Diagnose why our CSAT score dropped
A structured investigation of a falling satisfaction score that separates real problems from noise and points to fixes.
The prompt
Help me figure out why our customer satisfaction (CSAT) score dropped.
Score trend: [e.g. "91% to 84% over 6 weeks"]
Survey responses and volume: [NUMBER OF RESPONSES per period, response rate]
Breakdowns I have: [by channel, agent, team, ticket category, customer segment, product area]
Low-score comments: [PASTE A SAMPLE - remove personal data]
What changed in that period: [new product release, policy change, staffing changes, new tools, outages, pricing changes]
Other metrics in the same period: [first response time, resolution time, reopen rate, backlog]
Do this:
1. Check whether the drop is real: is the sample size large enough, and did response rate or survey method change?
2. Break the drop down: is it concentrated in one channel, category, team, or segment, or spread across everything?
3. Separate support-caused dissatisfaction (slow, unhelpful, rude, wrong answer) from product or policy dissatisfaction (they hated the answer, not the service). Use the comments for this.
4. Link the timing to the changes I listed.
5. Rank the likely causes by evidence strength, with what would confirm or rule out each.
Return: a short summary for my manager, a table (Cause | Evidence | Confidence | How to confirm | Fix), and 3 actions for the next two weeks. Don't invent data; if a breakdown I need is missing, tell me which one to pull.
Highlighted [brackets] are blanks to replace with your own details.
Fill in the blanks and use this promptMore Customer Support prompts
What to fill in
- e.g. "91% to 84% over 6 weeks"
- NUMBER OF RESPONSES per period, response rate
- PASTE A SAMPLE - remove personal data
- first response time, resolution time, reopen rate, backlog
How this prompt is built
This prompt follows the RTF structure (Role, Task, Format). RTF tells the model who to be, what to do and what shape the answer should take. It is the quickest structure that still gives noticeably better results than a bare request.
Related prompts
- Respond to an angry customer and de-escalateDe-escalates an upset customer, owns the issue, and offers a concrete resolution.
- Improve my customer service operationA plan to improve service quality, speed, and efficiency across channels.
- Respond to a guest who reports food poisoning or illnessA careful response, internal investigation checklist, and record-keeping steps after a foodborne illness complaint.
- Update support and chatbot content for a product changeFind every answer, article, and macro a product change affects, and get the rewritten content and a rollout checklist.