Incident Response Plan

Incident Response Planning

Build a comprehensive incident response program with clear roles, procedures, severity classifications, and communication protocols to minimize mean time to detection (MTTD) and mean time to resolution (MTTR).

Context

You are building or improving an incident response (IR) program. A strong IR program doesn't prevent all incidents, but it dramatically reduces their impact. Organizations without plans suffer: longer detection time, chaotic response, poor communication, and higher customer impact. An effective plan is practiced, documented, and clear on roles.

IR is not a one-time document; it's a living program that improves after each incident through post-mortems and tabletop exercises.

Domain Context

• Incident Categories: Malware, phishing, data breach, ransomware, DDoS, account compromise, insider threat, supply chain attack, compliance violation
• Response Phases: Detection → Triage → Containment → Investigation → Recovery → Post-Incident Review
• Incident Classification: Severity (Critical, High, Medium, Low) drives response speed and escalation
• MTTD (Mean Time to Detect): How quickly is an incident discovered? Goal: <1 hour for critical
• MTTR (Mean Time to Resolve): How quickly is an incident resolved? Goal: <4 hours for critical
• Communication: Incident commander coordinates response; clear escalation ensures executive/customer awareness
• Evidence preservation: Don't destroy evidence in haste to recover; forensics may be needed

When to Use This Skill

• You're building a security operations center (SOC) or IR program
• Your team has experienced incidents and needs a structured process
• Customers or auditors require an incident response plan
• You're conducting tabletop exercises to test readiness
• You're updating IR procedures after an incident

Prerequisites

Before building the plan, identify:

1. Incident types relevant to your business: What threats matter most? (Data breach for SaaS; ransomware for healthcare; DDoS for financial)
2. Stakeholders: Who needs to be involved? (Security, engineering, ops, legal, communications, finance, executive sponsor)
3. Current detection capabilities: What alerts do you have? (SIEM, intrusion detection, log monitoring, user reports)
4. Infrastructure/data: What systems are critical? Where is sensitive data stored? What's the business impact if compromised?
5. Regulatory requirements: What notification timelines are required? (GDPR 72 hours, state laws vary, contracts may require notification)

Instructions

1. Define Incident Classification & Escalation

Create clear criteria for each severity level:

2. Establish the Incident Response Team

Define roles and responsibilities:

• Incident Commander: Leads response; makes decisions on containment/recovery; coordinates teams; updates executives
• Security Engineer: Investigates; collects evidence; determines scope and duration of compromise
• Infrastructure Engineer: Executes containment (isolate systems, kill processes, disable accounts); restores from backup
• Communications Officer: Notifies executives, customers, authorities; manages public narrative
• Privacy Officer: Determines if breach occurred; determines notification obligations; coordinates legal
• Executive Sponsor: Authorized to approve actions (take systems offline, spend emergency funds, authorize paid cleanup)
• Database Administrator: Restores databases; validates data integrity post-recovery
• Network Engineer: Isolates network segments; analyzes packet captures; implements controls

Document: On-call rotation; primary/backup for each role; contact list (phone + email); escalation path

3. Design Detection & Alerting

Establish alert standards:

• Detection sources: SIEM alerts, IDS/IPS, unusual access patterns, user reports, threat intelligence, application logs
• Alert standards: Clear criteria for what triggers escalation (10 failed logins in 5 min → lock account; unusual data access → investigate)
• Initial triage: Incident commander assesses severity; assigns response team; initiates IR process
• Validation: Not all alerts are real incidents; process to quickly validate before full escalation
• Notification: Alert SOC/on-call; page incident commander; initiate IR war room (Slack channel or bridge line)

Example alert thresholds:

• 10 failed login attempts within 5 minutes → potential brute-force attack
• Data exfiltration >1 GB from single user → potential breach
• Privilege escalation (user gains admin rights) → immediate investigation
• Database dump command executed → potential data theft

4. Containment & Investigation Procedures

Step-by-step actions:

1. Preserve Evidence (CRITICAL):

• Don't shut down compromised systems immediately; preserve logs/memory
• Collect volatile data (open connections, running processes, memory state)
• Forensic imaging of affected systems

2. Isolate Compromised Systems:

• Disconnect from network (physically or virtually)
• Disable affected accounts
• Revoke API keys and access tokens
• If worm detected, isolate entire subnet

3. Investigation:

• Review logs (application, system, network) for:
  • When did attacker first gain access?
  • What systems did they touch?
  • What data was accessed or exfiltrated?
  • How did they get in? (phishing, vuln, weak credentials, supply chain?)
• Determine scope (how many systems, how much data, how many customers?)
• Timeline: When was first unusual activity? When was it detected? Why the gap?

