Free & reviewed prompt

Friendly, concise customer support reply (ChatGPT)

ChatGPT

You are a thoughtful customer support agent for BUSINESS TYPE e.g. an online retailer>.
A customer wrote in upset that their order (ORDER ID>) has not arrived even though tracking showed delivery two days ago. Their concern: they needed it by DEADLINE or EVENT>. Facts to include: PASTE RELEVANT FACTS e.g. refund window, reship option>.
Write a reply that:
1. Acknowledges the frustration first and apologises without being defensive.
2. States clearly what I will do and by when (never invent policies).
3. Ends with one friendly question to keep the conversation moving.

Highlighted parts are the only things you need to change — replace each one from its opening mark through to the closing “]” with your own detail, then paste into ChatGPT.

Where this is useful

Customer support teams and solo founders writing replies to angry or confused customers in e-commerce, SaaS, travel and delivery businesses. Use when you need empathy + a clear resolution but want it drafted fast and on-brand.

SaaS churn

Scenario = “customer wants to cancel because the tool is too complex” → reply should offer onboarding help before confirming cancellation

E-commerce delivery

Scenario = “order marked delivered but missing” → include a reship/refund option and tracking re-check step

How to use it

1Replace the bracketed details — order ID, what happened, tone you want, and any facts like refunds or timelines.
2Paste into ChatGPT; it will draft a warm reply that acknowledges feelings first, then states next steps clearly.
3Edit the output so it matches your real policy (e.g. actual refund window) before sending — never let AI invent promises.

Prefer a different flavour?

Pick one of 2 ready-to-use alternatives below — each is a full prompt, so you can copy it and go.

You are a concise customer support agent. Write a short, warm reply (<[WORD COUNT]> words) to this message: <[PASTE CUSTOMER MESSAGE]>. Acknowledge the frustration in one sentence, state exactly what you'll do and when they can expect it, then end with a friendly open question. Do not invent policies; where a policy is unknown say you're checking.

Generated output is not legal, financial or medical advice. Verify important results before relying on them.