Plans: Currently in Beta version; Available for selected merchant organizations | Platforms: All platforms
Overview
AI Return Reason replaces the return reason dropdown with a short conversation. The customer explains in their own words why they're returning an item. The agent asks a follow-up question when an answer is vague, asks for a photo when a claim needs evidence, and matches the conversation to a return reason you've already set up.
You get the same return reason field you have today, plus the full conversation behind it.
Return reasons page with the AI reasons settings card
Before you start
Applies to | Return requests submitted through your Return page only. POS, API, and TikTok Shop returns keep using reason dropdowns. |
Takes effect | Only on return requests submitted after you turn it on. Existing requests don't change. |
Existing setup | Your reasons, sub-reasons, and reason groups stay as they are. Nothing is migrated. |
How AI Return Reason differs from Reason Options
| Reason option | AI Return Reason |
What the customer does | Picks a reason from a list you built | Answers a question in their own words |
What you configure | A list of reason options per group | One short instruction per group (optional) |
What you get back | The reason they picked | A matched reason, plus the full conversation and photo findings |
Which reasons can be matched | Only the reasons attached to that group | Your entire reason library |
Both modes can run in the same return request. If one item matches a group that uses AI Return Reason and another matches a group that uses Reason option, the customer answers questions for the first item and picks from a dropdown for the second.
Turn On AI Return Reason
Use the setup flow
Go to Return policy > Return reason in your AfterShip Returns admin.
Click Set it up to open the setup modal.
AI reasons setup in Return Reason settings page
Follow the setup flow. You can test a conversation using your store's 10 best-selling products to see what customers will experience before anything goes live.
Complete the flow: Configure the initial question, enable AI suggested answers to help customers fill in the answer, and set the number of follow-up questions you want answered to get better understanding of the return reason.
Click Finish setup. Once activated, AfterShip Returns switches your Default return reason group to AI Return Reason mode.
Set up and test AI reasons, with the preview and Finish setup button
Check which group the setup flow switched. You can switch it back at any time.
Important: Your Default return reason group catches every return that no other group's conditions match. For most stores, that's the largest share of returns. To start smaller, switch the Default group back to Reason option and turn on AI Return Reason for a single, narrower group instead.
Turn it on for one group
Go to Return policy > Return reason in your AfterShip Returns admin.
Open the reason group you want, or create a new one.
Set Collection mode to AI Return Reason.
Click Save.
Reason group with Reason collection method set to AI conversation
Other groups stay in Reason option mode until you change them.
Write a conversation prompt
A conversation prompt is a short instruction for the agent. You can add one prompt to each group that uses AI Return Reason. Customers never see it.
The prompt tells the agent what to find out and how to map answers to your reasons. It's optional: without one, the agent matches using your reason and sub-reason names. A prompt usually improves accuracy, especially when some of your reason names are similar.
AI instructions box with Draft for me and Test
Include | Example |
The goal | "This group is large furniture. Find out whether the box was damaged or the product itself was." |
Matching logic | "Damage found on opening is shipping damage. Damage after use is a quality issue." |
When to ask for a photo, and of what | "For any damage claim, ask for a photo of the damaged area together with the product tag." |
Draft a prompt automatically
Click Draft for me in the prompt box. AfterShip Returns writes a first draft based on the group's name, its conditions, and your reason library. It uses only your store's data.
The draft only fills the box. Nothing is saved until you save the group.
You can edit, keep, or clear the draft.
If the box already has content, you confirm before AfterShip Returns replaces it.
Tip: Describe what the agent should find out, not what it should recite. An instruction like "Ask the customer whether it's A, B, or C" turns the conversation back into a dropdown read aloud.
Settings reference
Find these settings under Return policy > Return reason.
Note: AI return reason settings are not specific to individual reason groups. They apply universally to all reason groups for which AI reasons are enabled.
Setting | Default | Notes |
Initial question | Why are you returning this item? | Keep it neutral. A leading question like "Was the size wrong?" steers customers to the wrong reason. If you customize it, add your own translations in the content center. |
Suggested answers | On | Shows tappable suggestions under the first question, based on AfterShip's standard reasons and the item's details. Turn them off to show a plain text box. Suggestions reduce typing, but customers who tap instead of type give you less detail. |
Max follow-up questions | 2 (range 1 to 10) | Counts follow-up questions only, per item. With a limit of 2, the customer answers at most 3 questions. The agent stops as soon as it has enough information. |
Edit settings window with Initial question, Suggested answers, and Max follow-up questions
Demo
What the customer sees
One neutral question, a text box, and suggested answers (unless you turned them off).
