    AttackType,
    Severity,
    FailedAttempts,
    UniqueAccounts,
    TargetAccounts,
    Duration,
    StartTime,
    EndTime,
    Computers,
    FailureReasons

2.3 Successful Logon After Failed Attempts (Successful Breach)

// Find accounts where brute force succeeded
let TimeRange = 24h;
let FailedThreshold = 5;
let SuccessWindow = 30m;
let FailedLogons = SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4625
    | summarize 
        FailedCount = count(),
        LastFailed = max(TimeGenerated)
        by IpAddress, TargetUserName, Computer;
let SuccessfulLogons = SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4624
    | where LogonType in ("2", "3", "10") // Interactive, Network, RDP
    | project 
        SuccessTime = TimeGenerated,
        IpAddress,
        Account = TargetUserName,
        Computer,
        LogonType;
FailedLogons
| where FailedCount >= FailedThreshold
| join kind=inner (SuccessfulLogons) on IpAddress, $left.TargetUserName == $right.Account, Computer
| where datetime_diff('minute', SuccessTime, LastFailed) between (0 .. 30)
| project 
    Computer,
    Account,
    IpAddress,
    FailedAttempts = FailedCount,
    LastFailedAttempt = LastFailed,
    SuccessfulLogon = SuccessTime,
    TimeBetween = datetime_diff('minute', SuccessTime, LastFailed),
    LogonType,
    Severity = "Critical"
| order by SuccessfulLogon desc

2.4 Logon Type Analysis (Sentinel)

// Analyze logon type distribution and anomalies
let TargetAccount = "john.doe";
let TimeRange = 30d;
let BaselinePeriod = 14d;
// Get baseline logon type distribution
let Baseline = SecurityEvent
    | where TimeGenerated between (ago(TimeRange) .. ago(BaselinePeriod))
    | where EventID == 4624
    | where TargetUserName contains TargetAccount
    | summarize BaselineCount = count() by LogonType
    | extend BaselinePercentage = round(100.0 * BaselineCount / toscalar(
        SecurityEvent
        | where TimeGenerated between (ago(TimeRange) .. ago(BaselinePeriod))
        | where EventID == 4624
        | where TargetUserName contains TargetAccount
        | summarize count()
    ), 2);
// Get recent logon type distribution
let Recent = SecurityEvent
    | where TimeGenerated > ago(BaselinePeriod)
    | where EventID == 4624
    | where TargetUserName contains TargetAccount
    | summarize RecentCount = count() by LogonType
    | extend RecentPercentage = round(100.0 * RecentCount / toscalar(
        SecurityEvent
        | where TimeGenerated > ago(BaselinePeriod)
        | where EventID == 4624
        | where TargetUserName contains TargetAccount
        | summarize count()
    ), 2);
// Compare and identify anomalies
Baseline
| join kind=fullouter (Recent) on LogonType
| extend 
    LogonType = coalesce(LogonType, LogonType1),
    LogonTypeName = case(
        LogonType == "2", "Interactive (Console)",
        LogonType == "3", "Network",
        LogonType == "4", "Batch",
        LogonType == "5", "Service",
        LogonType == "7", "Unlock/Reconnect",
        LogonType == "8", "NetworkCleartext",
        LogonType == "9", "NewCredentials (RunAs)",
        LogonType == "10", "Remote Desktop",
        LogonType == "11", "CachedInteractive",
        strcat("Type ", LogonType)
    ),
    BaselineCount = coalesce(BaselineCount, 0),
    RecentCount = coalesce(RecentCount, 0),
    BaselinePercentage = coalesce(BaselinePercentage, 0.0),
    RecentPercentage = coalesce(RecentPercentage, 0.0)
| extend 
    PercentageChange = round(RecentPercentage - BaselinePercentage, 2),
    IsAnomaly = case(
        BaselineCount == 0 and RecentCount > 0, "NEW",
        abs(RecentPercentage - BaselinePercentage) > 20, "ANOMALY",
        "NORMAL"
    )
| order by IsAnomaly desc, abs(PercentageChange) desc
| project 
    LogonType,
    LogonTypeName,
    BaselineCount,
    BaselinePercentage,
    RecentCount,
    RecentPercentage,
    PercentageChange,
    IsAnomaly

