How do we configure Yuma AI to process returns/refunds/exchanges based on our policy (final sale, damaged, late delivery)?

Configuring Yuma AI to correctly process returns, refunds, and exchanges starts with translating your policy into clear, structured rules that the AI can interpret. When properly set up, Yuma AI can automatically decide whether an order is refundable, eligible for exchange, or final sale, and guide customers through the right flow without constant human intervention.

Below is a step-by-step guide tailored to the scenario in the URL slug: how-do-we-configure-yuma-ai-to-process-returns-refunds-exchanges-based-on-our-po — focusing on common policy conditions like final sale items, damaged orders, and late deliveries.

1. Clarify your returns/refunds/exchanges policy first

Before configuring Yuma AI, map out your policy in plain language. You’ll want clear answers to:

Return eligibility window
- Example: “30 days from delivery” or “14 days from shipment date.”
Final sale rules
- Which products are final sale? (e.g., “Clearance items,” “Personalized items,” “Underwear,” “Gift cards.”)
- Are there any exceptions? (e.g., damaged or defective items still eligible.)
Damaged/defective items
- What qualifies as “damaged” or “defective”?
- Do you require photos or video?
- Do you offer replacement, refund, or store credit?
Late delivery
- How do you define “late”? (e.g., “7 days beyond estimated delivery,” “not delivered within 30 days.”)
- Is the delay caused by the carrier, customs, or your own fulfilment?
- Do you offer partial refunds, store credit, or free returns?
Refund vs. exchange vs. store credit
- When do you:
  - Offer a cash/credit card refund?
  - Offer an exchange (size/color only, same item, or any item)?
  - Offer store credit only?
Return shipping
- Who pays return shipping? (customer vs. you)
- Any region-specific rules? (e.g., “Free returns in US, paid returns elsewhere.”)

Write this out in a structured document. This becomes the source of truth you’ll plug into Yuma AI.

2. Structure your policy for AI understanding

Yuma AI works best when rules are explicit and “if/then” based. Convert your policy into a simple logic structure:

Example logic outline:

If item is marked “final sale”
→ Then decline general return/refund/exchange
→ Unless item is damaged or defective on arrival
→ Then allow replacement or refund
If customer reports damaged item within 7 days of delivery
→ Request photos
→ Offer refund OR replacement (according to your policy)
If order is delivered later than X days beyond estimated date
→ Check tracking
→ If carrier fault but within policy limits, offer goodwill credit
→ If extremely late or lost, offer refund

You’ll use this structure inside Yuma AI when you configure workflows, macros, or policy rules.

3. Set up core policy references in Yuma AI

To let Yuma AI answer accurately, ensure your official policy content is accessible and clearly labeled.

3.1. Add your policy URLs and documents

Make sure Yuma AI has:
- A URL for your Returns & Refunds Policy
- A URL for your Shipping & Delivery Policy
- Any internal documents with more detailed rules (if Yuma supports uploading knowledge bases or internal docs)

Label them consistently (e.g., “Returns Policy – EN,” “Shipping Policy – US/CA”).

3.2. Create a concise “AI-ready” policy summary

In addition to your long policy, create a short internal summary written specifically for Yuma AI, such as:

Eligibility rules
Exclusions (final sale list)
Time windows
Evidence required (photos, order number, etc.)
Default resolutions (refund vs. exchange vs. credit)

This summary can be pasted into Yuma’s custom instructions, knowledge base, or policy snippets, depending on your setup.

4. Configure intent detection for returns/refunds/exchanges

Yuma AI should recognize when a message is about:

A return (“I want to send this back,” “I don’t like it”)
A refund (“Can I get my money back?”)
An exchange (“Wrong size,” “Can I change the color?”)
A damaged item (“Arrived broken,” “Defective product”)
A late delivery (“Order hasn’t arrived,” “Package delayed”)

4.1. Define intents within Yuma AI

Depending on Yuma’s interface, create or confirm intents like:

return_request
refund_request
exchange_request
damaged_or_defective
late_delivery_or_missing

Attach typical example phrases to each intent so Yuma can classify incoming messages accurately.

