Returns and Refunds in Live Chat: A Playbook

Handling returns and refunds in live chat goes well when the policy does most of the talking. A customer who wants to send something back is already a little disappointed, and the chat either repairs that or makes it worse. The fastest, fairest conversations happen when your return policy is clear, the general questions are answered instantly from it, and a person handles the parts that involve a specific order or money moving.

This playbook is for online shops and service businesses that take returns. It covers how to write a policy a chatbot can answer from, what an AI assistant should and should not do, how a human should run the order-specific part, and how to keep a returns conversation from turning into a complaint.

Why Returns Conversations Go Wrong

Most bad returns chats are not caused by rude agents. They are caused by uncertainty. The customer does not know whether the item qualifies, who pays for shipping, how long the refund takes or what to do next. The agent may not know either, if the policy is vague or lives in several places.

The usual failure points are easy to recognise:

A policy with gaps. “Returns accepted within 30 days” says nothing about opened items, sale items, gifts or faulty goods.
Different answers from different people. One agent offers free return shipping as a goodwill gesture, the next one refuses, and the customer notices.
No clear next step. The customer is told “you can return it” but not how, where or by when.
Silence after the parcel arrives. The return reaches the warehouse, nobody tells the customer, and a new chat starts: “Did you get my return?”

Every one of these is fixed before the chat begins, by writing things down.

Write a Return Policy a Chatbot Can Answer From

An AI assistant that answers from your website is only as good as the policy page it reads. Talkmio’s assistant, Mio, answers only from your own content and never invents prices or policies, so a clear page produces clear answers and a vague page produces a handoff. That is the right behaviour, but you get far more value when the page is complete.

Cover the cases customers actually ask about

The return window, and when it starts: delivery date or order date.
Condition rules: unworn, unopened, with tags, original packaging.
Items that cannot be returned, such as personalised goods, hygiene products or digital downloads, and why.
Who pays return shipping, in which cases, and how much.
Faulty or damaged items, which are usually handled differently from change-of-mind returns.
Exchanges versus refunds, and store credit if you offer it.
How refunds are paid: original payment method, timing after the return is received.
Gifts, sale items and international orders.

Write it in plain sentences

Write “You have 30 days from delivery to return an item” rather than “Returns must be initiated within the applicable period”. Short, direct sentences are easier for customers to read and easier for an assistant to quote accurately. Put each rule under a heading so the right section is easy to find.

Check the legal baseline

Consumer law sets minimums your policy cannot go below. In the EU, online buyers generally have a 14-day right of withdrawal for most goods, as explained on the European Commission’s Your Europe guide to returns. Rules differ elsewhere. Check what applies to you, and if in doubt ask a lawyer rather than a chatbot.

What an AI Assistant Should Handle in Returns and Refunds

A large share of returns questions are general and have the same answer for everyone. These are ideal for an assistant, because the customer gets the answer immediately, at any hour, and your team never sees them.

“How long do I have to return this?”
“Do I pay for return shipping?”
“Can I return a sale item?”
“How long does the refund take?”
“Can I exchange for a different size?”
“Where do I send it?”

The questions that should go to a person are the ones about a specific order or a specific payment: “I sent my return two weeks ago, where is my refund?”, “The item arrived broken”, “Can you make an exception, I’m one day late”. In Talkmio, the assistant hands a chat to your team when it cannot answer from your content, when the visitor asks for a person, complains, or asks about a specific order. Those triggers match the returns split almost exactly.

You can make the boundary explicit with a business instruction: “Explain the return policy. Never approve a return, promise a refund or make an exception. Hand order-specific and damage questions to the team.” Our guide to when an AI chatbot should hand off to a human covers how to draw that line in other areas too.

Who Handles What: A Quick Reference

A simple table helps everyone on the team, and it is a useful test for your assistant setup.

Situation	AI assistant	Person	Notes
Return window and conditions	Yes	—	Answered from the policy page
Return shipping cost	Yes	—	Only if the page states it
Starting a return for an order	Explains the steps	Yes, if a manual check is needed	Link to your returns portal if you have one
Damaged or faulty item	No	Yes	Ask for photos, open a ticket
Refund status for a specific order	No	Yes	Needs a lookup in your shop system
Late return, exception request	No	Yes	Judgement call, keep it consistent
Chargeback threat or angry complaint	No	Yes, senior if possible	Priority high

Running the Human Part Well

When a returns chat reaches a person, the customer has usually already read the policy. They want action, not a repeat of the rules.

Ask for everything once

Ask for the order number, the item and the reason in one message, plus photos if the item is damaged. Asking for one detail at a time drags the chat out and annoys people. If many of your chats are about orders, an optional order-number field in the pre-chat form saves a round trip, but keep it optional so pre-sale visitors are not blocked.

Use saved replies for the common outcomes

Most returns end in one of a few ways: return approved with instructions, exchange arranged, refund issued, return refused with reason, faulty item replaced. Write a saved reply for each. In the Talkmio Inbox, typing / inserts a quick reply, which you then adjust to the case. Our collection of canned responses that sound human has more examples.

