Skip to content
Tutorials Center
My learning

Topics

Lesson 23 of 37 ยท Incident Response

Post-Incident Review and Lessons Learned

Run a blameless post-incident review: build the timeline, find contributing factors, write actionable follow-ups and track them until they are done.

When the incident is closed it is tempting to move on. That is the moment to do the most valuable work of the whole process: a post-incident review, sometimes called a retrospective, post-mortem or lessons-learned meeting. Its purpose is to make the next incident shorter, cheaper or impossible. Teams that skip it fix the same problems repeatedly.

Make it blameless

A review that hunts for someone to blame produces silence, and silence hides the facts you need. A blameless review assumes that people acted reasonably with the information and tools they had, and asks how the system allowed the failure: why was the alert not seen, why did the process allow that access, why was there no backup? People still own actions and fixes; they are not punished for honest mistakes.

Ask "how" and "what" questions ("How did the alert get missed?") rather than "who" and "why did you" questions. The wording changes the answers.

When and who

Hold the review within a week of closing, while memories are fresh, for every incident of medium severity and above (and for any low-severity one that taught you something). Invite the responders, the owners of the affected systems, and someone from management who can approve changes. Keep it to about an hour with a facilitator who does not argue their own case.

Agenda

  1. Summary: what happened, in three sentences.
  2. Timeline: from first attacker action to closure, with the key decisions.
  3. What went well: detections that worked, fast decisions, good tools.
  4. What went badly: gaps, delays, confusion.
  5. Contributing factors: root and secondary causes.
  6. Actions: specific changes, each with an owner and a due date.

A good review answers

What happened? Facts and timeline
Why could it? Contributing factors
What changes? Owned, dated actions
How will we know? A measure of success

A review without actions is just a story.

1 / 2

Finding contributing factors

Incidents rarely have a single cause. Look at each stage of the attack and ask what could have stopped it earlier:

  • Prevention: was a patch missing, a setting weak, an account over-privileged?
  • Detection: did a rule exist? Did it fire? Was it noticed, or lost in noise?
  • Response: were the playbooks clear, the contacts correct, the tools reachable?
  • Recovery: were backups current and tested?

The "five whys" technique helps: keep asking why until you reach something you can change. Example: the attacker used a stolen password. Why did it work? No MFA on that portal. Why? The portal was added outside the standard process. Why? Nobody owns the inventory of external portals. The fix is an owner and an inventory, not a stricter lecture about passwords.

flowchart LR
    A["Stolen password worked"] --> B["No MFA on portal"]
    B --> C["Portal added outside standard process"]
    C --> D["No owner for the external-service inventory"]
    D --> E["Action: assign owner, build inventory, require MFA"]

Five whys: stop when the answer is something you can fix.

Writing actions that get done

A weak action: "Improve monitoring." A strong action: "Add a detection for service accounts logging in interactively (owner: Priya, due 30 June, test: replay the 10 May logs and confirm the alert fires)."

Good actions are specific, have one owner, a due date, and a way to verify. Rank them by impact and effort, and do not accept more than the team can finish: ten unfinished actions are worse than three finished ones.

Try it: track the follow-up actions

A review is only useful if the actions get closed. This stand-in lists open actions and flags overdue ones.

Python Action tracker

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Measuring the incident

Record numbers you can compare across incidents: time from attacker's first action to detection, from detection to containment, to full recovery, plus the number of systems and accounts affected and the business impact. These become your SOC metrics.

Python Detection and response intervals

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

A long gap between the first action and detection suggests detection work; a long gap between detection and containment suggests response-process work. The numbers show where to invest.

Feed improvements back

Turn findings into changes in four places:

  • Detections: new or tuned rules (alert tuning), replaying the incident's logs as a test.
  • Playbooks: clearer steps, correct contacts (playbook design).
  • Controls: patching, MFA, segmentation, backups.
  • People: training and exercises built on the real incident, with sensitive details removed.

Share a short, sanitized summary with the wider organization. People trust a security team that explains what it learned.

Common mistakes

  • Holding the review weeks later, when nobody remembers details.
  • Turning it into a hunt for blame.
  • Producing a long report that no one reads; one page with actions beats twenty pages of narrative.
  • Never checking whether the actions were done.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What does a blameless review focus on?

    Blamelessness keeps people honest so the real contributing factors surface.
  2. 2. Which is a well-formed follow-up action?

    Good actions are specific, owned, dated and verifiable.
  3. 3. What is the purpose of the 'five whys' technique?

    Repeated why-questions move from the symptom to a systemic cause you can change.
  4. 4. A long gap between the attacker's first action and detection points to which area of improvement?

    Dwell time before detection is mainly improved by better logging and detections.

Summary

  • Review every significant incident within a week, in a blameless setting.
  • Build the timeline, list what went well and badly, and find contributing factors with "five whys".
  • Write actions that are specific, owned, dated and testable, and track them to completion.
  • Measure the intervals, and feed the results back into detections, playbooks, controls and training.

Next: Threat Intelligence Fundamentals