🗺️ Phase 3: Lateral Movement Detection

3.1 Network Logon Chain Analysis (Defender XDR)

// Detect lateral movement patterns
let TimeRange = 24h;
let TargetAccount = "john.doe";
// Get all network logons
let NetworkLogons = DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where AccountName contains TargetAccount
    | where LogonType in ("Network", "RemoteInteractive")
    | where ActionType == "LogonSuccess"
    | project 
        Timestamp,
        SourceDevice = DeviceName,
        SourceIP = RemoteIP,
        Account = AccountName,
        LogonType
    | order by Timestamp asc;
// Detect pivot chains
NetworkLogons
| serialize
| extend 
    NextDevice = next(SourceDevice),
    NextTimestamp = next(Timestamp),
    TimeDelta = datetime_diff('second', next(Timestamp), Timestamp)
| where TimeDelta < 300 // Within 5 minutes
| where SourceDevice != NextDevice
| summarize 
    LateralMoveCount = count(),
    Path = make_list(strcat(SourceDevice, " -> ", NextDevice)),
    Timeline = make_list(Timestamp),
    FirstSeen = min(Timestamp),
    LastSeen = max(Timestamp)
    by Account
| where LateralMoveCount >= 2
| extend Severity = case(
    LateralMoveCount > 10, "Critical",
    LateralMoveCount > 5, "High",
    "Medium"
)
| order by LateralMoveCount desc

3.2 Lateral Movement with Process Correlation

// Correlate logons with suspicious process execution
let TimeRange = 24h;
let SuspiciousProcesses = dynamic([
    "powershell.exe", "cmd.exe", "wmic.exe", "psexec.exe", 
    "mmc.exe", "mshta.exe", "rundll32.exe", "regsvr32.exe"
]);
// Get network logons
let Logons = DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where LogonType == "Network"
    | where ActionType == "LogonSuccess"
    | project LogonTime = Timestamp, DeviceName, AccountName, RemoteIP;
// Get process creations
let Processes = DeviceProcessEvents
    | where Timestamp > ago(TimeRange)
    | where FileName has_any (SuspiciousProcesses)
    | project ProcessTime = Timestamp, DeviceName, FileName, ProcessCommandLine, AccountName;
// Correlate
Logons
| join kind=inner (Processes) on DeviceName, AccountName
| where datetime_diff('second', ProcessTime, LogonTime) between (0 .. 300)
| project 
    DeviceName,
    AccountName,
    RemoteIP,
    LogonTime,
    ProcessTime,
    TimeDelta = datetime_diff('second', ProcessTime, LogonTime),
    SuspiciousProcess = FileName,
    CommandLine = ProcessCommandLine
| order by LogonTime asc

3.3 Administrative Reconnaissance Detection

// Detect admin enumeration and reconnaissance
let TimeRange = 24h;
let ReconCommands = dynamic([
    "net view", "net user", "net group", "net localgroup",
    "whoami", "ipconfig", "systeminfo", "tasklist",
    "net accounts", "net share", "net session"
]);
DeviceProcessEvents
| where Timestamp > ago(TimeRange)
| where ProcessCommandLine has_any (ReconCommands)
| summarize 
    ReconCommandCount = count(),
    Commands = make_set(ProcessCommandLine),
    Devices = make_set(DeviceName),
    FirstSeen = min(Timestamp),
    LastSeen = max(Timestamp)
    by AccountName
| where ReconCommandCount >= 3
| extend 
    Severity = case(
        ReconCommandCount > 10, "Critical",
        ReconCommandCount > 5, "High",
        "Medium"
    )
| order by ReconCommandCount desc

3.4 Pass-the-Hash Detection (Sentinel)

// Detect potential pass-the-hash attacks
let TimeRange = 24h;
// Look for NTLM authentication without preceding local logon
let NTLMAuth = SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4776 // NTLM authentication
    | extend SourceComputer = Computer
    | project NTLMTime = TimeGenerated, TargetUserName, SourceComputer, Workstation;
let LocalLogons = SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4624
    | where LogonType == "2" // Interactive
    | project LogonTime = TimeGenerated, Account = TargetUserName, Computer;
// Find NTLM without recent local logon
NTLMAuth
| join kind=leftanti (
    LocalLogons 
    | where LogonTime > ago(24h)
) on $left.TargetUserName == $right.Account, $left.Workstation == $right.Computer
| summarize 
    SuspiciousNTLM = count(),
    Workstations = make_set(Workstation),
    SourceSystems = make_set(SourceComputer),
    Timeline = make_list(NTLMTime)
    by TargetUserName
| where SuspiciousNTLM >= 3
| extend Severity = "High"
| order by SuspiciousNTLM desc

🚨 Phase 4: Privilege Escalation & Admin Activity

4.1 Detect New Admin Rights Assignments

// Monitor for admin rights granted
let TimeRange = 7d;
union
    // Windows Events - Admin logons
    (SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4672 // Special privileges assigned
    | where PrivilegeList contains "SeDebugPrivilege" or PrivilegeList contains "SeBackupPrivilege"
    | extend EventType = "SpecialPrivileges"
    | project TimeGenerated, Account = TargetUserName, Computer, EventType, PrivilegeList),
    // Azure AD - Role assignments
    (AuditLogs
    | where TimeGenerated > ago(TimeRange)
    | where OperationName in ("Add member to role", "Add eligible member to role")
    | extend Account = tostring(TargetResources[0].userPrincipalName)
    | extend Role = tostring(TargetResources[0].modifiedProperties[0].newValue)
    | extend EventType = "RoleAssignment"
    | project TimeGenerated, Account, EventType, Role),
    // Defender for Identity - Group membership changes
    (IdentityDirectoryEvents
    | where Timestamp > ago(TimeRange)
    | where ActionType == "Group Membership changed"
    | where DestinationDeviceName contains "Admin" or DestinationDeviceName contains "Domain"
    | extend EventType = "GroupMembership"
    | project TimeGenerated = Timestamp, Account = AccountName, EventType, Group = DestinationDeviceName)
| order by TimeGenerated desc

