Find and fix the root causes of recurring IT incidents
Turns ticket data into a problem register with patterns, root-cause plans, workarounds, and permanent fixes for IT service teams.
The prompt
Help me run problem management on recurring IT incidents so they stop coming back.
Ticket data: [PASTE an export or summary - ticket ID, date, category, service, short description, resolution, time to resolve]
Time period: [e.g. "last 90 days"]
Our setup: [ITSM tool, service desk size, key services, ITIL practices already in place]
Biggest pain: [e.g. "the same VPN and password issues every week", "outages after every release"]
Steps:
1. Find patterns: group tickets by service, category, symptom, time of day or week, and recent changes. Rank clusters by volume, total handling time, and business impact, and show the evidence for each.
2. Pick 3-5 problem candidates worth investigating and write a problem record for each: description, affected services and users, incident count, suspected trigger.
3. Root-cause approach for each: which technique fits (5 Whys, fishbone, change correlation, fault tree), what data to pull, and who to involve.
4. Workarounds: a known-error entry (symptom, workaround, status) the service desk can use now.
5. Permanent fixes: options with effort, cost, and risk, and whether a change request is needed.
6. Prevention: links to change management, monitoring alerts, and self-service or knowledge articles that deflect repeat tickets.
Return a problem register: Problem | Evidence | Impact | Likely cause | Workaround | Permanent fix | Owner | Target date. Then a five-line summary for IT leadership.
Label every cause [HYPOTHESIS] until it's confirmed, and don't invent figures beyond the data I give you.
Highlighted [brackets] are blanks to replace with your own details.
Fill in the blanks and use this promptMore Projects & Operations prompts
What to fill in
- e.g. "last 90 days"
- HYPOTHESIS
How this prompt is built
This prompt follows the Few-Shot structure (Few-shot prompting). A few-shot prompt includes one or more worked examples of the input and the output you want, so the model can copy the pattern instead of guessing it.
Related prompts
- Find unnecessary approvals in a workflow and simplify sign-offsReviews every approval in a workflow, flags the ones that add delay without reducing risk, and proposes a leaner sign-off design with thresholds.
- Write a quality management plan for a project or featureA practical plan defining quality standards, testing, acceptance criteria, reviews, and defect handling before work starts.
- Plan a Windows imaging and laptop deployment rolloutA deployment plan for setting up company Windows devices at scale: tool choice, image and driver strategy, task sequence, testing, and rollout waves.
- Create a role-based IT setup checklist for new hiresBuilds a provisioning checklist per role so every new starter gets the right accounts, devices, and access on day one.