ChatGPT
Handle a Complaint Without Making It Worse
Reply to an upset customer with acknowledgement, a fact-based path, and a remedy you are allowed to give.
Angry Customer
A tighter de-escalation script for high-heat messages: calm tone, one remedy, one boundary.
The customer is angry. Reply in a way a reasonable person would accept. Message: {{message}} Facts: {{facts}} Allowed remedy: {{remedy}} If they crossed a line (abuse): {{abuse}} Produce: 1. What not to say (3 phrases) 2. The reply (max 120 words): feeling → fact → remedy → next step 3. If {{abuse}} is yes, add one boundary and still offer the remedy through a channel No sarcasm. No "calm down". No blame-the-customer.
Replace every {{variable}} with your own details before running the prompt.
4 variables
Reply to an upset customer with acknowledgement, a fact-based path, and a remedy you are allowed to give.
Cluster ticket themes into an FAQ with short answers, links, and the tickets that still need a human.
Reply to a public review: thank, address the point, take details off-thread, and skip the argument.