A Windows security alert usually takes 15 to 30 minutes to check properly, if you look at the right things in the right order. This guide is for the IT person at a small business who gets the alert, with no security team to hand it to.
The alert itself is a starting point, not a verdict. Microsoft Defender says what it saw. It rarely tells you how the file got there, who was logged in, or whether anything else changed. Those answers are already on the machine, in Windows event logs that most people never open.
Below are five checks, in order. Each one gives the exact log and event ID to read and a PowerShell command to run (as Administrator). At the end there’s a real example from our own test lab, and a one-page reference table.
New to alert triage? Our evidence-based guide to triaging security alerts covers the basics this article builds on.
Step 1: Read what Defender actually recorded
Start with Defender’s own log, not the pop-up. The pop-up is a summary. The log has the file path, the threat name, the user account and what Defender did about it.
The log is Microsoft-Windows-Windows Defender/Operational. Three event IDs matter most:
- 1116: malware or unwanted software was detected.
- 1117: Defender took action (quarantined, removed or blocked it).
- 5001: real-time protection was turned off. If you see this near the alert, treat it as serious. Someone or something may have disabled protection on purpose.
Microsoft lists every Defender event ID in its official event ID reference.
Get-WinEvent -FilterHashtable @{
LogName = 'Microsoft-Windows-Windows Defender/Operational'
Id = 1116, 1117, 5001
} -MaxEvents 20 | Format-List TimeCreated, Id, Message
Defender also keeps its own detection history:
Get-MpThreatDetection | Format-List InitialDetectionTime, ProcessName, Resources, ActionSuccess
Write down three things: the exact time, the full file path, and the user account. Every later step searches around that time.
A 1116 followed by a successful 1117 means Defender caught it. It does not mean the machine is clean. The file came from somewhere, and that is what the next steps find.
For a deeper walk-through of Defender detections, see how to investigate antivirus alerts.
Step 2: Check the file and the process that started it
The question here is simple: what launched this file? A file that Outlook or a browser saved is a different story from one that PowerShell started at 3 a.m.
If the file is still on disk (it may already be in quarantine), get its hash and signature:
Get-FileHash 'C:\path\from\step1.exe' -Algorithm SHA256
Get-AuthenticodeSignature 'C:\path\from\step1.exe' | Format-List Status, SignerCertificate
Search the SHA-256 hash on VirusTotal. A signed file from a known vendor is not automatically safe, but an unsigned file in a user’s AppData or Temp folder deserves a closer look.
Sysmon is a free Microsoft Sysinternals tool that records far more detail than Windows does by default.
Then find the parent process. If Sysmon is installed, its event 1 (process creation) records the full command line, the parent process and the file hashes:
Get-WinEvent -FilterHashtable @{
LogName = 'Microsoft-Windows-Sysmon/Operational'
Id = 1
StartTime = (Get-Date).AddHours(-2)
} | Where-Object { $_.Message -match 'step1.exe' } |
Format-List TimeCreated, Message
No Sysmon? Windows records Security event 4688 for new processes, but only if process-creation auditing is turned on, and it’s off by default. Turning it on (and installing Sysmon) before you need it is one of the cheapest things a small team can do.
Red flags: Word, Excel or Outlook starting powershell.exe or cmd.exe; PowerShell with -EncodedCommand or -WindowStyle Hidden; anything running from Temp. Our guide on investigating suspicious processes goes deeper on this.
Step 3: Check who logged in, and how
Malware often arrives through a login rather than a download: a guessed RDP password, a reused account, a stolen VPN credential. The Security log shows every one.
- 4624: successful logon.
- 4625: failed logon. One or two are normal typos. Dozens in a few minutes from one address is someone guessing.
Each logon event has a Logon Type. Two matter most on a small network:
- Type 3 (network): file shares, SMB, many remote tools.
- Type 10 (remote interactive): Remote Desktop. A type 10 logon from an address you don’t recognise is the single most important thing to catch.
Microsoft documents every logon type on its pages for event 4624 and event 4625.
# Failed logons in the last 24 hours, grouped by source address
Get-WinEvent -FilterHashtable @{
LogName = 'Security'
Id = 4625
StartTime = (Get-Date).AddHours(-24)
} | ForEach-Object {
([xml]$_.ToXml()).Event.EventData.Data |
Where-Object Name -eq 'IpAddress' | Select-Object -ExpandProperty '#text'
} | Group-Object | Sort-Object Count -Descending
Then look at the successful logons (4624) in the hour before the alert. The pattern to worry about: many failures from one address, followed by a success from that same address.
Step 4: Look for anything set up to come back
Attackers who get in usually leave a way back: a new account, a service, a scheduled task or a startup entry. Defender removing one file does not remove these.
| What to look for | Where | Event ID |
|---|---|---|
| New user account | Security log | 4720 |
| User added to a local group (e.g. Administrators) | Security log | 4732 |
| New service installed | System log | 7045 |
| New scheduled task | Security log (only if object-access auditing is on) | 4698 |
| Security log cleared | Security log | 1102 |
A 1102 (log cleared) on a small-business PC is almost never innocent. Treat it as a reason to escalate. See Microsoft’s page on event 1102.
