AI Smart Exchanges automatically surfaces intelligent, context-aware variant exchange recommendations directly in Loop's shopper portal. This article explains how AI Smart Exchanges works, how to enable it, and how to customize the experience for your shoppers. Whether you're trying to increase exchange rates, reduce refunds, or improve the shopper experience, AI Smart Exchanges uses return reason signals and historical data to guide shoppers toward the right variant — with no manual configuration required.

Note: This feature is still in beta. Please reach out to your Merchant Success Manager or support@loopreturns.com if you'd like to join.

Important: This feature is currently being tested to measure its impact on exchange rates and revenue retention. When enabled, smart exchanges will be shown to 50% of shoppers, while the other 50% will see your standard return experience.

AI Smart Exchanges is a feature that uses Loop's machine learning models to recommend the best-fit variant to a shopper during the exchange flow. Instead of defaulting to the first available variant, Loop analyzes why the shopper is returning the item and surfaces a personalized recommendation — like sizing up, suggesting a different color, or selecting an exact replacement for a damaged item.

Recommendations appear directly in the shopper portal, either as a nudge on the decision page (for merchants on Flow A) or on the product details page (for all eligible merchants).

Without intelligent exchange recommendations, shoppers are left to navigate variant options on their own — often defaulting to a refund when they don't know which size or color to choose. AI Smart Exchanges addresses this by:

• Reducing refunds and recovering revenue — Relevant, well-timed recommendations give shoppers confidence to exchange instead of refund.

• Improving the shopper experience — Recommendations are personalized to each shopper's return reason, reducing friction and confusion.

• Requiring zero merchant configuration — Loop's ML models generate recommendations automatically using your existing product catalog and return data.

AI Smart Exchanges is most impactful for merchants where fit and sizing are common return drivers. Here are a few examples of how it works in practice:

Size exchange: A shopper returns a medium jacket because it runs small. Their return reason is "Too Small." Loop recognizes the fit issue and pre-selects a size Large on the product details page.

Multi-dimensional variant: A shopper returns jeans that fit in the waist but were too long. Loop uses both the return reason and variant dimension data to surface the same waist size in a shorter inseam — rather than simply sizing up across the board.

Replacement: A shopper received a damaged item and selected "Item arrived damaged" as their return reason. Loop recognizes this as a replacement scenario and pre-selects the exact same variant the shopper originally ordered.

Color variant: A shopper returns a shirt in blue and notes in their comment that they wanted a darker color. Loop detects the color preference signal and surfaces the same size in a darker available color, prioritizing in-stock variants.

When a shopper initiates a return, Loop looks at three inputs to generate a recommendation:

1. Return reason — both the reason the shopper selected and any free-text comment they left (e.g., "too small" or "wanted a lighter color")

2. Historical exchange data — if other shoppers have exchanged the same product for a similar return reason, that pattern is factored in

3. Available variants — Loop identifies which variants best address the shopper's specific return reason and ranks them

The top-ranked option is pre-selected in the exchange UI, and a rationale is shown to the shopper (e.g., "size up," "try a lighter color," "straight replacement"). The system is also smart enough to handle combinations — if a shopper says the item was too small and they wanted a different color, Loop will size up and suggest an alternative color.

Note: Not every product will receive a recommendation. Loop only surfaces a recommendation when it has sufficient confidence in the suggestion — this ensures recommendations are genuinely helpful rather than adding noise to the experience.

Follow the steps below to enable AI Smart Exchanges:

1. Navigate to Returns Management > Shopper experience > Smart exchanges

   

2. Select the 'Join the beta' button and fill out the Google Form. Someone will reach out to you shortly after submission to finalize setup!

Once AI Smart Exchanges is enabled, recommendations will appear automatically in your shoppers' exchange flow — no additional configuration is needed. Reminder that the beta will operate as a 50/50 A/B test where half of your shoppers will see recommendations and half will not.

Shopper portal — Decision page nudge

Shoppers will see a nudge on the decision page encouraging them to consider an exchange before they reach the product details page. This early touchpoint is designed to increase exchange consideration before shoppers commit to a refund.

Shopper portal — Product details page

Shoppers will see the recommended variant pre-selected on the product details page, along with a brief rationale label explaining why that variant was suggested.

Customizing recommendation styling

Merchants can optionally customize the color of the recommendation UI components to match their brand.

1. Navigate to Returns Management > Shopper experience > Portal customizations > Smart Exchange Styling

   

2. Update the brand color settings and click Save.

Does AI Smart Exchanges require any manual setup or rules configuration? No. Recommendations are generated automatically by Loop's ML models using your existing product catalog and return data. There is no rules engine or manual configuration required.

Which merchants are eligible for AI Smart Exchanges? AI Smart Exchanges is available to merchants who have variant exchanges enabled. It works best for merchants where sizing and fit are common return drivers.

Will every product get a recommendation? No. Loop only surfaces a recommendation when its models have sufficient confidence in the suggestion. If no strong recommendation exists for a particular product and return reason, the exchange flow will proceed without one. This is intentional — it's better to show no recommendation than an irrelevant one.

What happens if the recommended variant is out of stock? If the recommended variant is out of stock, a recommendation will not be shown for that product and we may recommend a different one instead if there's another good fit for the shopper.

What types of exchanges does this feature use? AI Smart Exchanges currently supports variant exchanges only. Advanced exchanges are not in scope at this time.

Do translations work for Smart Exchanges? Yes, any copy shown in the shopper portal will automatically translate if it's a language we support.

Please reach out to support@loopreturns.com with any additional questions.