Use this prompt recipe for refund request reply for an upset customer. It gives the AI a clear job, visible source inputs and a review step so the result can be edited instead of blindly copied.
| Best for | support agents, founders, operations teams and customer success managers |
|---|---|
| Search intent | Readers usually need a prompt that drafts a helpful support response while respecting policy limits and escalation rules. |
| Expected output | Issue summary, Customer reply draft, Next step, Internal verification note |
When this recipe is useful
Use this recipe when the task is specific enough to benefit from AI structure but still needs human judgment. In the situation "for an upset customer", the prompt should make the audience, source material and quality bar explicit before asking for a finished answer.
The goal is not to make a generic refund request reply. The goal is to create a draft that reflects the real context, avoids unsupported claims and gives the reviewer a clear path for improving the result.
Source inputs to prepare
| Customer issue | Add the real detail for refund request reply for an upset customer. If this detail is unknown, ask the AI to return a missing-information question. |
|---|---|
| Known policy | Add the real detail for refund request reply for an upset customer. If this detail is unknown, ask the AI to return a missing-information question. |
| Order or account facts | Add the real detail for refund request reply for an upset customer. If this detail is unknown, ask the AI to return a missing-information question. |
| Tone and empathy level | Add the real detail for refund request reply for an upset customer. If this detail is unknown, ask the AI to return a missing-information question. |
| Escalation rules | Add the real detail for refund request reply for an upset customer. If this detail is unknown, ask the AI to return a missing-information question. |
Copy-ready prompt
Recommended output structure
- Issue summary: Make this section specific to refund request reply for an upset customer and easy for a human reviewer to check.
- Customer reply draft: Make this section specific to refund request reply for an upset customer and easy for a human reviewer to check.
- Next step: Make this section specific to refund request reply for an upset customer and easy for a human reviewer to check.
- Internal verification note: Make this section specific to refund request reply for an upset customer and easy for a human reviewer to check.
Review checklist
- The reply acknowledges the customer issue clearly.
- Refunds, timelines and exceptions match the policy.
- Missing details are requested instead of guessed.
- Escalation notes are separated from customer-facing copy.
Common mistakes
- Sounding empathetic while avoiding the answer.
- Inventing exceptions or timelines.
- Mixing internal notes into the customer reply.
- Using a macro without adapting it to the ticket.
How to adapt it
If the first output feels too broad, add one concrete example of the audience, one example of the tone you want, and one example of a claim the AI must avoid. This usually improves the next answer more than adding more adjectives.
If the output is too long, ask the AI to keep the structure but shorten each section around the decision the reader needs to make. If the output is too shallow, add source notes and ask for missing questions before another draft.
FAQ
When should I use this prompt?
Use it when you need refund request reply for an upset customer and want the AI output to stay tied to your real source inputs, audience and review rules.
What should I prepare first?
Prepare the practical inputs: Customer issue, Known policy, Order or account facts, Tone and empathy level. If one of those is missing, keep it marked as unknown instead of asking the AI to guess.
Can I publish the output directly?
Treat the output as a draft. Check facts, claims, names, dates, policies and promises before using it in public or customer-facing work.