[the_ad id="83613"]

Thirdwatch leverages AI to prevent fraud for e-commerce companies


Adarsh Jain is an avid e-commerce shopper and has been shopping online since 2005 (Indiaplaza days) when he used to order books online.

Two years ago, he was trying to order a TV for his father who lives in Greater Noida from one of the top two e-commerce sites. Much to his surprise, the order was declined by the company despite being prepaid. However, when Adarsh ordered the same product to his Gurugram address it came through.

On digging deeper, he found that fraud/abuse is a big problem in e-commerce, and companies sometimes use broader rules like blocking certain pin codes/cities/states and/or international credit from ordering which affect genuine users.

The quantum of prevalent fraud in e-commerce parlance made Adarsh think about it. “I saw an opportunity here and began discussing this with a long time friend Shashank Agarwal in late 2015,” said Adarsh.

Given their experience in building big data mobile analytics system the duo felt confident that they can solve this problem by using Artificial Intelligence (AI) and big data. After six months of brainstorming, the duo started ThirdWatch in April 2016.

How Thirdwatch prevents fraud in digital transactions?

ThirdWatch has a sole objective of solving fraud in digital transactions. It uses more than 200 signals on each transaction to generate a red or green flag in real time. “By superimposing data from our own knowledge base and from third-party databases we enhance the data captured from user transaction like zip code, IP address, and mobile number validity databases,” added Adarsh.

It also uses micro models for address profiling, device signature, account profiling and velocity profiling. The company captures the whole user behavior including the time taken to place an order, the device user is using to order, IP address of the user, the location of order amongst others.

“We are very stringent on data security and sign a strong non-disclosure agreement (NDA) with our clients. However, we do use networks effects to multiply the effectiveness of our platform by collating fraud patterns across the clientele base, and inputting values in our algorithm,” said Adarsh.

Adarsh is an alumnus of  IIT-BHU and has over 12 years of experience across various technology companies. He claims to be a part of big data and startups ecosystems since 2008. Shashank is an ethical hacker and programmer from the age of 15. Both founders hold multiple patents in the technology domain.

How is fraud in Indian e-commerce unique from the west?

Fraud in Indian ecosystem is very different from western Geographies. In the west dominant form of fraud is payment fraud as most orders are prepaid. In India, consumers love cash. Cash on Delivery (COD) still constitutes 60-70% orders for any e-commerce player.

Most dominant fraud in India is Return to Origin (RTO) which means while the user has ordered but delivery does not happen, and the items come back to the warehouse. Companies lose money in RTO on forward logistics, reverse logistics and the inventory cost when the item is in transit apart from other operating costs.

Besides RTO, other frauds prevalent are duplicate item return and promo code fraud.

So far traction, competition, and road ahead

Recently, Thirdwatch had raised an undisclosed amount of funding from India Angel Network (IAN). It was the first investment from the new corpus of Rs 450 crore by IAN.

Currently, the company processes about 5,000 transactions every day with a pipeline to reach 50,000 transactions on a daily basis. “Because of our NDA’s and agreements with our client we can’t disclose clients list,” added Adarsh.

Thirdwatch claims to be the only company leveraging AI to prevent fraud in real time in India. In US and Israel there are companies like SiftScience, Signifyd, Threatmetrix, and Forter, however their primary focus on prepaid orders.

Besides e-commerce, Thirdwatch will soon expand into preventing other financial frauds including Insurance and fin-tech fraud.

Thirdwatch: Website

About Author

Send Suggestions or Tips