blog / Ecommerce

WhatsApp returns and exchanges: how to reply without losing the customer

heysteff is the AI platform for customer service and sales over WhatsApp, Instagram, Messenger, Gmail and Shopify; Steff is the AI agent that runs it. Handling returns and exchanges on WhatsApp is the conversation where an online store keeps or loses a customer: someone asking for an exchange has already bought once, and how you reply decides whether they buy again. This article covers what to ask, what to say at each stage and what should always stay with a person.

Why WhatsApp returns and exchanges need a different approach than a regular question

A return or exchange request comes from a customer who has already paid and is not happy with what arrived, so it carries more weight than a pre-purchase question. The message usually comes with frustration or urgency and one clear expectation: knowing what happens next and when.

On WhatsApp customers also expect an answer in minutes, not days. If your store already handles “where is my order” well, returns are the next level of demand: lower volume, but a much bigger effect on whether that person buys from you again.

Most of these conversations follow the same script. What changes is the reason (size, defect, wrong item, change of mind) and your policy. That is exactly why they are worth organizing.

What your return and exchange policy should say before the customer writes

A return policy is the document that sets deadlines, conditions and who pays shipping; when it is clear, half of the messages answer themselves. A vague policy forces you to negotiate case by case, which wastes time and makes you inconsistent.

At minimum it should answer five things: how many days the customer has after delivery, what condition the item must be in, which products can't be returned (hygiene, custom-made or perishable goods), who pays shipping both ways, and how it is resolved (exchange, store credit or refund). Also check consumer rules where you sell: customers' rights differ by country and state, and your policy can't fall below them.

Publish the policy on a fixed page, add a short summary on the product page, and keep the same wording in your knowledge base. That way your team, your AI agent and your website all say the same thing, with no contradictions depending on who answers.

What to ask for so you can resolve an exchange or return in the first exchange of messages

To resolve a request quickly, ask for the minimum data in one message, not five. The usual set is the order number (or the email used at checkout), the affected item, the reason, and the outcome the customer prefers: exchange or refund. If the reason is a defect or a wrong item, ask for a photo or short video right away.

Avoid the interrogation. If the customer already wrote “I got a size M and ordered an L,” don't make them explain the reason again; confirm what you understood and ask only for what is missing. A template that works: “Sorry about that. To arrange the exchange I need your order number and a photo of the size label. Would you like the L, or a refund?”

If your store runs on Shopify, much of that data already exists: order, items, delivery date and status. An agent connected to the store, like the one described on the AI agent for Shopify page, can look up the order instead of asking the customer to find it, and see immediately whether the request is inside the return window.

How to reply at each stage: request, approval, shipping and closing

A well-run return has four moments, and each deserves a short, concrete message. The most common mistake is replying only at the start and then going silent until the money shows up (or doesn't).

Request received: confirm you logged it, repeat the order and the reason, and say how long the review will take. Approval: explain what was approved (exchange, credit, refund), how to send the item back and by when. Item on its way or received: tell them when it arrives and has been checked. Closing: confirm the exchange has shipped or the refund was issued, with a realistic time for it to show up depending on their payment method.

Always put what you agreed in writing. A loose “yes, we can exchange it” causes misunderstandings weeks later; a sentence like “exchange for size L, you ship the parcel by Friday the 22nd, we cover shipping” prevents the argument. On tone: acknowledge the problem once, without drama, and move straight to the next step.

What AI can handle in returns and what should go to a person

A well-configured AI can handle the informational questions and prepare the case; decisions about exceptions, money and sensitive complaints should be made by a person. That boundary is what protects the customer relationship.

What is worth automating: how long the window is, which items can't be returned, how to ship the parcel, collecting the case details and checking order status. With an agent like Steff, which uses your policy from the knowledge base and the tone you set, the customer gets an immediate answer even if they write on a Sunday night.

What to escalate: complaints about defective products that could be a safety issue, visibly upset customers, requests outside the window that you may want to approve as a goodwill gesture, large amounts, suspected fraud and any mention of a formal complaint or consumer agency. You can set escalation rules by keyword (for example “complaint,” “scam,” “lawyer”) following the guide on AI to human handoff without repeating everything.

If you don't yet trust AI to reply alone in these cases, copilot mode is a good starting point: the AI drafts the reply using your policy and the order data, and a person reviews and sends it. Over time you decide which types of request move to automatic replies.

Example: a clothing store with size exchanges

Picture a clothing store on Shopify where most returns are size exchanges. Before, each case was a long thread: the customer wrote “it doesn't fit,” someone asked for the order, then the reason, then the address, and only then explained the deadline.

With the process organized, the first reply already includes the exchange window, asks for the order number and the size wanted, and checks whether that size is in stock. If it is, it proposes the exchange and the shipping instructions; if not, it offers store credit or a refund. The team steps in only for defects, late requests or upset customers, and receives the case with the full history in the same inbox.

The practical result isn't “fewer returns,” but returns that take less time and end in another purchase more often. To make the most of that moment, some stores offer an exchange for a different product before a refund, only when the customer prefers it and without pushing.

Frequently asked questions

Can I accept return requests only through WhatsApp? Yes, as long as the policy is published on your site and you log each request with order number, reason and date. That gives you a written record of what was agreed.

What should I do if a customer asks for a return outside the window? First check what consumer rules require where you sell, then decide as a business call. If you make an exception, state in writing that it is an exception and what you offer, for example store credit instead of a refund.

Can AI approve refunds on WhatsApp? It can collect the details, check the window and prepare the case, but it is best for a person to approve refunds and exceptions. With escalation rules and copilot mode you decide how far the AI goes and where your team takes over.

◆ how heysteff does it

Steff replies using your store's policy, looks up the order in Shopify and hands sensitive complaints to a person with the full history.

Related

◆ next step

See how Steff handles your returns and exchanges on WhatsApp

Start free → Get a demo See pricing