Customer service prompt

ChatGPT Customer Service Prompt for Dropshipping

Use this prompt to draft customer service replies, reusable macros, and escalation rules for common dropshipping support situations.

The goal is to answer customers clearly without inventing order updates or promising refunds, replacements, delivery dates, discounts, or policy exceptions your store cannot actually support.

See how to use it

How to use this prompt

Give the assistant your real support rules first. Then add only the order facts needed for the situation and review the reply before it reaches the customer.

  1. Set the policy boundaries. Add your real shipping, return, refund, cancellation, replacement, and warranty rules before asking for customer-facing replies.
  2. Add the known case facts. Give the assistant the relevant order status, tracking information, dates, and customer request. Keep unknown information clearly unknown.
  3. Review or escalate. Check every promise against your policy. Send unusual, sensitive, expensive, or unclear cases to a human before replying.

Building reusable macros?

Use placeholders such as [customer name], [order number], [tracking link], and [next update date] instead of putting real customer information into the template. Add case-specific facts only when you actually handle the ticket.

Customer support

Dropshipping customer service prompt

Paste this into your AI assistant with your actual support policies and the common situations you handle. Keep private customer information out unless your approved privacy process allows it.

Supplier guide

What this prompt should give you

  • Reusable customer support macros grounded in your actual policies.
  • Clear internal notes showing what the support agent still needs to check.
  • Escalation rules for cases that should not be handled automatically.
  • A do-not-say list and final QA checklist for avoiding unsupported promises.
Full prompt
Act as a careful ecommerce customer service manager for a dropshipping store.

Your job is to help me create support macros and draft customer replies that match my real policies and the order information I actually provide.

Do not invent:

- Tracking updates
- Order status
- Processing status
- Delivery dates
- Supplier responses
- Refund approval
- Replacement approval
- Cancellation approval
- Discounts
- Store credit
- Compensation
- Policy exceptions
- Warranty coverage
- Investigation results

Do not promise an outcome simply because it would make the reply sound more helpful.

If information is missing, label it [needs verification].

If a decision requires human approval, say [manual review required].

If supplier action is needed, say [supplier follow-up required].

If you cannot inspect a URL, screenshot, tracking page, policy page, supplier message, or external source, say so clearly and ask me to paste the relevant information.

Do not claim that you checked an order, tracking page, or supplier response unless I actually supplied that information.

PRIVACY RULE

Do not ask me to provide unnecessary private customer data.

For reusable templates, use placeholders such as:

[customer name]
[order number]
[order date]
[tracking number]
[tracking link]
[current tracking status]
[product name]
[refund amount]
[next update date]

Do not require full payment information, passwords, authentication codes, or other unnecessary sensitive information.

Store:
[store name]

Support tone:
[friendly, calm, concise, practical, premium, casual, etc.]

Support email or channel:
[email, chat, social, etc.]

Target market:
[country or region]

Shipping policy:
[processing time, estimated shipping time, tracking rules, delayed-order policy, lost-order process]

Cancellation policy:
[when cancellation is allowed and what happens after processing begins]

Return policy:
[return window, eligible items, exclusions, return shipping rules]

Refund policy:
[when refunds are available, refund method, expected processing rules]

Damaged or defective item policy:
[required evidence, replacement/refund rules, time limits]

Wrong or missing item policy:
[what the customer should provide and what the store can offer]

Warranty or replacement policy:
[coverage, exclusions, duration]

Address-change policy:
[when an address can be changed]

Delivered-but-not-received policy:
[what checks happen before refund or replacement review]

Other support rules:
[chargeback handling, fraud review, high-value-order rules, escalation rules, goodwill limits]

Common support situations:
[list the issues your store handles most often]

Please create a customer service support kit with these sections:

1. Policy and authority audit

Summarize only the policies I actually provided.

Create three lists:

CAN BE PROMISED
[things clearly supported by my policy]

REQUIRES VERIFICATION
[things that depend on order status, tracking, supplier information, or another fact]

REQUIRES APPROVAL
[refunds, replacements, discounts, exceptions, or other actions that need a human decision]

Also list important policy gaps that I should fix before creating automated replies.

Do not fill missing policy rules with generic ecommerce practices.

2. Support tone rules

Create practical writing rules for customer replies.

Include:

- Tone
- Recommended reply length
- How to acknowledge frustration without over-apologizing
- When to apologize directly
- How to explain uncertainty
- How to give the next step clearly
- How to avoid defensive language
- How to avoid sounding robotic
- Words or phrases to avoid

Keep replies clear, calm, and specific.

Do not use fake empathy, exaggerated apologies, or corporate filler.

3. Reply structure

Create a default structure for most support replies:

1. Acknowledge the customer's question or problem
2. State what is currently known
3. State what is still being checked, if anything
4. Explain the next action
5. Give a realistic expectation for the next update only if supported
6. Close clearly

Do not add a delivery date or resolution deadline unless I provided one.

4. Macro library

Create customer-facing macros for the situations below when relevant to my policies.