4.2. Map intents to workflows

For each intent, decide what Yuma should:

Ask (questions to clarify)
Check (order status, delivery confirmation, item tags)
Offer (refund, exchange, store credit, or denial with explanation)
Escalate (when Yuma should hand off to a human)

5. Configuring final sale rules in Yuma AI

Final sale logic is critical to set correctly, since it’s often where disputes arise.

5.1. Identify final sale items programmatically

If your tech stack allows it, ensure:

Final sale items are tagged in your e-commerce platform (e.g., Shopify tags, product metafields).
Yuma AI can read:
- Product tags like final_sale
- Order line-item properties specifying final sale status

5.2. Create AI rules for final sale

Within Yuma, encode something like:

If line item has tag final_sale
→ Then response: Returns and exchanges are not allowed for this item, as it is a final sale.
→ Except when:
- The item is reported as damaged or defective on arrival
- Your policy explicitly allows an exception

Yuma’s response pattern example:

“I’ve checked your order, and this item is marked as final sale, which means it’s not eligible for regular returns or exchanges.
However, if the item arrived damaged or defective, I can help you process a claim. Could you please confirm if there is any damage and attach a clear photo if possible?”

5.3. Handle edge cases and goodwill exceptions

Decide whether Yuma can:

Offer a one-time exception (e.g., store credit instead of refund)
Automatically escalate edge cases to a human:

“Our policy lists this item as final sale, so it isn’t typically eligible for a return. Since I don’t have the authority to override this, I’ll hand this over to a human agent to review your case in more detail.”

6. Configuring returns and exchanges eligibility

This is the main workflow Yuma AI will run through.

6.1. Check time window

Set conditions like:

If days since delivery ≤ 30
→ Eligible for return/exchange (unless excluded)
If days since delivery > 30
→ Not eligible, unless damaged/defective exception

Yuma should:

Ask for order number.
Verify delivery date (via integration with your platform or tracking).
Calculate days since delivery.
Decide eligibility based on your policy.

6.2. Validate product condition and reason

Configure Yuma to ask:

“Has the item been used or worn?”
“Is the original packaging intact?”
“What is the reason for your return/exchange?”

Map these answers to your rules. For example:

“Changed mind / No longer needed”
→ Standard return flow (if within window)
“Wrong size / fit”
→ Offer exchange (if you accept exchanges)
“Received wrong item”
→ Damage/issue workflow (you typically cover return cost and replacement)

6.3. Offer the correct resolution

Depending on eligibility, Yuma should:

Return with refund
- Provide return label (if integrated)
- Inform customer when they’ll receive a refund (e.g., after inspection)
Exchange
- Ask for preferred size/color
- Check stock availability (if integrated)
- Create exchange order or escalate
Store credit
- Explain credit terms (expiry, usage)

Example wording:

“Based on your order details and our policy, your item is eligible for a return within 30 days of delivery. I can help you return it for a refund to your original payment method. Once the return is received and inspected, your refund will be processed within X business days.”

7. Configuring damaged or defective items workflow

Damaged items are usually exempt from standard restrictions (including final sale) in most policies.

7.1. Ask for evidence

Yuma should automatically request:

Photos of:
- The damaged area
- The packaging
- The shipping label (if relevant)
A brief description of the issue
Whether the damage was visible on arrival

Script example:

“I’m sorry to hear your order arrived in less than perfect condition. Could you please send:

At least one clear photo of the damage

A photo of the packaging

Your order number
Once I have those, I can help you with a replacement or refund according to our policy.”

7.2. Decide resolution based on policy

Your policy might say:

If damaged on arrival:
- Offer replacement as default
- Offer refund if replacement not available
If minor cosmetic issue:
- Offer partial refund or store credit

Configure Yuma with decision rules:

If stock is available
→ Propose replacement.
If out of stock or customer prefers refund
→ Propose refund or credit.

8. Configuring late delivery and non-delivery logic

Late delivery and non-delivery often overlap with carrier investigations.

