The Detection Lifecycle
Every detection rule goes through a lifecycle from initial concept to eventual retirement. Understanding this lifecycle is crucial for building an effective detection program.
Phase 1: Threat Intelligence
Before writing any detection, you need to understand what you're detecting. This comes from:
- Threat Reports: APT reports, malware analysis, incident writeups
- ATT&CK Framework: Understanding adversary tactics and techniques
- Internal Incidents: Learning from attacks on your own organization
- Industry Sharing: ISACs, threat feeds, peer sharing
Key Question: What specific behavior are we trying to detect?
Phase 2: Detection Design
With a clear understanding of the threat, design your detection:
- Define the Behavior: What specific actions indicate this threat?
- Identify Data Sources: What logs contain the evidence?
- Consider Evasion: How might attackers avoid detection?
- Plan for False Positives: What legitimate activity looks similar?
Key Question: What's the minimum set of conditions that reliably identifies this threat?
Phase 3: Development
Write the actual detection rule:
- Choose the right format (SIGMA, native SIEM query, etc.)
- Implement the detection logic
- Write clear documentation
- Create test cases
Key Question: Does this rule accurately capture the threat behavior?
Phase 4: Testing
Before production, validate your detection:
- Unit Testing: Does the logic work correctly?
- Lab Testing: Does it fire on simulated attacks?
- Baseline Testing: What's the expected alert volume?
Key Question: Will this detection work reliably in production?
Phase 5: Deployment
Push to production with care:
- Stage in a test environment first
- Monitor initial alert volume
- Have a rollback plan
- Communicate with the SOC
Key Question: Is the SOC ready to handle alerts from this detection?
Phase 6: Monitoring & Tuning
Once live, continuously improve:
- Track true/false positive rates
- Gather analyst feedback
- Refine detection logic
- Update documentation
Key Question: Is this detection providing value or creating noise?
Phase 7: Retirement
Eventually, detections become obsolete:
- Threat no longer relevant
- Better detection available
- Data source deprecated
Don't let stale rules clutter your environment.