Online Retail Customer Service
AI Chatbot for Online Stores: Help Customers Exchange the Wrong Size
Customers asking to change sizes? Learn how clearer exchange replies and AI-assisted enquiries can help your online-store team resolve the right request.

Reply.my Editorial Team
Practical guidance reviewed for clear implementation steps, responsible AI boundaries and honest claims. How we prepare our guides.

‘I love the shirt, but it is too small. Can I change it?’ Your customer has not rejected the product. They are asking whether buying from your shop can still work out. Yet the next reply is often a long policy, a request to send everything again, or a sales message for another item.
For an online-store owner, size exchanges can become a surprisingly large part of the inbox. One employee finds the order, another explains the conditions, and someone else has to check the replacement. The customer sees only the waiting—and wonders whether anyone understood the request.
An AI chatbot can help explain an online store's approved exchange information, clarify which item the customer means and pass unresolved questions to staff. It should not promise eligibility, replacement stock or a refund without reliable information and the appropriate approval. The aim is a useful next answer, not an automatic yes.
A customer who wants another size is asking a different question
An exchange request is not necessarily a refund request or a complaint about quality. Start with the outcome the buyer describes. If they like the product and want another size, explain the relevant exchange route before offering alternatives they did not ask for.
The distinction is practical. Shopify's returns and exchanges guidance separates receiving an item back, sending a replacement and refunding payment. These are related activities, not interchangeable outcomes. The documentation does not establish your store's policy or mean Reply.my performs those actions automatically.
A useful opening might be: ‘Which item would you like to exchange, and which size would you prefer?’ Ask for the order reference through your usual private support process when needed. If those details are already in the conversation, acknowledge them instead of starting the questionnaire again.
Listen for a different problem, too. ‘Too small’ could mean the customer chose the wrong size—or that the delivered label differs from the item ordered. A damaged or incorrect item needs the appropriate review, not an automatic reply about change-of-mind conditions.
Read the exchange conversation from the buyer's side
This fictional example describes an online clothing shop in Malaysia. The products and circumstances are invented, not a customer testimonial or a statement of Reply.my results.
Maya orders two tops. She messages on Instagram: ‘The blue one is tight. Can I swap for L?’ The shop responds with its full returns policy. Maya then asks on WhatsApp whether she must return both items. A different employee asks what she bought.
Nobody has refused to help, but the conversation has moved backwards. The important detail is that Maya wants to keep one top and exchange the other. Requiring her to interpret a general policy has not answered that question.
A more useful reply begins with the actual request: ‘You would like to keep the other top and request a size L for the blue one. We need to check the exchange conditions and replacement availability before confirming. Have I understood correctly?’ This is an example of wording to adapt, not a promise that every exchange can be accepted.
Suppose size L is unavailable. The next question is whether Maya wants to hear about another colour or wait for a staff update—not whether the business can quietly replace her choice. If she needs the top for an event, that timing may change her preference. Ask rather than assume.
The helpful part is not the length of the response. It is that each message moves the same request forward. By the time an employee makes the decision, the customer should not need to explain again which item she wants to keep.
Answer the uncertainty, not every paragraph of the policy
Customers usually need a small part of your policy at a particular moment. Someone asking where to send an approved return needs the correct instructions. Someone asking whether an exchange is possible first needs an explanation of the relevant conditions and any checks still outstanding.
Keep the complete policy accessible, but explain the applicable point in plain language. Use your current, approved terms; do not let an assistant invent a deadline, a postage charge or an exception. If staff need to review the circumstances, say what they are checking.
These four questions often reveal where the uncertainty sits:
- Which purchased item is involved, and what replacement does the buyer want?
- Has the request merely been received, or has the store approved the next step?
- Does the customer know the approved return instructions and any applicable costs?
- Is the requested replacement confirmed, or is availability still being checked?
Do not let ‘available’ become a promise you cannot keep
A product page showing size L is not the same as a replacement reserved for a particular exchange. Stock may need to be checked against the exact style, colour and location. A general answer about the range cannot settle that individual request.
The same caution applies to timing. Receiving a return request does not mean the returned item has arrived, been reviewed or triggered a replacement dispatch. Tell the customer which stage is known and which is not. Avoid a reassuring delivery date that nobody has checked.
If the customer asks about a parcel already sent, that becomes an order-status enquiry. Keep the original shipment and the proposed replacement distinct so a tracking link does not appear to answer the wrong question.
For stores selling through several storefronts or marketplaces, establish where the purchase was made before giving instructions. The conversation channel does not tell you which order process applies. A customer can ask on Messenger about a purchase made somewhere else.
Gift exchanges deserve particular care. The person messaging may not be the original purchaser. Explain the appropriate verification route without revealing the buyer's private details or sending refunds to a newly supplied account on the strength of a chat alone.
What would Reply.my take off your support team's plate?
The relevant starting point is repetitive conversation work: explaining approved information, asking for missing details and getting an unresolved request to a person. Reply.my's published plans include FAQ assistance, customer fields and human handover. An exchange-enquiry proposal should specify how those capabilities apply to your store.
Reply.my is an omnichannel service, not a WhatsApp-only chatbot. If shoppers contact you through Instagram, WhatsApp, Messenger or Telegram, confirm which channels and accounts your scope includes. A shared conversation workspace can help staff continue enquiries, but it should not be treated as proof that two profiles belong to the same person.
Keep the commercial question specific: can the proposed service reduce repeated explanations while preserving the customer's requested item and unresolved question? Ask to see an example using your own approved policy, with personal information removed.
Do not assume it includes live stock lookup, return labels, automatic refunds, warehouse updates or replacement reservations. Those require separate confirmation of systems, permissions and approval rules. The human-handover guide explains what makes the next employee's reply useful.
A better exchange experience may start with a better sentence
Before investing in more software, read a few completed exchange conversations. Did the buyer know what to send back? Did staff act on the correct variant? Did anyone promise an update that never arrived? These questions show whether the problem is unclear information, scattered conversations or a decision waiting on another team.
If exchange volumes are small, clear templates and one responsible employee may be sufficient. Where the same explanations repeat across many chats, compare Reply.my's plans against that actual workload—not a general promise to automate customer service.
A helpful exchange reply cannot guarantee a repeat purchase. It can avoid making a customer who still likes your product feel that they must start from zero to get help.
If your team keeps reopening chats to work out which size the buyer wanted, bring an anonymised example to Reply.my. We can discuss the enquiry-handling scope around that problem, including where your staff need to stay in control.
Discuss My Store's Exchange Enquiries