Optimizing Order Returns

Predict returns before order placement

Some variables to keep a log as user profiles:

• How many similar items are there in the cart
• Are the items of the right fit
• What has the customer’s return rate been for similar items in the past
• What is the product’s return rate
• Supplier return rate

These insights can be extended to an AI powered classification engine that can make probability of a return. Recommendations to prevent a return include:

1. Offering a targeted promotion
2. Suggesting better suited substitutes
3. Free delivery charges
4. Alternate supplier
5. Recommendation probability as an input for other personalisation algorithms.
6. Even dissuading serial returners from making the purchase

For example, it may increase shipping charges as a deterrent or offer a voucher as an incentive in return for making the purchase non-returnable.

Other descriptive analytics can be summarized for various departments from the model training:

• Forecast returns after order placement
• Link return rate to other business areas like sourcing, marketing, supply
• chain and merchandising.
• Stricter return policies