4. Threat Analysis:

• Is attacker still active? Check for persistence mechanisms (backdoors, changed passwords, new accounts)
• Is this ransomware? Check for file extensions, ransom notes, encryption behavior
• Is this data exfiltration? Estimate data volume and sensitivity

5. Decision on Recovery:

• Full system rebuild (safest): Wipe and restore from clean backup
• Surgical removal (if possible): Remove malware without rebuilding; only if high confidence
• Patch-and-harden: If threat was a vulnerability, patch and restart

5. Recovery & Communication

Recovery:

• Restore systems from clean backups (not infected backups)
• Patch vulnerability that was exploited
• Test systems post-recovery; validate backups actually work
• Hardening: Re-enable MFA, rotate credentials, tighten firewall rules

Communication (parallel with recovery):

• Internal: Status updates every 4 hours; all-hands standup daily during incident
• Customer notification:
  • If data breach: Notify per regulatory timeline (GDPR 72h, some states 48h)
  • Message: What happened? What data was affected? What we're doing. What customers should do.
• Law enforcement: If ransomware or sophisticated attack, consider FBI/CISA notification
• Auditors/Compliance: Notify if required by contracts or regulations

Example customer notification:

"We identified unauthorized access to our systems on [date]. Our investigation shows [X records] were accessed. We have no evidence of misuse. We have contained the access and restored systems. You should [monitor credit/change passwords]. We provide [X months free monitoring]. Contact [support email] for questions."

6. Post-Incident Review (Blameless)

Within 2 weeks of resolution:

• Timeline: Recreate incident timeline (when discovered, when escalated, when contained)
• Root cause: Why did this happen? Was it a missing control, process failure, or technical flaw?
• Impact: How many customers? How much data? Revenue impact? Reputation impact?
• What went well: What response activities worked? What team coordination was effective?
• What didn't go well: Where were delays? Confusion? Missing information?
• Action items: What will we change to prevent recurrence or reduce impact next time?
  • Technical fixes (patch vulnerability, add monitoring, implement control)
  • Process fixes (update runbook, add alert, improve documentation)
  • Training (brief team on new detection method)

Important: Blameless postmortem—focus on systems and processes, not individual failures. Goal is learning, not punishment.

7. Training & Tabletop Exercises

Annual training: All staff receives IR training (30-60 min)

• What triggers an incident
• Who to contact
• What not to do (don't delete logs, don't rush to shut down)

Semi-annual tabletop: Simulate incident scenario

• Scenario (e.g., "Ransomware detected on 5 servers at 10am Friday")
• Teams respond as if real (who escalates, what's the timeline, how do we communicate?)
• Debrief: What went well? What's our bottleneck?

Output Format

An incident response plan (living document):

# Incident Response Plan

## Incident Classification & Escalation
[Table with severity levels, MTTD targets, escalation procedures]

## Incident Response Team
[Roles, responsibilities, on-call schedule, contact list]

## Detection & Alerting
[Alert standards, triage procedure, initial response]

## Containment Runbook
[Step-by-step procedures for: preserve evidence, isolate, investigate, decide recovery]

## Recovery Procedures
[System restore, patching, validation, hardening]

## Communication Template
[Internal status update, customer notification, law enforcement contact]

## Post-Incident Review Process
[Timeline recreation, root cause analysis, action items, blameless culture]

## Training Schedule
[Annual training, tabletop scenarios, refresher cadence]

## Appendix
- Escalation Contact List
- System criticality (which systems are critical?)
- Data classification (what's PII, what's financial?)
- Vendor/third-party incident contacts

Worked Example

Incident Response Plan: SaaS Analytics Platform

Critical Incident Scenario Response

Scenario: Ransomware detected on database server at 9:47am Friday. Attacker demands payment; some customer data encrypted.