For every macro provide:

MACRO NAME

USE WHEN:
[when the macro applies]

CHECK BEFORE SENDING:
[facts the support agent must verify]

SUBJECT:
[email subject if relevant]

REPLY:
[customer-facing reply]

INTERNAL NOTE:
[what the agent should record or check]

ESCALATE WHEN:
[conditions requiring manual review]

Create macros for:

- Where is my order?
- Tracking has not updated
- Order is still processing
- Order is delayed
- Estimated delivery window has passed
- Tracking says delivered but customer cannot find the order
- Customer entered the wrong address
- Customer wants to change the address
- Customer wants to cancel
- Customer wants a refund
- Customer wants to return the product
- Product arrived damaged
- Product appears defective
- Wrong item received
- Missing item or accessory
- Customer says the product does not match the listing
- Customer asks about shipping before ordering
- Customer asks about returns before ordering
- Customer asks about warranty
- Customer is angry or frustrated
- Customer has contacted support repeatedly about the same unresolved issue

Skip macros that are irrelevant to the policies I supplied.

5. Order-delay decision tree

Create a simple decision tree for delayed-order tickets.

Use only my actual policy rules.

Include checks such as:

- Has the order finished processing?
- Is tracking available?
- Has tracking updated?
- Is the order still inside the stated delivery window?
- Has the stated delivery window passed?
- Is supplier follow-up required?
- Is carrier follow-up required?
- Is manual refund or replacement review required?

For every branch, show:

ACTION:
[what the agent does]

CUSTOMER MESSAGE:
[what can safely be said]

DO NOT PROMISE:
[unsupported outcome]

If my policies do not define the next action, label it [policy gap].

6. Refund and replacement decision rules

Using only my policies, create a decision framework for:

- Refund request
- Damaged product
- Defective product
- Wrong product
- Missing item
- Lost shipment
- Delivered-but-not-received claim
- Cancellation request

For each case show:

INFORMATION TO CHECK
POLICY RULE
POSSIBLE ACTION
WHO CAN APPROVE IT
CUSTOMER MESSAGE
WHEN TO ESCALATE

Do not approve a refund or replacement unless the information I provided clearly allows it.

7. Escalation matrix

Create three levels:

LEVEL 1: Standard support
Cases that can usually use an approved macro.

LEVEL 2: Manual review
Cases where a person needs to inspect the facts or approve the outcome.

LEVEL 3: Priority escalation
Cases that should receive prompt human attention.

Consider escalating cases involving:

- Large refund or replacement value
- Repeated delivery failure
- Multiple failed replacements
- Chargeback or payment dispute
- Fraud concern
- Safety concern
- Threatened legal action
- Serious product defect
- Customer alleges injury or harm
- Privacy or data concern
- Repeated unresolved complaint
- Policy exception request

Do not provide legal conclusions.

For unusual or serious cases, recommend human review rather than inventing a resolution.

8. Supplier follow-up templates

Create short internal or supplier-facing templates for:

- Order still processing
- Tracking unavailable
- Tracking stopped updating
- Customer reports damage
- Wrong item sent
- Missing item
- Replacement requested
- Order may be lost

For each template specify what evidence or order information should be included.

Do not include unnecessary customer data.

9. Do-not-say list

Create a list of wording support agents should avoid.

Include unsupported phrases such as:

"Your order will definitely arrive tomorrow."

"We guarantee it will be delivered this week."

"Your refund has been approved."

"We have already contacted the carrier."

"The supplier confirmed..."

"We guarantee a replacement."

"This never happens."

"Your package is definitely not lost."

unless the relevant fact is actually known and the agent is authorized to say it.

For each risky statement, give a safer alternative.

10. Missing-information language

Create reusable phrases for situations where something is still unknown.

Examples should cover:

- Waiting for supplier confirmation
- Waiting for tracking information
- Policy requires manual review
- Customer needs to provide evidence
- Order information is incomplete
- No confirmed resolution yet

Make uncertainty clear without sounding dismissive.

11. Privacy and data-minimization checklist

Create guidance for using AI-assisted support without putting unnecessary customer information into the prompt.

Separate:

USE ONLY WHEN NEEDED
[minimum order information required for the task]

REDACT OR OMIT
[information that usually does not need to be included]

NEVER REQUEST FOR THIS WORKFLOW
[passwords, authentication codes, complete payment credentials, or other unnecessary secrets]

Do not assume that an AI tool is approved to receive customer data merely because it can process it.

12. Macro personalization rules

Explain what should be personalized in a customer reply.

Examples:

- Customer name
- Product
- Order situation
- Known tracking status
- Actual next step
- Relevant policy

Explain what should not be invented merely to make the reply sound personal.

13. Final reply QA checklist

Before a reply is sent, check:

FACTS
- Is the order status accurate?
- Is the tracking status accurate?
- Are dates accurate?
- Is the product correct?

POLICY
- Does the reply match the shipping policy?
- Does it match the cancellation policy?
- Does it match the return policy?
- Does it match the refund policy?
- Does it match the replacement or warranty policy?