4.2 Explicit Credential Usage (RunAs)

// Detect runas and explicit credential usage
let TimeRange = 7d;
let TargetAccount = ""; // Leave empty for all accounts
SecurityEvent
| where TimeGenerated > ago(TimeRange)
| where EventID == 4648 // Logon with explicit credentials
| where isempty(TargetAccount) or TargetUserName contains TargetAccount
| extend 
    SourceAccount = Account,
    TargetAccountUsed = TargetUserName,
    TargetServer = TargetServerName
| summarize 
    Count = count(),
    TargetServers = make_set(TargetServer),
    FirstSeen = min(TimeGenerated),
    LastSeen = max(TimeGenerated)
    by SourceAccount, TargetAccountUsed, Computer
| where Count >= 3
| extend 
    Severity = case(
        TargetAccountUsed contains "admin" or TargetAccountUsed contains "svc", "High",
        Count > 10, "Medium",
        "Low"
    )
| order by Severity desc, Count desc

4.3 Service Account Interactive Logon Detection

// Service accounts should never log on interactively
let ServiceAccounts = dynamic(["svc-", "service-", "sa-"]); // Adjust patterns
let TimeRange = 7d;
DeviceLogonEvents
| where Timestamp > ago(TimeRange)
| where LogonType in ("Interactive", "RemoteInteractive")
| where AccountName has_any (ServiceAccounts)
| summarize 
    InteractiveLogons = count(),
    Devices = make_set(DeviceName),
    IPs = make_set(RemoteIP),
    FirstSeen = min(Timestamp),
    LastSeen = max(Timestamp)
    by AccountName
| extend Severity = "Critical"
| order by InteractiveLogons desc

🌍 Phase 5: Azure AD / Cloud Authentication Analysis

// Detailed Azure AD authentication analysis
let TargetAccount = "john.doe@contoso.com";
let TimeRange = 7d;
SigninLogs
| where TimeGenerated > ago(TimeRange)
| where UserPrincipalName == TargetAccount
| extend 
    Country = tostring(LocationDetails.countryOrRegion),
    City = tostring(LocationDetails.city),
    State = tostring(LocationDetails.state),
    Latitude = tostring(LocationDetails.geoCoordinates.latitude),
    Longitude = tostring(LocationDetails.geoCoordinates.longitude),
    DeviceOS = tostring(DeviceDetail.operatingSystem),
    DeviceBrowser = tostring(DeviceDetail.browser),
    IsCompliant = tostring(DeviceDetail.isCompliant),
    IsManaged = tostring(DeviceDetail.isManaged),
    TrustType = tostring(DeviceDetail.trustType),
    Status = case(ResultType == "0", "Success", "Failed"),
    FailureReason = case(
        ResultType == "50053", "Account Locked",
        ResultType == "50055", "Password Expired",
        ResultType == "50057", "Account Disabled",
        ResultType == "50126", "Invalid Credentials",
        ResultType == "50074", "Strong Auth Required",
        ResultType == "50076", "MFA Required",
        ResultType == "50079", "User needs to enroll MFA",
        ResultType == "50158", "External Security Challenge",
        ResultType == "53003", "Blocked by CA Policy",
        ResultType == "70008", "Session Expired",
        ResultDescription
    ),
    RiskLevel = tostring(RiskLevelDuringSignIn),
    RiskState = tostring(RiskState)
| project 
    TimeGenerated,
    UserPrincipalName,
    Status,
    FailureReason,
    AppDisplayName,
    IPAddress,
    Country,
    City,
    DeviceOS,
    DeviceBrowser,
    IsCompliant,
    IsManaged,
    AuthenticationRequirement,
    ConditionalAccessStatus,
    RiskLevel,
    RiskState,
    MfaDetail = AuthenticationDetails
| order by TimeGenerated desc

