|
Listen : Audio version of this article
|
Online markets are growing rapidly and are estimated to reach 4 trillion dollars by next year. Customers find online platforms easier to use and time-saving for a lot of reasons, but all the online sites have been facing one issue unanimously: customers returning almost half of the products they buy online, which results in the dropping of sales and profits for these e-commerce sites.
Myntra and Google have come up with a new machine learning model that is trained on a data from its customer’s preferences, product views, and reviews, along with the body shapes of the customers to predict their return probability, right before they purchase any item from the site.
Researchers have conducted various surveys on Myntra which took the customers to answer basic questions, and they have found that out of all the returns, 53% of them occur due to size related issues, 4% because there are similar items in the cart.
They have also concluded that shoppers with more than one item in their cart have returned the items more than 72% and the ones with just one time have returned only 9% of the item, or sometimes their chances of returned slimed down to none.
Insights from this research have resulted in the boost of the algorithm, with a dual model to estimate both cart and item return probability. The data set contained all the information to things like brand, cart size and order day and time, payment mode and purchase frequency. This resulted in the websites to an accurate prediction of their orders and returns.
This Artificial Intelligence has proved itself very accurate in the test runs, In a live test conducted with 100,000 users, the count of orders dipped slightly (by 1.7%) compared with a control set, but the return rate has dropped by 3%. The team looks forward to improving this algorithm and reduce their overall returns.