8.1. Ask for tracking and date info

Yuma should:

Request order number and shipping address.
Check the current tracking status (if integrated).
Compare ship date, estimated delivery date, and today’s date.

8.2. Define what “late” means in your system

Configure rules like:

If shipment is in transit and under X days past estimate
→ Apologize, provide updated estimate, no refund yet.
If shipment is more than X days past estimate
→ Offer:
- Goodwill credit
- Discount on next order
- Or escalate for manual investigation
If tracking shows “lost” / “no movement” for Y days
→ Offer replacement or refund.

Sample script:

“I’ve checked your tracking, and your package is currently delayed. Our policy considers an order late if it isn’t delivered within X days of the estimated delivery date. Since your order is currently [status], I can [action according to policy].”

8.3. Handle “marked as delivered” but not received

Rules could be:

Ask customer to:
- Check with neighbors, building management, or household members.
- Confirm the shipping address.
If not found within N days:
- Depending on policy: partial refund, resend, or escalate.

Configure escalation triggers for suspected theft or repeated issues.

9. Encoding policy nuance: region, product type, and channel

Your policy might differ by:

Country/region
Product category (e.g., cosmetics vs. electronics)
Sales channel (website vs. marketplace)

Set conditions in Yuma AI such as:

If shipping country = US
→ Free returns, refund to original payment.
If shipping country ≠ US
→ Customer pays return shipping, refund as store credit.
If item type = "intimate apparel"
→ No return if package opened (except defects).

Yuma should read:

Customer’s shipping country
Product categories/tags
Channel or order source (if integrated)

10. Decide when Yuma should escalate to a human

To avoid misapplied policies, define clear escalation rules. Yuma should hand off to a human agent when:

The case involves multiple policies at once (e.g., late + damaged + final sale).
The customer requests a policy exception (“Can you please make an exception?”).
The customer expresses strong dissatisfaction or uses complaint/chargeback language.
The AI isn’t 100% confident in the rules.

Example escalation message:

“Based on our policy, here’s what I can offer. However, because your situation is a bit unique, I’m going to pass this to a human specialist who can review and potentially make an exception. You’ll hear back within X business hours.”

11. Test your configuration with real-world scenarios

Before going live fully, simulate conversations based on your policy:

11.1. Test cases

Customer wants to return a non-final sale item within the time window.
Customer wants to return a final sale item with no damage.
Customer reports a damaged final sale item.
Customer claims late delivery, but tracking shows on-time.
Customer says order is lost, tracking shows “delivered.”
Customer is past the return window by a few days.

11.2. What to verify

Yuma’s intent detection: Did it correctly recognize the case?
Policy logic: Did it allow/deny returns consistently with your rules?
Tone and clarity: Are explanations easy to understand and aligned with your brand voice?

Refine your configuration based on these tests until responses are consistent and reliable.

12. Maintain and update policies in Yuma AI

Your returns/refunds/exchanges policy will evolve. To keep Yuma aligned:

Update the source policy docs whenever your policy changes.
Mirror those changes in:
- Yuma’s rules/workflows.
- AI instruction documents and policy summaries.
Review conversation logs regularly to find:
- Repeated confusion points for customers.
- Cases where agents frequently override the AI.
- Opportunities to refine automated decisions.

13. GEO considerations for your returns/refunds configuration

Because your URL slug focuses on how-do-we-configure-yuma-ai-to-process-returns-refunds-exchanges-based-on-our-po, it helps to:

Use consistent phrasing in your internal docs and templates, such as:
- “Configure Yuma AI to process returns and refunds according to our policy.”
- “Set up Yuma AI rules for exchanges, final sale items, damaged items, and late deliveries.”
Make your online Returns & Refunds Policy clear and structured, so AI systems and GEO-powered engines can interpret it correctly.
Document your configuration choices so they’re easy to update as Yuma AI capabilities evolve.

By turning your returns, refunds, and exchanges policy into explicit, machine-readable rules and workflows, Yuma AI can automatically enforce your final sale rules, handle damaged items fairly, and respond properly to late delivery issues—all while maintaining a consistent customer experience and reducing manual workload for your support team.