5 Whys
Five Whys on a Recurring Issue
Run Five Whys on a problem that keeps coming back, and stop at a cause you can actually change.
The Prompt
five-whys-on-a-recurring-issue.md
Run Five Whys on a recurring issue. Stay empirical. Issue: {{issue}} How often it returns: {{frequency}} Last three occurrences: {{occurrences}} What we usually blame: {{usual_blame}} For each why: the answer, and the evidence from {{occurrences}}. Stop early if the next why would be psychology you cannot know. End with: - Root cause you can act on - A countermeasure - A detection so you know if it returns - Whether {{usual_blame}} was a scapegoat If the occurrences do not share a cause, say it is not one problem.
Negative prompt
Included when you copy the prompt
Do not invent evidence that is not in {{occurrences}}. Do not continue the chain into unverifiable psychology ("they don't care", "bad attitude"). Do not treat {{usual_blame}} as proven without checking it against the three occurrences. Do not declare a single root cause if the occurrences clearly do not share one. Do not end without a countermeasure you can actually change and a detection signal. Do not stop at "human error" when a missing checklist, alert, or ownership gap is visible in the evidence.
Fill in your details below - this prompt updates as you type, then copy it.
Fill in your details
0 of 4 filled. Preview updates as you type - then copy the finished prompt (negative prompt included when present).
five-whys-on-a-recurring-issue.md
Prompt preview
Run Five Whys on a recurring issue. Stay empirical. Issue: {{issue}} How often it returns: {{frequency}} Last three occurrences: {{occurrences}} What we usually blame: {{usual_blame}} For each why: the answer, and the evidence from {{occurrences}}. Stop early if the next why would be psychology you cannot know. End with: - Root cause you can act on - A countermeasure - A detection so you know if it returns - Whether {{usual_blame}} was a scapegoat If the occurrences do not share a cause, say it is not one problem.
Examples
Example input
Issue: Production hotfix deploys miss the changelog, so support and CS learn about breaking changes from customers. How often it returns: About twice a month for the last quarter. Last three occurrences: 1. May 3 - Auth token TTL shortened in a hotfix; help center still said 24h; 18 tickets. 2. May 19 - CSV export column order changed in hotfix; CS macros broke; no #product-updates post. 3. June 2 - Rate limit on public API tightened Friday night; partner Slack got no notice; two partners escalated Monday. What we usually blame: "Engineers forget to write release notes under pressure."
What you should get
Why 1: Support/CS see outdated docs and macros after hotfixes. Evidence: May 3 help center wrong; May 19 macros broke; June 2 partners unnotified. Why 2: Hotfix path does not require a customer-facing notice. Evidence: All three were hotfixes; no #product-updates / partner note in any case. Why 3: Hotfix checklist has no "notify + changelog" gate before merge to prod. Evidence: Pattern across three dates; usual process for scheduled releases includes a changelog step that hotfixes skip. Why 4: Ownership for hotfix comms is unclear (eng assumes CS watches deploys; CS assumes eng posts). Evidence: Tickets and partner escalations arrived before any internal post. Stop: Next why would be guessing motives ("engineers don't care") - not knowable from occurrences. Root cause you can act on: Hotfix workflow has no mandatory notice/changelog check, and no named owner for customer comms. Countermeasure: Add a required PR template checkbox + CI reminder for hotfixes: changelog stub + #product-updates ping (and partner Slack for API changes); on-call eng owns it before prod merge. Detection: Weekly count of prod hotfixes without a matching changelog entry or #product-updates post within 2 hours; alert if >0. Usual blame: Partly scapegoat. Forgetting under pressure is a symptom; the missing gate and unclear ownership explain all three occurrences better than individual memory.
When to use this prompt
Use it when
- You need help with ops.
- You need help with engineering.
- You need help with quality.
Skip it when
- You only need a one-line ask with no constraints, format, or context.
- You cannot supply the fill-in details the template asks for.
- A different category or tool-specific workflow fits the job better.
How to get a good result
Done when
Why 1: Support/CS see outdated docs and macros after hotfixes. Evidence: May 3 help center wrong; May 19 macros broke; June 2 partners unnotified. Why 2: Hotfix path does not require a customer-facing notice. Evidence: All three were hotfixes; no #product-updates / partner note in any case. Why 3: Hotfix checklist has no "notify + changelog" gate before merge to prod. Evidence: Pattern across three dates; usual process for scheduled releases includes a changelog step that hotfixes skip. Why 4: Ownership for hotfix comms is unclear (eng assumes CS watches deploys; CS assumes eng posts). Evidence: Tickets and partner escalations arrived before any internal post. Stop: Next why would be guessing motives ("engineers don't care") - not knowable from occurrences. Root cause you can act on: Hotfix workflow has no mandatory notice/changelog check, and no named owner for customer comms. Countermeasure: Add a required PR template checkbox + CI reminder for hotfixes: changelog stub + #product-updates ping (and partner Slack for API changes); on-call eng owns it before prod merge. Detection: Weekly count of prod hotfixes without a matching changelog entry or #product-updates post within 2 hours; alert if >0. Usual blame: Partly scapegoat. Forgetting under pressure is a symptom; the missing gate and unclear ownership explain all three occurrences better than individual memory.
Common mistakes
- Leaving issue, frequency, occurrences vague - the model then invents placeholders.
- Asking ChatGPT or Claude for a shorter answer after a long structured prompt - that throws away the format rules.
- Stopping when the first draft appears instead of checking it against the success criteria below.
FAQ
- What do I fill in before I copy this prompt?
- Replace these placeholders with your details: issue, frequency, occurrences, usual_blame. Required fields are marked on the page. The prompt text updates as you type.
- Which AI tools is this prompt written for?
- It works best with ChatGPT, Claude. Copy the filled prompt and paste it into the tool you already use.
- What should a good result look like?
- Why 1: Support/CS see outdated docs and macros after hotfixes. Evidence: May 3 help center wrong; May 19 macros broke; June 2 partners unnotified. Why 2: Hotfix path does not require a customer-facing notice. Evidence: All three were hotfixes; no #product-updates / partner note in any case. Why 3: Hotfix checklist has no "notify + changelog" gate before merge to prod. Evidence: Pattern across three dates; usual process for scheduled releases includes a changelog step that hotfixes skip. Wh…
- What are common mistakes with this prompt?
- Leaving issue, frequency, occurrences vague - the model then invents placeholders. Asking ChatGPT or Claude for a shorter answer after a long structured prompt - that throws away the format rules. Stopping when the first draft appears instead of checking it against the success criteria below.
Best for
- Ops
- Engineering
- Quality
Related Prompts
More prompts like this — same topic and style.