Scheduled tasks and startup entries can be checked directly, whatever auditing is set to:
# Tasks outside Microsoft's own folders
Get-ScheduledTask | Where-Object { $_.TaskPath -notlike '\Microsoft\*' } |
Select-Object TaskName, TaskPath, State, @{n='Action';e={$_.Actions.Execute}}
# Programs set to start at logon
Get-ItemProperty 'HKLM:\Software\Microsoft\Windows\CurrentVersion\Run',
'HKCU:\Software\Microsoft\Windows\CurrentVersion\Run'
Anything pointing into AppData, Temp or ProgramData with a random-looking name goes on your list.
Step 5: Check what the machine is talking to
Malware that stays usually phones home. Check live connections with Get-NetTCPConnection and match each one to the program that owns it:
Get-NetTCPConnection -State Established |
Where-Object RemoteAddress -notmatch '^(127\.|::1|10\.|192\.168\.)' |
Select-Object RemoteAddress, RemotePort,
@{n='Process';e={(Get-Process -Id $_.OwningProcess).Path}} |
Sort-Object Process
Browsers, Teams and Windows Update will fill most of the list. You’re looking for a program you don’t recognise, or one running from a user folder, holding an outside connection open.
For the past, not just right now:
- Sysmon event 3 records network connections per process.
- Sysmon event 22 records DNS lookups. Odd, random-looking domain names stand out quickly.
- The Windows Firewall log (
%SystemRoot%\System32\LogFiles\Firewall\pfirewall.log) shows blocked and allowed connections, but only if logging was switched on beforehand. It’s off by default.
A real example from our test lab
In October 2026 we tested ForenClarity against a Windows 10 Pro machine on our own lab network, attacking it from a Kali Linux virtual machine. One of the tests was a Remote Desktop password-guessing attack: the attacker tried to log in as Administrator over RDP using a short list of 20 wrong passwords.
What Windows recorded: a run of 4625 (failed logon) events, all with Logon Type 10 (Remote Desktop) and all from the same source address, seconds apart. No 4624 success followed, because none of the passwords were right.
What a person would have seen without a tool: nothing. Windows doesn’t pop up a warning for failed logons. The evidence was in the Security log, but only for someone who went looking.
What ForenClarity showed: one incident for that attacker’s address, not 20 separate alerts. It said in plain language who was trying, which account they targeted, and that the attempts came over Remote Desktop. The incident has a manual Block this port button. It adds one Windows Firewall rule for that single address on port 3389, not a blanket closure for everyone, and nothing is blocked until the administrator clicks it. We tested both the block and the restore against a real attacking machine.
The lesson for any small team, with or without our product: the evidence for most attacks is already sitting in the event logs. The hard part is noticing it and connecting the events.
Decide: false alarm, contained, or escalate
After the five checks, the alert falls into one of three outcomes.
| Outcome | What you found | What to do |
|---|---|---|
| False alarm | Known, signed software; expected parent process; no odd logons, persistence or connections | Note it, and add an exclusion only if it keeps recurring and you’re sure |
| Caught and contained | Real malware, Defender removed it (1117), it came from a download or attachment, steps 3 to 5 are clean | Find how it arrived, warn the user, run a full scan, keep your notes |
| Escalate | Any of: 5001 or 1102 near the alert, a type 10 logon you can’t explain, a new admin account, unknown services or tasks, unknown outside connections | Disconnect the network cable or Wi-Fi but don’t shut the machine down (memory holds evidence). Change passwords from a different, clean machine. Call a professional incident-response provider. |
If you’re unsure, treat it as an escalation. A false alarm costs an hour; a missed intrusion costs much more. For more on why tools sometimes disagree, see why security tools disagree and false positives happen.
This guide helps you investigate. It does not replace antivirus, EDR, a SIEM or professional incident response.
Quick reference
| Event ID | Log | Means |
|---|---|---|
| 1116 | Defender/Operational | Malware detected |
| 1117 | Defender/Operational | Defender took action |
| 5001 | Defender/Operational | Real-time protection turned off |
| 4624 | Security | Successful logon (check Logon Type) |
| 4625 | Security | Failed logon |
| 4688 | Security | New process (if auditing is on) |
| 4720 | Security | User account created |
| 4732 | Security | User added to a local group |
| 4698 | Security | Scheduled task created (if auditing is on) |
| 1102 | Security | Security log cleared |
| 7045 | System | New service installed |
| 1 / 3 / 22 | Sysmon/Operational | Process start / network connection / DNS lookup |
FAQ
Defender says the threat was removed. Am I done?
Not yet. Removal deals with one file. Steps 3 to 5 tell you whether it was the whole story.
Do I need Sysmon?
You can do steps 1, 3 and 4 without it. Sysmon makes steps 2 and 5 far easier. It’s free from Microsoft Sysinternals.
Should I just reinstall Windows?
If you escalated, wait for the incident-response provider first. A reinstall destroys the evidence of how the attacker got in, and they may simply come back the same way.
About ForenClarity
ForenClarity is an evidence-review helper for small IT teams that don’t have a dedicated security specialist. It reads the Windows evidence described above and explains incidents in plain language. It’s currently in development for Windows. To join the pilot, contact us.
References
- Microsoft Defender Antivirus event IDs and error codes (Microsoft Learn)
- Event 4624: An account was successfully logged on (Microsoft Learn)
- Event 4625: An account failed to log on (Microsoft Learn)
- Event 1102: The audit log was cleared (Microsoft Learn)
- Sysmon (Microsoft Sysinternals)
- Get-NetTCPConnection (Microsoft Learn)
- VirusTotal (file hash lookup)
Related: How to triage security alerts: an evidence-based guide.