Module 11 Activity

Scenario

Something broke in production. This activity turns it into a permanent improvement rather than a fixed bug.

What you build

A timeline with measured gaps, a root-cause test, a blameless review with owned actions, and a scheduled check that the fix held.

Steps

1. Build the incident timeline and measure the gaps: deploy to detection, detection to diagnosis, diagnosis to recovery.
2. Report the metric lag - how long between users noticing and monitoring showing it - and treat it as a finding.
3. Write the test for the ROOT CAUSE, not the symptom. Confirm it would have failed in CI before the change shipped.
4. Run a blameless review asking what made the wrong action look right. End with owned, dated actions.
5. Set a review date four weeks out and check the metric on it.
6. Add the resulting cases to the eval set and report what fraction of your set now comes from real failures.

Evidence to hand in

• The timeline with all three gaps measured.
• The metric lag as a stated finding.
• The root-cause test, and evidence it would have failed pre-deploy.
• The review actions with owners and dates.
• The four-week check.
• The fraction of your eval set that came from incidents.

Review checklist

• The gaps are measured, not narrated.
• Metric lag is treated as its own finding.
• The test targets the root cause and would have failed in CI.
• Every action has an owner and a date.
• The four-week review date was set when the incident closed.

Mark complete