The first question is required, like a required dropdown. There's no minimum length.
Each answer can be up to 500 characters, with the remaining count shown.
The agent may ask a follow-up question based on what the customer wrote.
The agent may ask for a photo and explains what to capture, such as the damaged area, the outer box and label, or the whole item.
After the first question, the customer can end the conversation at any time and continue submitting.
The customer never sees photo findings, and photo findings never block a return.
First question with suggested answers on the Return page
Photo request on the Return page
Review the results
Open a return request in your AfterShip Returns admin to see:
The full conversation: every question, answer, uploaded photo, and timestamp.
The matched return reason, plus a sub-reason when the agent is confident. If the agent is confident about the main reason but not the sub-reason, it records the main reason only.
Photo findings, if the agent requested photos.
You can't edit the conversation or the matched reason, the same as return reasons today. AI Exchange also uses the conversation as context for its exchange recommendations.
Return request with the matched reason and photo verdict on the item
Photo checks
When a claim needs evidence, the agent asks for a photo and tells the customer what to capture. It then reports whether the photo supports the claim.
Verdict | Meaning |
Confirmed | The photo supports the claim. |
Inconclusive | The photo is unusable or doesn't show enough to tell. |
Mismatch | The photo doesn't match the claim. |
The agent also reports item condition as three separate values, each with a confidence level.
Condition | Possible values |
Tag attached | Attached, Removed, Not applicable, Undetermined |
Product condition | Good, Minor damage, Major damage, Functional defect, Wrong item, Undetermined |
Packaging condition | Perfect, Acceptable, Damaged, Not applicable, Undetermined |
AI return reason window with photo assessment, item condition, and conversation details
If a photo is unusable, the agent doesn't ask for another one. It records Inconclusive and lets the customer submit.
Condition findings appear on the return request for you to review. They don't trigger any automatic action.
How findings affect risk review
Inconclusive or Mismatch raises a Medium Risk signal in RMA Review Agent.
Confirmed raises nothing and doesn't lower an existing risk assessment.
A separate setting controls whether these Medium Risk cases are flagged for manual review. Your existing auto-flag setting applies to High Risk only.
How reasons are matched
AI Return Reason only matches reasons that already exist in your reason library. It never creates a new one.
The agent can match any reason in your library, not only the reasons attached to the group the item matched. Groups that use AI Return Reason don't have their own reason lists.
Your reason library supports up to 200 reasons.
If nothing matches, the return reason shows as Uncategorized. This is a system value, so you can't edit or remove it. You'll see it on the return request, in Analytics, in return reason filters and CSV exports, and as a workflow condition.
Languages
Questions and photo instructions follow the customer's language on your Return page. Reason names use your existing content center translations.
If you customize the initial question, add your own translations for it. You can't configure follow-up questions or photo instructions, because the agent writes them in response to each answer.
FAQs
1. Will this change return requests that were already submitted?
No. AI Return Reason applies only to requests submitted after you turn it on.
2. Do I need to change my existing reasons or groups?
No. AI Return Reason uses your reasons, sub-reasons, and groups as they are. Existing groups stay in Reason option mode until you change them.
3. What happens to POS, API, or TikTok Shop returns?
They keep using the reason dropdown from the matching reason group. AI Return Reason applies only to requests submitted through your Return page.
4. Can a customer skip the questions?
The first question is required. After that, the customer can end the conversation and continue at any time.
5. Can I edit the reason the agent matched?
No. Return reasons are fixed once submitted, the same as today.
6. Why did the agent match a reason that isn't in the item's group?
AI Return Reason matches across your whole reason library. Groups that use AI Return Reason don't have their own reason lists.
7. Why is the sub-reason empty?
The agent records a sub-reason only when it's confident. If it's confident about the main reason but not the sub-reason, it records the main reason only.
8. Can I require a photo for a particular reason?
No. The agent decides when to ask for a photo based on what the customer describes. The Required photo setting on individual reasons doesn't apply to AI Return Reason conversations. To guide photo requests, describe when a photo is needed in the group's conversation prompt.
9. Do customers see whether their photo passed?
No. Only you see photo findings.
Need help?
For any further questions or help, please contact our chat support team or reach out to us at [email protected].