5.2 Impossible Travel Detection

// Detect impossible travel scenarios
let TargetAccount = "john.doe@contoso.com";
let TimeRange = 7d;
let MaxTravelSpeedKmH = 900; // ~airplane speed
SigninLogs
| where TimeGenerated > ago(TimeRange)
| where UserPrincipalName == TargetAccount
| where ResultType == "0" // Successful only
| extend 
    Country = tostring(LocationDetails.countryOrRegion),
    City = tostring(LocationDetails.city),
    Latitude = toreal(LocationDetails.geoCoordinates.latitude),
    Longitude = toreal(LocationDetails.geoCoordinates.longitude)
| where isnotnull(Latitude) and isnotnull(Longitude)
| order by TimeGenerated asc
| serialize
| extend 
    PrevLatitude = prev(Latitude),
    PrevLongitude = prev(Longitude),
    PrevCity = prev(City),
    PrevCountry = prev(Country),
    PrevTime = prev(TimeGenerated)
| where PrevLatitude != Latitude or PrevLongitude != Longitude
| extend 
    TimeDiffMinutes = datetime_diff('minute', TimeGenerated, PrevTime),
    // Haversine formula for distance
    dLat = radians(Latitude - PrevLatitude),
    dLon = radians(Longitude - PrevLongitude),
    a = sin(dLat/2) * sin(dLat/2) + cos(radians(PrevLatitude)) * cos(radians(Latitude)) * sin(dLon/2) * sin(dLon/2),
    c = 2 * atan2(sqrt(a), sqrt(1-a)),
    DistanceKm = 6371 * c // Earth radius in km
| extend 
    RequiredSpeedKmH = round(DistanceKm / (TimeDiffMinutes / 60.0), 2)
| where RequiredSpeedKmH > MaxTravelSpeedKmH
| project 
    TimeGenerated,
    UserPrincipalName,
    PreviousLocation = strcat(PrevCity, ", ", PrevCountry),
    CurrentLocation = strcat(City, ", ", Country),
    DistanceKm = round(DistanceKm, 2),
    TimeDiffMinutes,
    RequiredSpeedKmH,
    PreviousIP = prev(IPAddress),
    CurrentIP = IPAddress,
    Severity = "Critical"
| order by TimeGenerated desc

5.3 Anomalous Application Access

// Detect access to unusual applications
let TargetAccount = "john.doe@contoso.com";
let TimeRange = 30d;
let BaselinePeriod = 21d;
// Get baseline applications
let BaselineApps = SigninLogs
    | where TimeGenerated between (ago(TimeRange) .. ago(BaselinePeriod))
    | where UserPrincipalName == TargetAccount
    | where ResultType == "0"
    | summarize by AppDisplayName;
// Get recent applications
SigninLogs
| where TimeGenerated > ago(BaselinePeriod)
| where UserPrincipalName == TargetAccount
| where ResultType == "0"
| where AppDisplayName !in (BaselineApps)
| summarize 
    AccessCount = count(),
    IPAddresses = make_set(IPAddress),
    Locations = make_set(LocationDetails.countryOrRegion),
    FirstAccess = min(TimeGenerated),
    LastAccess = max(TimeGenerated)
    by AppDisplayName
| extend Severity = case(
    AppDisplayName contains "Admin" or AppDisplayName contains "Graph", "High",
    "Medium"
)
| order by FirstAccess desc

5.4 Conditional Access Policy Failures

// Analyze Conditional Access policy blocks
let TimeRange = 7d;
SigninLogs
| where TimeGenerated > ago(TimeRange)
| where ConditionalAccessStatus == "failure"
| extend 
    PolicyDetails = parse_json(ConditionalAccessPolicies)
| mvexpand PolicyDetails
| extend 
    PolicyName = tostring(PolicyDetails.displayName),
    PolicyResult = tostring(PolicyDetails.result)
| where PolicyResult == "failure"
| summarize 
    BlockCount = count(),
    Users = make_set(UserPrincipalName),
    IPAddresses = make_set(IPAddress),
    Countries = make_set(LocationDetails.countryOrRegion),
    Applications = make_set(AppDisplayName)
    by PolicyName
| order by BlockCount desc

🎯 Phase 6: Behavioral Analytics (UEBA)

6.1 User Risk Score Analysis

// Analyze user behavior anomalies
let TargetAccount = "john.doe@contoso.com";
let TimeRange = 7d;
BehaviorAnalytics
| where TimeGenerated > ago(TimeRange)
| where UserPrincipalName == TargetAccount or UserName contains TargetAccount
| extend 
    ActivityType = ActivityType,
    RiskScore = InvestigationPriority,
    AnomalyDetails = UsersInsights
| summarize 
    TotalAnomalies = count(),
    HighRiskEvents = countif(InvestigationPriority > 7),
    ActivityTypes = make_set(ActivityType),
    AnomalyReasons = make_set(UsersInsights),
    AvgRiskScore = avg(InvestigationPriority),
    MaxRiskScore = max(InvestigationPriority)
    by UserPrincipalName
| extend Severity = case(
    MaxRiskScore > 8, "Critical",
    MaxRiskScore > 5, "High",
    "Medium"
)

