TrolyPage

An autonomous Marketing & Sales operating system for businesses that want AI-powered omni-channel growth.

Contact Info

Follow Us

AI Chatbots & Automation 5 min read

50 ChatGPT Prompts for Customer Support and Sales Replies

TrolyPage

Nov 19, 2026
9 views
50 ChatGPT Prompts for Customer Support and Sales Replies

Here are 50 ChatGPT (or Claude, Gemini) prompts for customer support and sales replies, in 6 groups: drafting answers, handling objections, follow-ups, summaries and handovers, building chatbot knowledge, and quality review. Replace the [brackets] with your real data, and always tell the AI not to invent facts.

1. Drafting replies (10)

  1. "You're a support agent for [store], selling [category]. Using only this data: [data], answer: '[question]'. Under 60 words, friendly, end with a question. If the data doesn't cover it, say you'll check."
  2. "Write 3 versions of a reply to '[question]': concise, warm, formal. Use only: [data]."
  3. "A customer is [height/weight] and asks which size of [product] to buy. Size chart: [chart]. Recommend a size and explain briefly."
  4. "Write a one-sentence greeting for first-time messages that asks which product they're interested in."
  5. "Answer each part of this multi-part question as bullets, using [data]: '[question]'."
  6. "Rewrite this reply to sound natural and remove filler, keeping every number: '[reply]'."
  7. "Write an away message for after hours that sets expectations and asks for details."
  8. "Explain our return policy [policy] in plain language in 3 sentences."
  9. "Translate this reply into [language], keeping prices, dates and commitments unchanged: '[reply]'."
  10. "Turn this long policy into a short chat answer for the question '[question]': [policy]."
50 prompts in 6 groups

2. Handling objections (8)

  1. "Customer says 'too expensive'. Product: [name], $[price], benefits: [list]. Acknowledge, explain value, suggest [cheaper option]. Don't offer discounts."
  2. "Customer says 'I'll think about it'. Write a no-pressure reply asking what they're unsure about."
  3. "Customer compares us to a cheaper competitor. Reply without criticizing them, highlighting [warranty/returns/quality]."
  4. "Customer worries it won't look like the photos. Reassure using [policy] and offer a real video."
  5. "Customer asks for a discount; current offers: [offers]. If none applies, decline politely and suggest [bundle]."
  6. "List 5 reasons customers hesitate on [product] with a short reply for each."
  7. "Customer asks if we're legit. Write a reply referencing [reviews, policies, company details] without overclaiming."
  8. "Customer is upset about a delay. Write an apology with a concrete next step: [step]."

3. Follow-ups (8)

  1. "Customer asked about [product] yesterday and went quiet. Write a helpful nudge under 40 words, no pressure."
  2. "Write a check-in 3 days after delivery of [product] with one usage tip."
  3. "Write a polite review request after purchase, with where to leave the review."
  4. "Write a back-in-stock message for a customer who asked about [product]."
  5. "Write 3 replenishment reminders for [consumable] bought [X] weeks ago, each with a different angle."
  6. "Write a follow-up for an abandoned order conversation that asks if anything blocked them."
  7. "Write a renewal reminder for [service] expiring on [date], friendly and factual."
  8. "Write a win-back message for a customer inactive for [months], offering help, not a hard sell."

Note: on Messenger and Instagram, standard follow-ups must be sent within 24 hours of the customer's last message.

4. Summaries and handovers (8)

  1. "Summarize this chat in 3 lines: what the customer needs, what we told them, what's pending: [chat]. Don't guess."
  2. "Extract order details: product, size, qty, name, phone, address. Mark missing fields 'missing'."
  3. "Write a shift handover note from this chat: status, next action, caution."
  4. "Rate customer sentiment (positive/neutral/negative) with one-line reasoning."
  5. "List open questions the customer asked that we haven't answered yet."
  6. "Identify any promise we made in this chat (dates, refunds, discounts)."
  7. "Turn this complaint thread into a concise escalation for a manager."
  8. "Tag this conversation with up to 3 labels from: [label list]."

5. Building chatbot knowledge (8)

  1. "Convert this product description into a structured entry: name, price, variants, materials, care, returns. Don't add facts."
  2. "Group these 50 customer questions into 8–10 topics with a representative question each: [list]."
  3. "Rewrite this policy as short, single-condition sentences: [policy]."
  4. "List questions customers might ask about [product] that this data can't answer: [data]."
  5. "Write 10 ways customers might ask about shipping cost, including typos and slang, for testing."
  6. "Create an FAQ document from these chats; remove all personal details: [chats]."
  7. "Write explicit 'we do not' statements from this policy (exclusions)."
  8. "Draft hand-off rules for a support bot in [industry]."

6. Quality review (8)

  1. "Compare this bot answer with the source data and list anything incorrect or unsupported."
  2. "Review these 20 chats and find slow or inaccurate replies; suggest better versions."
  3. "Check this reply for promises we can't guarantee (same-day delivery, full refunds)."
  4. "Write a 10-point checklist for evaluating support conversations."
  5. "Score this reply for clarity, accuracy and tone from 1–5 with reasons."
  6. "Find inconsistent answers to the same question across these chats."
  7. "Suggest 5 knowledge-base additions based on these escalated chats."
  8. "Write test questions designed to tempt a bot into inventing a discount."

Using prompts inside an AI agent

With an AI agent connected to your channels, you configure knowledge and rules once instead of pasting data into every prompt. Prompts from groups 1–2 become voice and handling rules; group 6 becomes your weekly review.

FAQ

Is it safe to paste customer chats into ChatGPT?

Remove names, phone numbers and addresses first, and follow your privacy obligations.

Written by the TrolyPage team. Updated November 2026.