The fledgling online fashion retailer was no match for the fraudulent ingenuity of the engineering students from Kanpur. The students would order branded clothes, stitch the labels on inferior locally made replicas, and claim a refund on the imitations citing bad fit. It took hundreds of transactions over two months for the retailer to realise what was up, reports ET.
 
According to the report, such frauds and damaged originals account for 10% of all product returns in the domestic online fashion industry, and etailers including Flipkart, Myntra and smaller startups are fighting back with artificial intelligence and other advanced technology. 
 
According to the industry estimate, around 30% of all fashion products ordered online are returned for various reasons. This translates to an estimated loss of about Rs 870 crore per year. This costing include products not delivered because of logistics issues or rejected by a buyer without taking delivery. 
 
Inspite of such huge losses, the fashion industry has no other potions but to allow for returns to encourage customers used to trying on clothes before buying to also shop online. 
 
Flipkart, which along with its fashion units Myntra and Jabong dominates the online fashion market, deploys “machine learning models to predict the intent of the user to accept the order at the time of placing an order… (and) to detect fraudulent or invalid addresses,” Sandeep Kohli, senior director of engineering at the company told ET.
 
“If the model predicts lack  of user’s intent, various levers like asking the customer to pay upfront are deployed. This helps in reducing the return fraud,” Kohli added. 
 
As per the report, Myntra, which allows for ‘no questions asked returns’ and a high return cycle of 30 days, analyses customer data to minimise its rate of product returns.
 
“Based on the customer’s purchase and returns history, we have notifications to inform them in case they are buying the wrong size within a different brand,” chief executive Ananth Narayanan told ET. “This inhouse technology was launched a couple of months ago and we expect this to lead to a 2% drop (in returns).” 
 
The report also says that Myntra’s product return rate is 12- 14% of all orders, according to industry estimates, and another 10-12% of its orders fall under the category of ‘return to original,’ or RTOs, which means these products couldn’t be delivered or were not accepted by the buyer.
 
For Flipkart, overall product returns and RTOs in the fashion category add up to 15-20%, according to industry estimates. 
 
In 2016, Amazon filed a patent application in the US for a 3D modelling system that would make size recommendations to users, according to a news report.