By: Ram on Jul 23, 2022
AI/ML for Product Returns and Cancellations in Ecommerce
As a customer, we enjoy having easy return policy while shopping online. And of course, it has played an important role toward ecommerce adoption in India and across the world. Most of the customers read and review return policies before placing the orders online.
As we know, most of the ecommerce businesses in India are not profitable. Flipkart posted 26% increase in Revenue for FY21, but the losses went up by 49% YoY (source). Order cancellations and returns involve various operational costs to the ecommerce business. Controlling these costs without impacting customer experience is one of the important focus areas for the ecommerce business.
Multi-functional teams – Product, Data Science/Analytics, Tech, Category, Customer Service (CS), and Seller/Delivery Partner teams to operations team work together to set up frameworks, algorithms, and processes to manage the costs around Returns and Cancellations.
Before we get into details around role of Data Science & Machine learning team in manging these costs, let’s look at the overall customer journey.
A customer explores products, decide to buy and add the products to the cart and then confirm the order – after selecting the delivery address and making the payment (in case of pre-paid order).
Once the order is placed, remaining steps of order journey start and customer waits to receive the products in most of the cases. Some of these steps are – order gets allocated to a store or warehouse after a few checks, seller team packs the orders, assign the parcel to delivery partner and the delivery partner handovers the parcel to the customer in the shortest time.
But in some of the cases, the journey is affected and that involves the costs we have mentioned above.
Costs and Challenges involved in each of the above stages of order cancellations and returns
Key questions for business are to identity orders that will fall into each of the above segments or genuine order and then define action strategy to manage the % of orders for each of these segments. And this must be done even before confirmed to the customers and sellers.
Can you identify attributes that can help us in putting an order to the above segments or a genuine order?
In the next blog, we will pick one of this and find ways to improve operational efficiencies leveraging AI/ML.