By Adarsh Jain
Fraud is as old as Commerce, with E-Commerce becoming popular fraud has also adopted to the new digital avatar. Fraud management is an essential component of running an e-commerce business because it is intrinsically connected to a merchant’s bottom line.
Amazon does not ship anything above Rs 7000 to 20% of Indian population i.e. anybody living in Madhya Pradesh, Bihar, Uttarakhand or Jammu and Kashmir cannot order a TV, Washing machine or even a laptop. One of voucher gifting companies was hit by fraudsters with around INR 10 million fraud, they got lucky when one of the main hacker’s associates ratted him out (due to personal jealousy) to company’s CEO else it could have led to existential crisis for the company.
Fraud is bad for the whole ecosystem, merchant, customer and payment gateway. In 2016 alone, e-commerce companies around the world lost an estimated $15 billion to fraud. It always goes under reported as it is negative publicity and can affect customer sentiment and valuation. Dealing with fraud incurs management time loss, money loss and requires attention to deal with fraud.
Some common user frauds popular in India are
- Fake for fun COD orders
- Denial of delivery (Receive the correct item but claim it is fake or brick …)
- Return fake item (Place return request and return cheap replica)
- Payment fraud this can be further classified into
- Order via stolen credit card
- Credit card chargeback
- Data tempering on payment page
Users are mostly interested in doing fraud on items with high resale value like mobiles. Mobile recharge is also very popular with fraudsters.
Challenge is, fraudsters don’t follow any rules and fraud keeps on evolving. Starting from Just for fun fake Cash On Delivery orders fraud has become sophisticated with bots crawling e-commerce sites to exploit any known vulnerabilities. What is required is automatic screening of all orders and no rejection of genuine orders.
What can stop fraud is hidden in the recent successes of AlphaGo (against Lee Sedol) and Libratus (against top Poker players) which use Neural Networks what is popularly known as Deep Learning. Deep Learning can detect features on its own. evolve with more data, work with incomplete information and even recognise when the other side is trying to bluff. To effectively fight fraud we would like to have such a ally on our side.