PROMISES
- Is a delivery date being promised without evidence?
- Is a refund being promised without approval?
- Is a replacement being promised without approval?
- Is a discount or compensation being offered without authority?
- Is a supplier or carrier action being claimed without confirmation?

CLARITY
- Does the customer know what happens next?
- Is there unnecessary filler?
- Does the reply hide a delay or problem?
- Does the reply sound defensive?

PRIVACY
- Does the reply or AI input include unnecessary private information?

ESCALATION
- Should this case be reviewed by a person instead?

14. Support-library maintenance plan

Tell me when these macros should be reviewed.

Create a checklist for changes involving:

- Supplier processing times
- Shipping estimates
- Carriers
- Return window
- Refund rules
- Cancellation rules
- Replacement rules
- Warranty
- Store contact information
- Target countries
- New recurring support problems

Do not assume an old macro is still correct after a policy or supplier change.

15. Final recommendation

Give me:

READY TO USE AS A TEMPLATE
[macros supported by the policies I supplied]

NEEDS POLICY CLARIFICATION
[macros that depend on missing policy rules]

NEEDS HUMAN APPROVAL
[decisions that should not be automated]

NEEDS SUPPLIER INFORMATION
[cases where current fulfillment evidence is missing]

PRIORITY FIXES
[the most important support-system gaps to address first]

The goal is to help the customer understand what is happening and what comes next without making promises the store cannot keep.

What you still need to check yourself

A well-written reply can still be wrong. Check the real order information, policy, and authority behind the answer before sending it.

Pay particular attention to replies involving money, delivery promises, replacements, policy exceptions, or information that depends on a supplier.

  • Confirm the customer's current order and tracking status before replying.
  • Match the response to your current shipping, cancellation, return, refund, and warranty policies.
  • Do not promise a refund, replacement, discount, or policy exception without the required approval.
  • Do not state that a supplier or carrier has taken action unless that action was confirmed.
  • Escalate unusual, high-value, repeated, safety-related, or disputed cases manually.
  • Update macros when supplier processing, shipping, or store policies change.
  • Keep customer communication clear about delays instead of hiding bad news behind vague wording.
  • Avoid putting unnecessary customer information into an AI tool or reusable prompt.
Keep the limits clear

What this prompt cannot decide for you

AI can draft support language and organize a workflow. It does not have authority to make store decisions simply because the reply sounds confident.

  • It cannot know the current order or tracking status unless you provide reliable information.
  • It cannot approve refunds, replacements, discounts, cancellations, or policy exceptions.
  • It cannot prove that a supplier or carrier has taken an action that was not confirmed.
  • It cannot decide whether an unusual dispute, safety concern, or legal issue is resolved.
  • It cannot guarantee that an old support macro still matches your current policies.
  • It cannot make unnecessary customer-data sharing appropriate simply because the information helps write a reply.

Customer service prompt FAQs

Quick answers before you use the prompt to create macros or draft a real customer reply.

Can ChatGPT write customer service replies for dropshipping?

Yes. ChatGPT can draft support replies, macros, internal notes, and escalation rules. The final response still needs to match your real policies and the customer's actual order information.

What should I paste into this customer service prompt?

Start with your shipping, cancellation, return, refund, damaged-item, replacement, and warranty policies plus your preferred support tone. For individual tickets, add only the order facts needed to handle that situation.

Can I send AI-written customer service replies without reviewing them?

Review them first. Check the order facts, policy rules, and any promises about delivery, refunds, replacements, discounts, or next steps before sending the reply.

Should I paste customer data into ChatGPT?

Only provide customer information when your privacy process and the AI tool you are using allow it, and avoid including data that is not necessary for the support task. Reusable macros should use placeholders instead.

When should an AI-drafted support reply be escalated?

Escalate when the outcome needs human approval, important facts are unclear, the case falls outside your normal policy, or it involves an unusual dispute, safety concern, repeated failure, or another sensitive issue.

Need a different AI workflow?

Browse the full prompt library for product research, supplier checks, pricing, validation, product pages, ads, SEO, and more. Or open the AI Dropshipping Hub for the broader workflow.

Do Dropshipping helps you understand how dropshipping really works and decide if itโ€™s right for you. If it is, weโ€™ll help you move forward with practical guides, supplier research, and tools for product research and ecommerce.

Contact

Address: Keurenplein 4 (A7484) 1069 CD, Amsterdamย (NL)

Follow

Disclaimer: The content on Do Dropshipping is intended to inform, inspire, and guide your ecommerce journey. We research carefully and aim to keep information accurate and current, but it is not legal, financial, tax, or professional advice and may not fit your exact situation.

By using this site, you agree to our Terms & Conditions and acknowledge that actions you take based on our content are your responsibility. To the fullest extent permitted by law, Do Dropshipping is not liable for any direct or indirect issues arising from how you use the information here.

This is the official website of Do Dropshipping and reflects our personal views and experiences.

Affiliate Disclosure: Some links on this site are affiliate links, which means we may earn a small commission at no extra cost to you if you make a purchase.

Site Editor:ย Richard