How to handle returns and refunds without drowning your support team

The holiday peak doesn't end on Cyber Monday. It moves. The orders shipped in November and December come back in January as returns, exchanges and refund questions, just when the team is tired and temporary staff have gone home.

Returns are not the problem. A store that sells clothing, shoes or gifts will always have them, and a generous returns policy sells more. The problem is when every return goes through a person: the customer writes to ask if they can return, then asks how, then sends photos, then asks where the label is, then asks where the refund is. That's five messages for one return.

This guide shows how to cut that to zero messages for most returns and one well-prepared conversation for the rest.

Why returns overwhelm support

Returns create support work in predictable ways:

• The policy is unclear. Customers can't tell whether sale items, opened products or gifts qualify, so they ask.
• There is no self-service. The only way to start a return is to email you.
• Details arrive in pieces. Order number in one message, photos in the next, reason in a third.
• Exceptions are decided ad hoc. Different agents give different answers, which customers notice and quote back.
• Refunds are invisible. Once the parcel is sent back, customers have no idea when the money will arrive, so they ask again.
• Gift returns are awkward. The person returning didn't place the order and doesn't have the order number.

Fixing these is mostly about process. Tools then make the process run by itself.

Step 1: Write a policy customers can understand

Your returns page is the source your agents, your AI agent and your customers all read. Put the answers people need at the top:

1. The window: "30 days from delivery", plus the holiday extension if you offer one.
2. The condition: unworn, with tags, in original packaging, or whatever applies to your products.
3. What can't be returned: final sale, personalised items, hygiene products.
4. Who pays return shipping, and whether exchanges are free.
5. How long refunds take after the return arrives, and how they are paid (original payment method or store credit).
6. How to start a return: one link.

Then add the legal detail underneath. If you sell to customers in the EU, remember they have a 14-day right of withdrawal on most online purchases. Since June 19, 2026, EU rules also require online shops to offer a clearly labelled "withdraw from contract" function so customers can cancel online without paper forms.

If two agents could read your returns policy and give a customer different answers, the policy isn't finished.

Step 2: Let customers start a return without writing in

A self-service returns flow removes the biggest share of returns contacts. At minimum, customers should be able to:

• find their order by order number and email (or a gift receipt code);
• choose the items and a reason from a list;
• see straight away whether the return is accepted under the policy;
• get a return label or drop-off instructions;
• choose a refund, an exchange or store credit.

Your store platform may offer basic returns. Dedicated platforms such as Loop (for Shopify) and Narvar (for large retailers) add rules, exchanges and logistics. Link the returns portal from your order emails, your help center and the first message of your chat widget.

Step 3: Collect the details automatically when customers do ask

Some customers will still ask in chat, by email or on Instagram. Don't make agents gather information one message at a time. A chat flow can:

1. Ask for the order number or email and look up the order.
2. Check the order date against the return window.
3. Ask which items and why, using the same reason list as your portal.
4. Ask for photos if the item is damaged or faulty.
5. Within policy: send the returns portal link or label. Outside policy: create a ticket for a person, with everything attached.

With an AI agent in front of it, customers can ask in their own words ("the shoes are too small, can I swap them?"). In Tidio, Lyro answers the policy questions and a Flow handles the steps above, so agents only see the cases that need a decision.

Step 4: Decide the exceptions before January

Exceptions are where returns get slow and inconsistent. Write the rules down once and give agents the authority to apply them:

Situation	Default decision	Who decides
Within window, item in condition	Accept automatically	Self-service
A few days outside the window	Accept as a goodwill gesture, once per customer	Any agent
Damaged or faulty on arrival	Replace or refund, no return needed for low-value items	Any agent, with photos
Final sale or personalised item	Decline politely, offer store credit if faulty	Any agent
High-value item outside policy	Case by case	Team lead
Repeated returns from the same customer	Review before approving	Team lead

Put these rules into your AI agent's content and your saved replies too, so the answer is the same at 3am as at 3pm.

Step 5: Offer exchanges and credit first

A refund ends the relationship; an exchange keeps the sale. In your returns flow and your replies:

• Show exchange as the first option when the reason is size, colour or fit.
• Make exchanges cheaper for the customer: free return shipping or a small bonus credit.
• Suggest the right alternative using the return reason ("runs small, try one size up").
• Ship the replacement quickly, before the original arrives back if the customer's history allows it.

Step 6: Show customers where their refund is

Once returns are flowing, "where is my order?" becomes "where is my refund?". Prevent it the same way:

• confirm when the return is created, with the label and the timeline;
• notify when the parcel is received and inspected;
• notify when the refund is issued, with how long the bank usually takes;
• make return and refund status visible in chat, so a bot can answer it.

Step 7: Learn from return reasons

Return reasons are free product feedback. Review them every month:

• "Too small" on one product means the size guide or product description is wrong.
• "Not as pictured" points to photos or colours that need fixing.
• "Arrived damaged" points to packaging or a carrier problem.

Fixing the top three reasons reduces returns, which reduces tickets, which is the cheapest support improvement there is.

Common mistakes

• Hiding the returns policy to discourage returns. Customers ask instead, then buy less next time.
• Requiring an email to start every return. It doubles the work for both sides.
• Inconsistent exceptions. Customers compare notes on social media.
• Slow refunds with no updates. They create the angriest tickets of the year.
• Ignoring gift returns. Give gift recipients a way to return without the buyer's order number.

Before the January wave

Returns season starts with the orders you ship in November, so prepare in the autumn, before the peak takes over:

1. October: rewrite the top of your returns page, including your holiday return window, and publish it before Black Friday.
2. Early November: set up or tidy your returns portal and link it from every order email.
3. Before Black Friday: build the returns flow in chat, train your AI agent on the policy, and write the exceptions table for agents.
4. December: add refund status notifications and plan staff for the first two weeks of January.

Returns start with delivery. If order status questions are still filling your inbox, read how to reduce "where is my order?" tickets. To compare returns platforms, see alternatives to Loop Returns.