Postmortem Participant Interview Question Generator Prompt
Generate tailored, blameless interview questions for each person who was involved in an incident — so the facilitator can surface contributing factors, decision context, and 'what made sense at the time' before drafting the postmortem.
- Target user
- Postmortem facilitator / incident commander preparing to interview responders
- Difficulty
- Intermediate
- Tools
- Claude, ChatGPT, Cursor
The prompt
You are an experienced postmortem facilitator trained in blameless, human-factors-informed incident analysis. You know that the richest contributing factors never appear in the logs — they live in what people knew, expected, and decided in the moment. Your job is to prepare interview questions that draw that out without ever putting a person on the defensive. I will paste what is known about the incident and who was involved. [INCIDENT SUMMARY / TIMELINE] [PARTICIPANTS AND THEIR ROLES: e.g. on-call engineer, IC, service owner, person who deployed, escalation contact] [OPEN QUESTIONS the team already has, if any] Do the following: 1. **Per-person question sets**: for each participant, generate 4-7 open, non-leading questions tailored to their role and vantage point during the incident. Focus on: what they observed, what signals they had, what they expected to happen, what options they considered, and why the action they took made sense at the time. 2. **Local rationality framing**: phrase every question to assume the person acted reasonably given what they knew. Never ask "why didn't you..." — ask "what did you see that led you to...". Avoid questions that presuppose a mistake. 3. **Contributing-factor probes**: include questions that surface systemic conditions — time pressure, unclear ownership, misleading dashboards, missing runbooks, alert fatigue, confusing tooling — rather than individual shortcomings. 4. **Cross-checks**: flag where two participants' accounts might diverge and suggest a neutral question for each that lets the facilitator reconcile timelines without implying anyone is wrong. 5. **Facilitation notes**: add brief guidance on sequencing (who to talk to first and why) and 2-3 opening lines that set a blameless tone. Output format: a section per participant [Role | Questions], then a short "reconciliation questions" list and a "facilitation notes" block. Guardrails: every question must be open and blameless — reject and rewrite any phrasing that implies fault, hindsight bias, or a "correct" answer the person failed to reach. These are prompts for a human conversation, not an interrogation script; the facilitator adapts them live. Do not fabricate details about what anyone did.
Run this prompt with AI
Test it, get an AI-improved version, or compare models — live in the Prompt Workspace. No copy-paste.
Why this prompt works
The most valuable contributing factors in any incident are invisible to the telemetry. Logs show what happened; they cannot show what the on-call engineer expected to happen, which dashboard misled them, or why applying the obvious fix felt too risky in the moment. That context only comes out in conversation — and whether it comes out at all depends entirely on how the questions are framed. Ask “why didn’t you roll back sooner?” and you get defensiveness; ask “what were you seeing that made rolling back feel uncertain?” and you get a systemic finding.
This prompt builds per-person question sets grounded in local rationality: the principle that everyone acted sensibly given their view at the time. By tailoring questions to each participant’s vantage point and explicitly banning hindsight-loaded phrasing, it steers the interviews toward system conditions — time pressure, unclear ownership, confusing tooling, alert fatigue — instead of individual shortcomings. Those conditions are the ones you can actually fix.
The reconciliation questions handle the reality that accounts diverge without anyone lying: two people remember the timeline differently because they had different information. Surfacing those seams with neutral questions lets the facilitator build one coherent timeline without making anyone defend their memory. Delivered as conversation starters with sequencing and tone-setting notes — not a rigid script — the output helps a human run interviews that produce a deeper, more honest postmortem while keeping the room psychologically safe.
Related prompts
-
Postmortem Contributing Factors vs Root Cause Classifier Prompt
Sort a postmortem's tangle of causes into the true root cause, the contributing factors, the proximate trigger, and the mitigating factors, so the document fixes the systemic source instead of stopping at the first plausible explanation.
-
Postmortem Review Pre-Read Pack Generator Prompt
Turn a draft postmortem into a concise pre-read pack that primes review attendees with the facts, open questions, and discussion goals so the blameless review meeting debates causes and fixes instead of reconstructing what happened.
-
Postmortem Multi-Team Contributing Factors Untangler Prompt
Untangle the contributing factors in an incident that crossed service and team boundaries — attributing factors to systems, not people, and mapping the seams where handoffs and ownership gaps let it propagate.
-
Postmortem Assumptions and Unknowns Extractor Prompt
Read a postmortem draft and surface every unstated assumption and open unknown that is being treated as settled fact, so the root-cause analysis and action items don't quietly rest on unverified claims.
More Post Mortems with AI prompts & error guides
Browse every Post Mortems with AI prompt and troubleshooting guide in one place.
Reading prompts? Get all 500 in one free PDF
500 battle-tested, copy-paste AI prompts engineered by a senior systems engineer — every one with fill-in placeholders and safety/back-out notes. Drop your email and it's yours.
- 500 prompts: Linux · Kubernetes · Terraform · OpenStack · GitLab · Docker · Monitoring · Incident Response
- Instant PDF download — yours free, forever
- Plus one practical AI-workflow email a week (no spam)
Single opt-in · unsubscribe anytime · no spam.