6.2 Peer Group Comparison

// Compare user activity to peer group
let TargetAccount = "john.doe@contoso.com";
let TimeRange = 7d;
// Get target user's department/group
let UserDepartment = IdentityInfo
    | where AccountUPN == TargetAccount
    | summarize arg_max(TimeGenerated, Department)
    | project Department;
// Compare logon patterns
let TargetActivity = DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where AccountName contains TargetAccount
    | summarize TargetLogons = count();
let PeerActivity = IdentityInfo
    | where Department in (UserDepartment)
    | where AccountUPN != TargetAccount
    | join kind=inner (
        DeviceLogonEvents
        | where Timestamp > ago(TimeRange)
    ) on $left.AccountName == $right.AccountName
    | summarize PeerLogons = count() by AccountUPN
    | summarize AvgPeerLogons = avg(PeerLogons), StdDev = stdev(PeerLogons);
TargetActivity
| extend 
    AvgPeerLogons = toscalar(PeerActivity | project AvgPeerLogons),
    StdDev = toscalar(PeerActivity | project StdDev)
| extend 
    DeviationFromPeers = round((TargetLogons - AvgPeerLogons) / StdDev, 2),
    IsAnomaly = case(
        abs((TargetLogons - AvgPeerLogons) / StdDev) > 3, "Significant Anomaly",
        abs((TargetLogons - AvgPeerLogons) / StdDev) > 2, "Moderate Anomaly",
        "Normal"
    )

🔧 Phase 7: Advanced Hunting Techniques

7.1 Multi-Stage Attack Detection

// Detect complete attack chain: Initial Access -> Lateral Movement -> Privilege Escalation
let TimeRange = 24h;
let Stage1_InitialAccess = DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where LogonType == "RemoteInteractive"
    | where ActionType == "LogonSuccess"
    | project 
        Stage = "Initial Access",
        Timestamp,
        DeviceName,
        AccountName,
        RemoteIP;
let Stage2_LateralMovement = DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where LogonType == "Network"
    | where ActionType == "LogonSuccess"
    | project 
        Stage = "Lateral Movement",
        Timestamp,
        DeviceName,
        AccountName,
        RemoteIP;
let Stage3_PrivilegeEscalation = SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4672
    | project 
        Stage = "Privilege Escalation",
        Timestamp = TimeGenerated,
        DeviceName = Computer,
        AccountName = TargetUserName,
        RemoteIP = IpAddress;
// Combine stages
union Stage1_InitialAccess, Stage2_LateralMovement, Stage3_PrivilegeEscalation
| order by AccountName, Timestamp asc
| serialize
| extend 
    NextStage = next(Stage),
    NextTimestamp = next(Timestamp),
    NextDevice = next(DeviceName)
| where AccountName == next(AccountName) // Same account
| extend TimeDelta = datetime_diff('minute', NextTimestamp, Timestamp)
| where TimeDelta < 60 // Within 1 hour
| summarize 
    AttackPath = make_list(Stage),
    Timeline = make_list(Timestamp),
    Devices = make_list(DeviceName),
    SourceIPs = make_set(RemoteIP)
    by AccountName
| where array_length(AttackPath) >= 2
| extend Severity = "Critical"

7.2 Account Reconnaissance and Exploitation

// Detect recon followed by account compromise
let TimeRange = 48h;
// Stage 1: Reconnaissance
let Recon = DeviceProcessEvents
    | where Timestamp > ago(TimeRange)
    | where ProcessCommandLine has_any ("net user", "net group", "whoami", "net localgroup", "dsquery")
    | summarize 
        ReconCommands = make_set(ProcessCommandLine),
        ReconTime = min(Timestamp),
        ReconDevice = any(DeviceName)
        by InitiatingProcessAccountName;
// Stage 2: Account Usage
let AccountUsage = DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where ActionType == "LogonSuccess"
    | summarize 
        LogonTime = min(Timestamp),
        LogonDevice = any(DeviceName)
        by AccountName;
// Correlate
Recon
| join kind=inner (AccountUsage) on $left.InitiatingProcessAccountName == $right.AccountName
| where datetime_diff('hour', LogonTime, ReconTime) between (0 .. 24)
| project 
    AccountName,
    ReconDevice,
    ReconTime,
    ReconCommands,
    LogonDevice,
    LogonTime,
    TimeBetween = datetime_diff('hour', LogonTime, ReconTime),
    Severity = "High"

7.3 Data Exfiltration via Compromised Account

// Detect unusual data access patterns
let TimeRange = 7d;
let BaselinePeriod = 21d;
let TargetAccount = "john.doe@contoso.com";
// Get baseline file access
let Baseline = OfficeActivity
    | where TimeGenerated between (ago(TimeRange + BaselinePeriod) .. ago(TimeRange))
    | where UserId == TargetAccount
    | where Operation in ("FileDownloaded", "FileAccessed", "FileSyncDownloadedFull")
    | summarize BaselineCount = count();
// Get recent file access
let Recent = OfficeActivity
    | where TimeGenerated > ago(TimeRange)
    | where UserId == TargetAccount
    | where Operation in ("FileDownloaded", "FileAccessed", "FileSyncDownloadedFull")
    | summarize 
        RecentCount = count(),
        Files = make_set(OfficeObjectId),
        Operations = make_set(Operation),
        UniqueFiles = dcount(OfficeObjectId)
    | extend BaselineCount = toscalar(Baseline);
Recent
| extend 
    PercentIncrease = round((RecentCount - BaselineCount) * 100.0 / BaselineCount, 2),
    IsAnomaly = case(
        RecentCount > BaselineCount * 3, "Critical",
        RecentCount > BaselineCount * 2, "High",
        "Normal"
    )
| where IsAnomaly != "Normal"