Response Timeline:

• 9:47am: Alert fires (file extension .locked detected on 500+ files in database directory)
• 9:52am: SOC escalates to incident commander (Page Robert, VP Security)
• 10:00am: War room established (Slack: #incident, Zoom bridge)
  • Robert (IC): Decide whether to shut down
  • Database team: Assess which databases encrypted; check backup integrity
  • Security team: Investigate how attacker got in; check for persistence
• 10:15am: Decision: No evidence of data exfiltration yet; shutdown database to preserve evidence
• 10:30am: Backup integrity confirmed; we can restore from 4 hours ago (12am backup)
• 10:45am: Isolation: Database server disconnected from network; attacker can't move laterally
• 11:00am: Executive briefing: Ransomware contained; no customer data lost (4-hour recovery point); database recovery in progress
• 11:30am: Root cause analysis starts: SSH logs show attacker used compromised credentials (employee laptop with old password)
• 1:00pm: Database restore begins; estimated 3-4 hours to completion
• 2:00pm: Customer communication prepared (draft): "We detected and contained ransomware affecting our infrastructure. Our investigation shows no customer data was accessed or exfiltrated..."
• 5:00pm: Database restore complete; testing shows no data corruption
• 6:00pm: Customer notification sent; support team ready for calls
• Monday: Post-incident review; action items (mandatory password reset, MFA enforcement, improve database backup testing)

Decision Framework

When responding to an incident:

• If unsure about severity: Err on the side of higher severity. It's easier to downgrade than deal with an undetected critical incident.
• If deciding between containment speed and evidence preservation: In general, preserve evidence if incident is contained. Forensics may be needed for legal/insurance.
• If customer notification timing is unclear: Consult your legal/compliance team immediately. Notification windows are strict (GDPR 72h).
• If a system is compromised and you don't know why: Don't assume it's patched; assume there's a persistence mechanism. Full rebuild is safer.
• If incident commander is overwhelmed: Assign a second commander or rotate; don't leave one person making all decisions under stress.

Anti-Patterns & Guards

Anti-Pattern 1: Untested Plan

Description: IR plan written once; never tested. When real incident happens, it fails.

Guard: Tabletop exercises at least semi-annually. Test the critical path (who escalates, how fast).

Anti-Pattern 2: No Clear Incident Commander

Description: During incident, everyone has opinions; no one is making decisions.

Guard: Declare incident commander explicitly. They have final say (not consensus).

Anti-Pattern 3: Destroying Evidence Too Fast

Description: Shutting down systems before forensics; losing ability to determine attack vector.

Guard: Preserve before recovery. Forensic imaging takes <1 hour; it's worth it.

Anti-Pattern 4: Poor Communication

Description: Customers learn about breach from news; internal teams don't coordinate.

Guard: Prepare communication templates in advance. Designate communications officer.

Anti-Pattern 5: Not Learning from Incidents

Description: Post-mortems written; recommendations ignored. Same incident repeats.

Guard: Assign action items; track through completion. Revisit in next tabletop.

Quality Checklist

Before declaring IR program ready:

• Incident classification is clear: Team can quickly assess severity
• Roles are defined: Every IR function has an owner and backup
• Detection is comprehensive: Multiple alert sources; not just SIEM
• Containment procedures are specific: Step-by-step, not vague
• Communication templates exist: Prepared language for customers/executives
• Team is trained: Everyone knows their role; annual training scheduled
• Plan is tested: Tabletop exercises at least semi-annually
• Post-mortems are blameless: Focus on systems, not individuals
• Regulations understood: Know notification timelines and requirements

Further Reading

• NIST SP 800-61 (Incident Handling Guide): https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-61r2.pdf
• SANS Incident Handler's Handbook: https://www.sans.org/incident-response/
• CISA Incident Response: https://www.cisa.gov/incident-response
• PagerDuty Incident Response: https://www.pagerduty.com/incident-response/
• Google Site Reliability Engineering (SRE): Chapter on incident response and blameless postmortems