Say what happens next, with a timescale

“I’ve approved the return. Print the label from the e-mail I just sent, drop the parcel at any post office within 14 days, and we refund within 5 working days of receiving it.” The customer should leave the chat knowing exactly what to do and when to expect money. Use the timescales from your policy, not hopeful ones.

Note the case for the next person

Returns often take more than one conversation. Write an internal note, visible only to your team, with what was agreed. If the customer comes back a week later, whoever answers can see the history in Contacts instead of asking the same questions again.

When a Return Becomes a Ticket

A chat is fine for a return that is settled in one conversation. Anything that waits on something, such as a parcel in transit, a warehouse inspection or a supplier decision on a faulty item, is better tracked as a ticket.

In Talkmio, any conversation can become a numbered ticket with priority and status. Give the customer the ticket number. On Pro and higher plans, the e-mail channel lets follow-ups continue by e-mail in the same thread, so the customer does not need to come back to the website to check on a refund. Our guide to tickets vs live chat explains when to switch.

Set the ticket status to Pending while you wait for the parcel, and close it only when the refund or replacement is done. That way the open list shows real work, and nothing falls through the gap between “approved” and “refunded”.

Keep Returns and Refunds in Live Chat Calm

Returns are emotional more often than they look. The customer may feel misled by a product photo, worried about money, or annoyed at the effort. A few habits keep the tone right.

Acknowledge first, then act. “That’s frustrating, sorry it arrived damaged” before “please send a photo”.
Do not argue with the reason. If the policy allows change-of-mind returns, the reason does not matter. Asking “why” can feel like a challenge. Ask only when it helps, and say why you are asking.
Be consistent with exceptions. Decide in advance who may approve a late return and when. Inconsistency creates more complaints than strictness.
Offer the better option when there is one. If an exchange solves the problem faster than a refund, say so, but never push it.

If a returns chat does turn heated, the techniques in our guide on handling angry customers in live chat apply directly.

Use Returns Chats to Reduce Returns

Every returns conversation tells you something about the product page, the sizing chart or the packaging. Read a sample each month and tag the reason: wrong size, not as pictured, damaged in transit, changed mind, arrived late. Patterns appear quickly.

Many “wrong size” returns on one product point to a sizing chart that needs fixing or a “runs small” note.
“Not as pictured” points to photos or descriptions that oversell.
“Damaged in transit” points to packaging or a carrier problem.

Fix the cause on the website, then add the answer to your knowledge base so the assistant can answer pre-sale questions such as “does this run small?” before the order is placed. Research from the Nielsen Norman Group on e-commerce product pages shows how much missing detail drives uncertainty, and that uncertainty is what later becomes a return.

Measure the Returns Workload

Track a few numbers so you know whether the setup is working:

Share of chats about returns and refunds. If it grows faster than orders, look at products and policy.
Share answered by the assistant alone. Talkmio’s reports show this overall. If returns questions keep being handed off, the policy page is probably missing an answer.
First-reply time on handoffs. Order-specific returns questions waiting hours generate repeat contacts.
Repeat contacts per return. Customers asking “did you get my parcel?” is a sign that you need status updates after the return arrives.

Frequently Asked Questions

Can an AI chatbot approve returns and refunds?

It should not. An assistant is good at explaining your return policy, the window, conditions and steps, because those answers are the same for everyone. Approving a specific return, issuing a refund or making an exception requires a look at the order and a judgement call, so hand those conversations to a person.

What should a return policy include so a chatbot can answer from it?

State the return window and when it starts, item condition rules, non-returnable items, who pays return shipping, how faulty goods are handled, exchanges versus refunds, how and when refunds are paid, and rules for gifts, sale items and international orders. Write each rule in plain sentences under its own heading.

How do I handle a customer asking about a refund that has not arrived?

Ask for the order number, check when the return was received and when the refund was issued, and tell the customer exactly what you found and when to expect the money. If you are waiting on something, turn the chat into a ticket, give them the number and follow up by e-mail.

Should returns be handled in live chat or by e-mail?

Start in whichever channel the customer chose. Simple returns can be settled in one chat. Returns that wait on a parcel, an inspection or a supplier are better tracked as tickets with e-mail follow-up, so the customer gets updates without coming back to the website.

How can live chat help reduce the number of returns?

Returns chats reveal why products come back. Tag the reasons each month, fix sizing charts, photos and descriptions that cause them, and add the answers to your knowledge base so an assistant can answer pre-sale questions such as “does this run small?” before the customer orders.

The Bottom Line

Returns and refunds in live chat become fast and fair when the policy is complete and written in plain language, an assistant answers the general questions from it, and a person handles anything involving a specific order or money. Ask for everything once, state the next step with a timescale, and track longer cases as tickets. Then use the reasons behind returns to fix product pages. To set up an AI live chat that answers your return policy and hands order questions to your team, start free at app.talkmio.com.

Returns and Refunds in Live Chat: A Policy-First Playbook