📊 Phase 8: Workbooks and Dashboards

8.1 Account Usage Overview Dashboard (KQL for Workbook)

// Comprehensive account usage metrics
let TimeRange = 7d;
let TargetAccount = "john.doe@contoso.com";
// Logon success rate
let LogonMetrics = union
    (SecurityEvent | where TimeGenerated > ago(TimeRange) | where EventID in (4624, 4625)),
    (DeviceLogonEvents | where Timestamp > ago(TimeRange))
| where AccountName contains TargetAccount or TargetUserName contains TargetAccount
| summarize 
    TotalAttempts = count(),
    Successful = countif(EventID == 4624 or ActionType == "LogonSuccess"),
    Failed = countif(EventID == 4625 or ActionType == "LogonFailed")
| extend SuccessRate = round(Successful * 100.0 / TotalAttempts, 2);
// Geographic distribution
let GeoDistribution = SigninLogs
    | where TimeGenerated > ago(TimeRange)
    | where UserPrincipalName == TargetAccount
    | where ResultType == "0"
    | summarize Count = count() by Country = tostring(LocationDetails.countryOrRegion)
    | top 10 by Count;
// Device distribution
let DeviceDistribution = DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where AccountName contains TargetAccount
    | summarize Count = count() by DeviceName
    | top 10 by Count;
// Hourly activity pattern
let HourlyPattern = union
    (SecurityEvent | where TimeGenerated > ago(TimeRange) | where EventID == 4624),
    (DeviceLogonEvents | where Timestamp > ago(TimeRange) | where ActionType == "LogonSuccess")
| extend Hour = hourofday(coalesce(TimeGenerated, Timestamp))
| summarize Count = count() by Hour
| order by Hour asc;
// Combine all metrics
print Metrics = pack_all(
    LogonMetrics,
    GeoDistribution,
    DeviceDistribution,
    HourlyPattern
)

8.2 Real-Time Monitoring Query

// Live monitoring of account activity
let MonitorAccounts = dynamic(["admin", "svc-", "da-"]); // High-value accounts
union
    (SecurityEvent
    | where TimeGenerated > ago(5m)
    | where EventID in (4624, 4625, 4648, 4672)
    | where TargetUserName has_any (MonitorAccounts)
    | extend Source = "SecurityEvent"),
    (DeviceLogonEvents
    | where Timestamp > ago(5m)
    | where AccountName has_any (MonitorAccounts)
    | extend Source = "DefenderEndpoint"),
    (SigninLogs
    | where TimeGenerated > ago(5m)
    | where UserPrincipalName has_any (MonitorAccounts)
    | extend Source = "AzureAD")
| order by coalesce(TimeGenerated, Timestamp) desc
| take 100

🚀 Phase 9: Automated Response Queries

9.1 Automated Threat Hunting - Scheduled Query

// Run every 15 minutes to detect active compromises
let TimeRange = 15m;
let Threats = union
    // Brute force attempts
    (SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4625
    | summarize FailedAttempts = count() by IpAddress, TargetUserName
    | where FailedAttempts > 10
    | extend ThreatType = "Brute Force", Severity = "High"),
    // Impossible travel
    (SigninLogs
    | where TimeGenerated > ago(TimeRange)
    | where ResultType == "0"
    | extend Country = tostring(LocationDetails.countryOrRegion)
    | order by UserPrincipalName, TimeGenerated asc
    | serialize
    | where UserPrincipalName == prev(UserPrincipalName)
    | where Country != prev(Country)
    | where datetime_diff('minute', TimeGenerated, prev(TimeGenerated)) < 60
    | extend ThreatType = "Impossible Travel", Severity = "Critical"),
    // Lateral movement
    (DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where LogonType == "Network"
    | summarize DeviceCount = dcount(DeviceName) by AccountName
    | where DeviceCount > 5
    | extend ThreatType = "Lateral Movement", Severity = "High"),
    // Privilege escalation
    (SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4672
    | where TargetUserName !endswith "$" // Exclude computer accounts
    | extend ThreatType = "Privilege Escalation", Severity = "High");
Threats
| summarize 
    ThreatCount = count(),
    Details = make_bag(pack_all())
    by ThreatType, Severity
| where ThreatCount > 0

