Home · Business · Admin & Office · Documents & templates
As of 13 August 2026, AI can write a response to a customer complaint.
This still needs a person who signs their name to it.
Can you do it?
5 minutesto a draft.
30 minutesto something you’d act on.
Cost, all in£0
Skill neededchat-fluent
Who has to check ityou
What the alternative costsThe supplied tool list gives no price for a human complaint-response service.
If this goes wrong: you make an unsupported promise or admission, worsen the dispute, or miss an issue that needs escalation.
What to actually do
Hand it to a person
The route this page recommends
A person who owns the outcome does this end to end, worth it when the failure is dear.
Use a tool built for this
Second choiceDo it yourself
The distant thirdA chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.
How to actually do it
- Open the original complaint and your customer record, then copy the full complaint into the prompt without adding guesses about the customer's motive.
- Gather the order or service record, contact history, relevant policy or terms, and the remedy you are authorised to offer.
- Paste the verified facts and relevant policy text into the matching sections of the prompt, removing unnecessary personal data.
- Ask the chatbot to produce the customer-facing response and the separate list of claims, promises and policy references to check.
- Compare every name, date, event, amount, policy reference and promised action in the draft with your records and current policy before sending.
- Check that the proposed remedy is authorised and that any legal admission, compensation issue, data concern or unresolved dispute is escalated to the responsible manager or adviser.
- Send the checked response through your normal customer-service channel and record the response, remedy and any follow-up deadline in the customer record.
Prompt
Write a response to the customer complaint below. Customer complaint: [PASTE THE FULL COMPLAINT] Verified facts: [LIST THE ORDER, SERVICE, DATES, CONTACTS, ACTIONS ALREADY TAKEN, AND ANY OTHER FACTS] Relevant company policy or terms: [PASTE ONLY THE RELEVANT POLICY OR TERMS] Remedy I am authorised to offer: [STATE THE REFUND, REPLACEMENT, CREDIT, EXPLANATION, INVESTIGATION, OR OTHER ACTION] Reply requirements: - Use a calm, respectful and plain British English tone. - Acknowledge the customer's specific concerns without sounding defensive. - Use only the verified facts and policy text supplied here. Invent nothing. - Do not admit legal liability, promise compensation, or state that a law or policy was breached unless that conclusion is explicitly supplied above. - Do not disclose internal processes or personal data that the customer does not need. - State what we will do next, who will do it, and any confirmed timescale. If a timescale is not supplied, write [TIMESCALE TO CONFIRM] rather than inventing one. - If the facts are incomplete or the remedy is not authorised, flag the missing point before drafting the final response. - Keep the reply under 250 words and include a suitable subject line. Return the final customer-facing response first, followed by a short list of claims, promises and policy references that I must check before sending.
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- AI cannot decide whether the complaint reveals a wider product, service or safeguarding problem that your business should investigate.
- AI cannot take responsibility for an admission, refund, compensation promise or escalation caused by the wording.
- AI cannot reliably judge when an ambiguous complaint needs a manager, legal adviser or another specialist.
- AI cannot know whether the proposed remedy is commercially fair unless you provide the authority, policy and relevant context.
- AI cannot confirm that the customer's underlying account of events is true.
Even on a YES, the friction has a name: stakes of error, judgement under ambiguity and legal accountability.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 2 |
| Inputs | 2 |
| Verification | 1 |
| Liability | 1 |
| Effort delta | 2 |
| Total | 8 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can ChatGPT write a customer complaint response?
- Yes. Give it the complaint, verified facts, relevant policy and the remedy you are authorised to offer, then check every factual statement and promise before sending. It drafts the reply, but your business remains responsible for it.
- Is it safe to use AI to reply to a customer complaint?
- It is suitable for a first draft when you remove unnecessary personal data and provide only verified information. Do not let it invent a remedy, make a legal admission or decide whether a serious or ambiguous case needs escalation.
- How do I get AI to write a professional complaint response?
- Ask for a calm British English reply that acknowledges the specific concerns, uses only supplied facts, states the authorised remedy and flags missing information. Ask for a separate checking list of claims, promises and policy references before the customer-facing text.
- Should I let AI send a customer complaint response automatically?
- No, not for complaints where the reply could create a financial, legal or reputational commitment. Have a person compare the draft with the customer record and current policy, and escalate cases involving liability, compensation or unresolved facts.
Nearby answers
Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.
The newsletter
AI news, new answers and product picks, straight to your inbox.