Verified against ChatGPT · 2026-08-09
Write a refund response that states the decision in the first sentence, not the third paragraph
Produces a refund reply — approved, partial, or denied — that leads with the decision and the reason, instead of burying it under policy explanation the customer has to dig through to find their answer.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are writing a refund decision reply to a customer. The decision has already been made — your job is to communicate it clearly, not to re-litigate it. DECISION Partial refund: $34 of the $89 order, for the damaged item only. REASON FOR THE DECISION Two of three items arrived undamaged and are past our return window; the third arrived with a cracked lid, confirmed by the photo the customer sent. ORDER/CASE CONTEXT Order #48213, placed July 14, delivered July 19, complaint filed August 3. POLICY BASIS (if denial or partial) Damaged items are refunded in full within 30 days of delivery with photo proof; undamaged items are non-refundable after 14 days. TONE Warm but unambiguous — this customer has been polite throughout, so no defensiveness is needed, but the denial part still needs to be clear. HOW TO STRUCTURE THIS State the decision in the first sentence — approved, partial, or denied — with the dollar amount or item if applicable. Do not make the customer read a paragraph of context before finding out what happened to their money; that's the single most common failure in refund replies and the one customers complain about most. If the decision is a denial or partial refund, give the specific policy reason in one clear sentence, not a paraphrase of the whole policy document — quote the relevant condition plainly enough that the customer could check it themselves, without sounding like you're reading from a manual. If it's a partial refund, explain what portion is and isn't covered and why, so it doesn't read as an arbitrary number. Close with the concrete next step and timing (when the money will actually show up, or what the customer needs to do if they want to appeal), not a vague "please let us know if you have questions." WHAT NOT TO DO Do not soften a denial into ambiguous language that a customer could mistake for a maybe — an unclear denial generates a second contact, which is worse for everyone than a clear one delivered kindly. Do not apologize for a decision that was correctly made under policy; apologize for the situation, not for enforcing a rule that exists for a reason, unless the reason genuinely was our error. OUTPUT FORMAT 1. The reply, 80-150 words depending on whether an explanation of policy is needed. 2. If it's a denial, one alternate line offering a partial or goodwill gesture only if Damaged items are refunded in full within 30 days of delivery with photo proof; undamaged items are non-refundable after 14 days. allows for one — otherwise omit this line entirely rather than implying flexibility that doesn't exist.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Refund replies are one of the clearest cases where GPT-5.1's default structure actively works against the goal: left unconstrained, it tends to build a reply in the order a human explains a decision out loud — context, then reasoning, then verdict — because that's the natural narrative shape for justifying something. But a customer reading about their own money wants the verdict first and everything else is, to them, supporting detail they'll skim or skip; forcing decision-first ordering overrides the model's narrative default with the actual information hierarchy the recipient has. Quoting the specific policy condition rather than paraphrasing the whole policy matters because a vague paraphrase ("per our return policy") reads as a brush-off that could be hiding anything, while a specific, checkable condition ("damaged items within 30 days, undamaged after 14") gives the customer something concrete enough to either accept or specifically dispute, which resolves more cases in one exchange instead of triggering a follow-up asking what the policy actually says. The instruction against softening a denial into ambiguous language addresses a real tendency in the model to hedge negative outcomes — phrases like "we may not be able to" or "this could be difficult to process" feel kinder in isolation but functionally create a second support contact when the customer interprets ambiguity as room to negotiate, which is worse for resolution time than a clearly stated no delivered with warmth. Restricting the apology to the situation rather than the decision itself is a subtle but important distinction GPT-5.1 doesn't make on its own — it will reflexively apologize for enforcing a correctly applied policy, which subtly undermines the policy's legitimacy every time it's invoked.
What you get back
We've approved a $34 refund for the item that arrived with a cracked lid — the photo you sent confirmed the damage, and that's covered in full under our 30-day damage policy. The other two items in the order arrived undamaged and are outside our 14-day return window for non-defective items, so those aren't eligible. The $34 will post back to your original payment method within 3-5 business days.
Verified against
ChatGPT GPT-5.1 · 2026-08-09
Changelog
- 2026-08-09 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