9.2 Incident Creation Query (For Automation Rules)

// Trigger incident when account compromise indicators detected
let TimeRange = 1h;
let CompromiseIndicators = union
    // Failed then successful
    (SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID in (4625, 4624)
    | where TargetUserName !endswith "$"
    | summarize 
        Failed = countif(EventID == 4625),
        Success = countif(EventID == 4624),
        IPs = make_set(IpAddress)
        by TargetUserName
    | where Failed > 10 and Success > 0
    | extend Indicator = "Brute Force Success"),
    // Service account interactive
    (DeviceLogonEvents
    | where Timestamp > ago(TimeRange)
    | where AccountName startswith "svc-" or AccountName startswith "service-"
    | where LogonType == "Interactive"
    | extend Indicator = "Service Account Interactive"),
    // Off-hours admin
    (SecurityEvent
    | where TimeGenerated > ago(TimeRange)
    | where EventID == 4672
    | extend Hour = hourofday(TimeGenerated)
    | where Hour < 6 or Hour > 20
    | extend Indicator = "Off-Hours Admin");
CompromiseIndicators
| summarize 
    Indicators = make_set(Indicator),
    Count = count()
| extend 
    Severity = case(Count >= 3, "High", "Medium"),
    Title = strcat("Account Compromise Detected - ", Count, " indicators"),
    Description = tostring(Indicators)

🎓 Pro Tips and Best Practices

Query Optimization

// ❌ SLOW - Filtering after query
SecurityEvent
| where TimeGenerated > ago(7d)
| where EventID == 4624
| where TargetUserName == "john.doe"

// ✅ FAST - Use FilterHashtable approach
SecurityEvent
| where TimeGenerated > ago(7d) and EventID == 4624 and TargetUserName == "john.doe"

// ✅ FASTER - Put most selective filter first
SecurityEvent
| where TargetUserName == "john.doe" // Most selective
| where TimeGenerated > ago(7d)
| where EventID == 4624

Time Range Best Practices

// ❌ Don't use absolute dates (not reusable)
| where TimeGenerated between (datetime(2025-11-29) .. datetime(2025-11-30))

// ✅ Use relative dates (reusable)
| where TimeGenerated > ago(1d)

// ✅ Use parameters for flexibility
let TimeRange = 7d;
| where TimeGenerated > ago(TimeRange)

Null Handling

// Always handle potential nulls
| extend Country = tostring(LocationDetails.countryOrRegion)
| where isnotnull(Country) and Country != ""

// Use coalesce for fallbacks
| extend AccountName = coalesce(TargetUserName, AccountName, "Unknown")

Summarization Tips

// Use make_set for unique values (no duplicates)
| summarize IPs = make_set(IPAddress) by AccountName

// Use make_list for all values (with duplicates)
| summarize Commands = make_list(ProcessCommandLine) by AccountName

// Use make_bag for key-value pairs
| summarize Details = make_bag(pack("IP", IPAddress, "Time", TimeGenerated))

🔔 Alert Rules (Analytic Rules)

High-Priority Alert: Admin Account Brute Force Success

// Alert when brute force against admin succeeds
let TimeRange = 1h;
let FailedThreshold = 10;
let AdminAccounts = IdentityInfo
    | where GroupMembership has "Admin"
    | summarize by AccountName;
SecurityEvent
| where TimeGenerated > ago(TimeRange)
| where EventID in (4625, 4624)
| where TargetUserName in (AdminAccounts)
| summarize 
    Failed = countif(EventID == 4625),
    Success = countif(EventID == 4624),
    arg_max(TimeGenerated, *)
    by TargetUserName, IpAddress
| where Failed >= FailedThreshold and Success > 0
| extend 
    Severity = "High",
    Tactics = "CredentialAccess",
    Techniques = "T1110"

Alert: Impossible Travel Detected

// Alert on impossible travel (adjust for your org)
let TimeRange = 6h;
let MaxSpeedKmH = 900;
SigninLogs
| where TimeGenerated > ago(TimeRange)
| where ResultType == "0"
| extend 
    Lat = toreal(LocationDetails.geoCoordinates.latitude),
    Lon = toreal(LocationDetails.geoCoordinates.longitude)
| where isnotnull(Lat)
| order by UserPrincipalName, TimeGenerated asc
| serialize
| extend 
    PrevLat = prev(Lat),
    PrevLon = prev(Lon),
    PrevTime = prev(TimeGenerated),
    PrevIP = prev(IPAddress)
| where UserPrincipalName == prev(UserPrincipalName)
| extend 
    TimeDiff = datetime_diff('minute', TimeGenerated, PrevTime),
    dLat = radians(Lat - PrevLat),
    dLon = radians(Lon - PrevLon),
    a = sin(dLat/2) * sin(dLat/2) + cos(radians(PrevLat)) * cos(radians(Lat)) * sin(dLon/2) * sin(dLon/2),
    Distance = 6371 * 2 * atan2(sqrt(a), sqrt(1-a)),
    RequiredSpeed = Distance / (TimeDiff / 60.0)
| where RequiredSpeed > MaxSpeedKmH
| project 
    TimeGenerated,
    UserPrincipalName,
    PreviousLocation = strcat(prev(LocationDetails.city), ", ", prev(LocationDetails.countryOrRegion)),
    CurrentLocation = strcat(LocationDetails.city, ", ", LocationDetails.countryOrRegion),
    DistanceKm = round(Distance, 2),
    TimeDiffMinutes = TimeDiff,
    RequiredSpeedKmH = round(RequiredSpeed, 2),
    PreviousIP = PrevIP,
    CurrentIP = IPAddress
| extend 
    Severity = "High",
    Tactics = "InitialAccess",
    Techniques = "T1078"

