Published : 3 years ago, on 31/03/2021

By Mairtin O’Riada, CIO and co-founder, Ravelin

If there is one business sector that has thrived during the pandemic, it’s ecommerce. According to eMarketer, worldwide retail ecommerce sales recorded a 27.6% growth rate in 2020, with sales reaching over $4 trillion.

The rapid growth has given rise to new challenges, with the buoyant ecommerce sector attracting an increasingly high level of fraudulent activity. With most ecommerce companies focused on creating frictionless experiences for their customers, including customer support and easy refund policies, they are often less aware of the danger coming from their very own customer base — that of ‘friendly fraud’ and refund abuse.

Not all fraud is committed intentionally, with consumers often being confused or ill-informed rather than malicious. Regardless of intention, each dollar of fraud costs retailers $2.94 in fees in lost merchandise, security and other associated costs. It’s therefore essential for merchants to detect and prevent fraud to protect their business.

Let’s look at two increasingly common types of fraud originating from customers and explore how you can prevent both of them.

Abuse of refund policies

Because of Covid-19, many companies have simplified their refund terms and conditions to drive growth. Simultaneously, they’ve introduced a higher level of risk by inviting people to commit so-called refund abuse, which takes advantage of a company’s refund policy.

According to Ravelin’s Online Merchants Perspectives report, refund abuse is the fastest growing type of online fraud, with just over half of merchants reporting an increase in refund abuse in the past year.

The significant rise in refund abuse may be related to changing delivery patterns, such as the rise in contactless delivery of goods introduced to protect the health and safety of customers and delivery staff. By leaving goods outside the customer’s front door, it may be more difficult for merchants to prove that the goods have indeed been delivered, or, for instance, for food services to provide evidence that a meal was still warm when delivered.

The unfriendly nature of friendly fraud

Friendly fraud differs from refund abuse in that it involves a process of chargeback via the issuing bank associated with a debit or credit card.

This means that a customer orders goods but then disputes the transaction with their issuing bank via chargeback rather than requesting a refund directly from the merchant. Designed to protect the cardholder’s safety, chargeback involves a forced retrieval of funds from the merchant by the issuing bank, which are then given back to the customer.

Managing chargebacks can be a long, costly and resource-heavy process for merchants. The issuing bank or payment service provider will charge and also require a certain degree of admin to provide documentation and/or dispute the chargeback request. PayPal charges a non-refundable fee of $20 whenever a customer files a chargeback. If chargeback requests rise above an acceptable threshold, the bank can simply close the account, leaving the merchant unable to accept card payments.

On occasion, a customer may initiate chargeback without any malicious intent. But, the relative simplicity of this otherwise legitimate mechanism means it is often misused by fraudsters. In fact, it is estimated that friendly fraud makes up anywhere between 60% and 80% of all chargebacks.

The growth of friendly fraud has been confirmed by our research, where around 40% of merchants reported an increase of this form of fraud. Friendly fraud is now the third most common type of fraudulent activity, behind online payment fraud and account takeover.

Tackling fraud with machine learning

Fraudulent customer activity in the shape of friendly fraud or refund abuse is becoming a major concern for merchants that can have a significant impact on their bottom line. To prevent financial loss, they can adopt a number of risk mitigation strategies.

An important measure is to have a strong customer service that communicates proactively. Merchants should also provide real-time delivery tracking, ensure timely delivery and design clear refund policies. Such steps will prevent confusion or inadvertent attempts at fraud.

However, in this time of rapidly growing online transactions, it’s important to supplement customer service measures with proactive monitoring, detection and prevention of fraud. AI and machine learning-powered analytics helps spot unusual patterns by analysing online transactions and customer behaviour.

When it comes to chargebacks, machine learning models have the potential to identify suspicious behaviour and predict fraud prior to the transaction taking place. With refund abuse, they can help identify serial returners and their linked accounts through network analysis. These insights enable merchants to set limits on the number of refunds per customer, or prevent certain customers from requesting a refund for a period of time.

But our research has shown that few merchants are currently looking at payment data and customer activity to determine and prevent fraud. Using machine learning-enabled modelling and link analysis they have an opportunity to unlock new insights into fraudster activity to boost their fraud detection and prevention success.