Réaliser des analyses a posteriori sans recherche de coupable

When an incident happens, it's natural to look for causes. A blameless post-mortem is a structured review process designed to learn from incidents without assigning individual blame.

Its core purpose is to understand what happened, why it happened, and how to prevent recurrence, focusing on systemic improvements rather than personal failures.

The term 'blameless' can be misunderstood. It doesn't mean a lack of accountability, but rather a shift in focus:

• From: Who made the mistake?
• To: What in our system, process, or tools allowed this mistake to lead to an incident?

We assume everyone involved acted with the best intentions and did the best they could with the information and tools available at the time.

Beyond just fixing the immediate issue, a blameless post-mortem aims to achieve several critical objectives:

• Understand: Comprehend the incident's full timeline and impact.
• Identify: Uncover contributing factors and root causes.
• Learn: Extract lessons for preventing similar incidents.
• Improve: Generate actionable items for system and process enhancements.
• Communicate: Share findings transparently with stakeholders.

A truly blameless environment requires psychological safety. This means participants feel safe to speak up, admit errors, and share information without fear of punishment or humiliation.

It's crucial for open, honest discussions that lead to genuine learning and not just superficial fixes.

Effective post-mortems start with thorough preparation:

• Gather Data: Collect logs, metrics, alerts, monitoring screenshots, and communication records.
• Timeline: Construct a preliminary timeline of events.
• Participants: Invite key individuals involved in the incident or its resolution, and relevant stakeholders.
• Neutral Facilitator: Designate someone to lead the discussion who wasn't directly involved in the incident.

The facilitator is key to a successful blameless post-mortem. Their responsibilities include:

• Setting the blameless tone at the start.
• Guiding the discussion to stay on track and focused on facts.
• Ensuring all voices are heard and contributions are valued.
• Intervening if blame starts to creep into the conversation.
• Summarizing key points and ensuring clear action items.

A typical post-mortem discussion flow:

1. Incident Overview: Briefly state what happened.
2. Timeline Walk-through: Review events chronologically.
3. What Went Well: Identify successful actions and learnings.
4. What Went Wrong: Discuss failures, surprises, and areas for improvement.
5. Root Cause Analysis: Dig deeper into the 'why'.
6. Action Items: Define concrete next steps.

Keep the discussion focused on the system, not individuals.

To move beyond symptoms, use techniques like the '5 Whys'. This involves repeatedly asking 'Why?' to peel back layers of causality.

Example:

• Problem: Website was slow.
• Why? Database queries were inefficient.
• Why? No proper indexing on a new table.
• Why? Developer forgot to add indexes.
• Why? Code review process missed database schema changes.
• Why? No automated check for index changes in CI/CD pipeline.

The post-mortem's value comes from its outcomes: concrete, actionable items. Each action should be:

• Specific: Clearly defined.
• Measurable: How will we know it's done?
• Assignable: Who is responsible?
• Realistic: Achievable.
• Time-bound: Has a deadline.

Follow up regularly to ensure actions are completed and improvements are implemented.

You've learned that conducting blameless post-mortems is crucial for continuous improvement. Key takeaways:

• Shift focus from individual blame to systemic issues.
• Foster psychological safety for open discussion.
• Prepare thoroughly with data and a neutral facilitator.
• Structure the discussion to uncover root causes (e.g., 5 Whys).
• Generate specific, actionable items for follow-up.

By embracing this approach, your team can turn incidents into powerful learning opportunities.