📚 Investigation Playbook Cheatsheet

Quick Investigation Steps

Identify Account Type

IdentityInfo | where AccountUPN == "user@domain.com" | project AccountName, AccountDomain, Department, Manager

Get Recent Activity Summary

union DeviceLogonEvents, SigninLogs
| where Timestamp > ago(7d) or TimeGenerated > ago(7d)
| where AccountName == "user" or UserPrincipalName == "user@domain.com"
| summarize count() by bin(coalesce(Timestamp, TimeGenerated), 1h)
| render timechart

Check for Failed Logons

SecurityEvent | where EventID == 4625 | where TargetUserName == "user"
| summarize count() by IpAddress | top 10 by count_

Review Recent Admin Activity

SecurityEvent | where EventID == 4672 | where TargetUserName == "user"
| where TimeGenerated > ago(24h)

Check Lateral Movement

DeviceLogonEvents
| where AccountName == "user"
| where LogonType == "Network"
| summarize DeviceCount = dcount(DeviceName), Devices = make_set(DeviceName)

🔗 Integration with SOAR

Logic App/Playbook Trigger Query

// Query for automated playbook execution
let Compromised = SecurityEvent
    | where TimeGenerated > ago(5m)
    | where EventID == 4625
    | summarize FailedCount = count() by TargetUserName, IpAddress
    | where FailedCount > 20;
Compromised
| extend 
    Action = "DisableAccount",
    Priority = "High",
    AssignedTo = "SOC-L2"

Remember: Always validate queries in a test environment first. Adjust thresholds based on your organisation's baseline. Document all customisations.

PreviousEnhanced Windows Event Log Investigation Guide NextProgram Execution Artifacts Investigation Guide

Last updated 23 hours ago

Microsoft Sentinel & Defender XDR - SOC Analyst Playbook

🎯 Overview

📊 Data Sources Quick Reference

Available Tables

Event ID to Table Mapping

🔍 Phase 1: Initial Triage Queries

1.1 Quick Account Profile (Sentinel)

1.2 Quick Account Profile (Defender XDR)

1.3 Is This Account Under Attack? (Quick Check)

🔐 Phase 2: Authentication Analysis

2.1 Complete Authentication Timeline (Sentinel)

2.2 Failed Logon Analysis - Brute Force Detection

2.3 Successful Logon After Failed Attempts (Successful Breach)

2.4 Logon Type Analysis (Sentinel)

🗺️ Phase 3: Lateral Movement Detection

3.1 Network Logon Chain Analysis (Defender XDR)

3.2 Lateral Movement with Process Correlation

3.3 Administrative Reconnaissance Detection

3.4 Pass-the-Hash Detection (Sentinel)

🚨 Phase 4: Privilege Escalation & Admin Activity

4.1 Detect New Admin Rights Assignments

4.2 Explicit Credential Usage (RunAs)

4.3 Service Account Interactive Logon Detection

🌍 Phase 5: Azure AD / Cloud Authentication Analysis

5.1 Comprehensive Azure AD Sign-in Analysis

5.2 Impossible Travel Detection

5.3 Anomalous Application Access

5.4 Conditional Access Policy Failures

🎯 Phase 6: Behavioral Analytics (UEBA)

6.1 User Risk Score Analysis

6.2 Peer Group Comparison

🔧 Phase 7: Advanced Hunting Techniques

7.1 Multi-Stage Attack Detection

7.2 Account Reconnaissance and Exploitation

7.3 Data Exfiltration via Compromised Account

📊 Phase 8: Workbooks and Dashboards

8.1 Account Usage Overview Dashboard (KQL for Workbook)

8.2 Real-Time Monitoring Query

🚀 Phase 9: Automated Response Queries

9.1 Automated Threat Hunting - Scheduled Query

9.2 Incident Creation Query (For Automation Rules)

🎓 Pro Tips and Best Practices

Query Optimization

Time Range Best Practices

Null Handling

Summarization Tips

🔔 Alert Rules (Analytic Rules)

High-Priority Alert: Admin Account Brute Force Success

Alert: Impossible Travel Detected

📚 Investigation Playbook Cheatsheet

Quick Investigation Steps

🔗 Integration with SOAR

Logic App/Playbook Trigger Query