Identifying Actionable Improvements
What you'll learn: How to transform postmortem findings into concrete, prioritized actions that actually prevent the same incident from happening again.
From "What Went Wrong" to "What We'll Do"
After writing an effective incident narrative (the previous lesson's focus), you have a clear understanding of what broke and why. But a postmortem without action items is just storytelling. Identifying actionable improvements means translating your root cause analysis into specific, measurable tasks that close the gaps revealed by the incident.
What Makes an Improvement "Actionable"?
Think of this like turning a doctor's diagnosis into a treatment plan. "Patient has high blood pressure" isn't actionable—"Prescribe 10mg medication daily and schedule follow-up in 30 days" is.
Good action items are:
- Specific: Not "improve monitoring" but "add alert for queue depth > 1000 on payment-processor service"
- Owned: Assign a person or team responsible for each item
- Time-bound: Set realistic deadlines (avoid vague "someday")
- Preventative: Address root causes, not just symptoms
Prioritization Matters
You'll typically identify 5-15 potential improvements per incident. Not all are equally urgent. Prioritize by:
- Impact: Would this prevent similar high-severity incidents?
- Effort: Quick wins vs. long-term architectural changes
- Error budget alignment: Does this help you meet your SLOs?
Categorize items: "Do immediately," "Schedule this quarter," "Consider for roadmap."
Key Takeaway: Actionable improvements turn postmortem insights into tangible preventative measures—specific, owned, deadline-driven tasks that close the gaps revealed by your incident, prioritized by impact on reliability goals.