YundaDeskExchange requests look calmer than refund requests, but they can jam a support queue just as fast.
Wrong size, color mismatch, or a customer asking for another SKU: these cases arrive every day. The hard part is checking policy, inventory, price difference, logistics, and customer emotion at the same time. A useful exchange request script works like this: AI checks eligibility from the knowledge base and asks for missing information; price differences, inventory holds, reshipments, refunds, and exceptions go to a human.
Check eligibility first: do not promise too early
When a customer says “I want an exchange”, answering “sure” too fast creates risk. Eligibility may depend on order date, item condition, category restrictions, campaign rules, packaging, and return-shipping cost.
We have received your exchange request. I will first check whether it matches our current exchange policy. Please send your order number, the item/size/color you want to exchange to, and photos showing the current item condition. If this involves a price difference, inventory, or special handling, I will pass it to a human agent for confirmation.
I understand you want the right item as soon as possible. To avoid an inaccurate promise, I will first review the order status and after-sales policy. Anything requiring human approval will be handed over with the full context.
Size exchanges: acknowledge first, then collect facts
Size issues are common and easy to turn into reusable scripts. Avoid blaming the customer, but still check policy before promising an exchange.
| Scenario | Exchange script |
|---|---|
| Size runs small | Sorry the size did not fit as expected. I will first check whether this item supports a size exchange. Please send the order number, current size, desired size, and whether the item is unused with tags and packaging intact. |
| Size runs large | I understand you want a better fit. Please send the order number and the size you want. I will check the after-sales window, item condition, and inventory first. |
AI can suggest a likely size from the knowledge base, but it should not decide for the customer. The final exchange choice still needs customer confirmation.
Color swaps: do not argue about what is “normal”
Color-mismatch conversations escalate quickly. The customer sees the physical item; the brand sees lighting, screens, photography, and batch variation. Starting with “this is normal color difference” sounds like blame-shifting.
Sorry the color you received did not match your expectation. Please send your order number, photos of the item, a photo of the outer label, and a screenshot of the product page you compared it with. I will check the batch, color option, and current exchange policy first.
Color differences need to be reviewed against the product page, photo conditions, and physical batch. If this involves exchange, reshipment, or any fee adjustment, I will send it to a human for approval.
| Avoid saying | Say this instead |
|---|---|
| This is normal color variation, so it cannot be exchanged | We need to review item photos and order details first, then confirm whether the exchange policy applies |
| Product photos are for reference only | When the page and received item feel different, we will review it carefully. Please send photos first |
The point is to move from argument to facts, policy, and the next decision owner.
Price difference and inventory: AI can flag, not execute
Exchange requests often raise two operational questions: how to handle a price difference, and whether the target SKU is in stock. This is where small mistakes become expensive.
The item you want may involve a price difference. I can record the target item and current order information first, but the final amount, payment method, or difference return arrangement needs human review.
I will first check whether the target size/color appears available. Inventory can change as orders come in, so whether it can be reserved or reshipped needs confirmation by a human agent in the system.
Practical rule: if it affects money or inventory, the conversation should not stop with AI.
That rule can be split into a simple conversation-risk model: let AI handle low-risk information collection, organize context for inventory and logistics uncertainty, and hand off anything involving money, reshipment, or exception approval.
Segment exchange requests by handling risk (illustrative)
Logistics scripts: explain what the customer does and what you do
Cross-border exchange logistics are long. Customers need to know what happens after they ship an item back, without hearing timelines you cannot control.
| Stage | Exchange script |
|---|---|
| Pending confirmation | I will first confirm whether this order qualifies for exchange. Once confirmed, a human teammate will share the return address, packaging requirements, and next steps. |
| Customer ships back | Please keep the tracking number after shipping and send it to us. Once the returned package arrives and passes inspection, we will continue the exchange process. |
| Reshipment | Once the new item ships, tracking information will be shared. Cross-border delivery may be affected by customs, carrier handoff, and local delivery. |
If the customer asks to receive the new item before returning the old one, AI should not approve it directly:
I understand you want to use the new item as soon as possible. Whether we can ship first depends on order risk, inventory, and after-sales policy. I will send your request and order context to our team for human review.
Ready-to-use exchange scripts
Store these by scenario in the knowledge base. Define use, required fields, and handoff triggers.
General opening
We have received your exchange request. I will first check the order and after-sales policy. Please send your order number, the item/size/color you want to exchange to, and photos showing the current item condition.
Likely eligible
Based on the information you shared, this order may qualify for exchange. The next step requires human confirmation of inventory, price difference, and return method. I will pass the current context to our team.
Missing information
We are still missing some information, so I cannot confirm exchange eligibility yet. Please send the order number, item photos, packaging/tag status, and the exact item you want instead.
Not eligible under policy
Under the current after-sales policy, this case does not support exchange. If you believe there is a special situation, please add details and photos, and I will send it to a human agent for review.
Make exchange scripts smarter over time
Writing scripts is only the start. Agent follow-ups, AI corrections, and new rules should become learning suggestions you confirm. After owner approval, they can enter the knowledge base, while staying traceable, testable, and revertible.
Use this checklist for exchange knowledge base entries:
- Define eligible and ineligible exchange conditions
- Separate size, color mismatch, wrong item, and defect handling
- List required fields: order number, target SKU, photos, packaging status, tracking number
- Mark price differences, refunds, inventory holds, and reshipments as human handoff cases
- Keep one policy across languages and channels instead of rewriting it per inbox
If your channels already flow into one workspace, exchange conversations, customer identities, order history, and social IDs can live in the same customer record. For the foundation behind that setup, see what an omnichannel inbox really solves.
Smooth exchange handling is not full automation. It is AI answers first, humans back up: AI checks eligibility, collects information, and guides the customer; humans handle price differences, inventory, approvals, and exceptions.