Write a Blameless Post-Mortem from Your Incident or Project Failure Timeline

Paste your incident notes and get a structured blameless post-mortem with a timeline, a five-whys root cause analysis and preventive actions owned by roles, not people.

You getSummary, impact and timeline

Prompt

Prompt
You are an experienced incident review facilitator who runs blameless post-mortems for engineering, operations and project teams.

I will paste my notes about an incident or failed project right after this prompt: timeline fragments, chat logs, metrics or a rough summary. Turn them into a blameless post-mortem.

Rules:
1. Build the timeline in chronological order with timestamps. Mark any time you inferred with "(approx.)".
2. Describe actions by role or system, never by a person's name, and never name a person as the root cause. If my notes blame someone, reframe it as the condition that made the error possible.
3. Run five whys on the main failure. Each why must cite a fact from my notes. Stop early and say so if the notes do not support the next step.
4. Every preventive action has an owner role, a due window, and a type: Prevent, Detect or Mitigate.
5. Do not invent metrics, customer counts or quotes. If impact numbers are missing, write "Unknown - confirm".

Output format:
1. Summary (3 sentences: what happened, impact, current status)
2. Impact
3. Timeline (table: Time | Event | Source)
4. Root cause analysis (numbered five whys)
5. Contributing factors
6. What went well
7. Preventive actions (table: Action | Type | Owner role | Due | Priority)
8. Open questions

End with one line naming the biggest gaps in my notes the team should fill before sharing this.

Example output

At a glanceCheckout outage timeline, 12 March47-minute outage from config deploy to revert
  1. Config deploy14:02Gateway timeout lowered from 30s to 3s
  2. Errors spike14:09Error rate on /checkout rises to 38%
  3. Complaints reported14:21 (approx.)Support lead flags customer complaints in #ops
  4. On-call paged14:26On-call engineer paged via PagerDuty
  5. Cause identified14:41Timeout change identified as likely cause
  6. Config reverted14:49Errors drop to baseline

Example for: a 47-minute checkout outage at a mid-size online retailer after a config change, notes pasted from Slack.

1. Summary

On 12 March, checkout requests failed for 47 minutes after a payment gateway timeout setting was changed during a routine deploy. Roughly 1,900 checkout attempts returned errors. Service is fully restored and the setting has been reverted.

2. Impact

  • Failed checkouts: ~1,900 (from error dashboard)
  • Lost revenue: Unknown - confirm with finance
  • Support tickets: 64

3. Timeline

TimeEventSource
14:02Config deploy lowers gateway timeout from 30s to 3sDeploy log
14:09Error rate on /checkout rises to 38%Dashboard
14:21 (approx.)Support lead reports customer complaints in #opsSlack
14:26On-call engineer pagedPagerDuty
14:41Timeout change identified as likely causeSlack
14:49Config reverted; errors drop to baselineDeploy log

4. Root Cause Analysis

  1. Why did checkouts fail? Gateway calls timed out after 3 seconds.
  2. Why was the timeout 3 seconds? A config change meant for the search service was applied to the payments config file.
  3. Why did it reach payments? Both services share one config file with similar key names.
  4. Why wasn't it caught in review? The diff showed only the value change, not which service block it belonged to.
  5. Why did detection take 17 minutes? No alert existed on checkout error rate; paging relied on manual reports.

5. Contributing Factors

  • Deploy happened during peak afternoon traffic
  • No staging test covers gateway latency

6. What Went Well

  • Revert took under 10 minutes once the cause was found
  • Support escalated quickly with clear customer examples

7. Preventive Actions

ActionTypeOwner roleDuePriority
Split shared config into per-service filesPreventPlatform lead2 weeksHigh
Add alert at 5% checkout error rateDetectSRE on-call owner1 weekHigh
Add latency test for gateway in stagingPreventQA lead1 monthMedium
Document one-click config rollbackMitigatePlatform lead2 weeksMedium

8. Open Questions

  • Were any orders charged but not confirmed?
  • Is peak-hour deploy freeze policy needed?

Gaps to fill: exact revenue impact, and confirmation of when the first customer complaint arrived.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Forcing role-based language and reframing blame into conditions keeps the review blameless and focused on systems, not people.
  • Requiring each 'why' to cite a fact from your notes stops the model from guessing a tidy but invented root cause.
  • Tagging actions as Prevent, Detect or Mitigate shows whether you fixed the cause or only made the next incident shorter.

Use within a few days of an outage, missed launch or failed project. Have your Slack thread, alert times and rough notes ready to paste into ChatGPT, Claude or Gemini, then review the draft with the team before circulating.

Common edits

  • Ask for a one-paragraph executive summary version to share with leadership.
  • Replace the five whys with a fishbone breakdown for complex project failures with many causes.

Prompt details

By
Prompts-GPT editors
Published

Did this prompt work?

Keep exploring

More in Business