What’s going on in Brazilian e-commerce? How big is the fraud market and how online retailers can deal with it? Read 5 insights about e-commerce in Brazil provided by our expert in this field, Leandro Oliveira. At Nethone, he's responsible for recognizing high-risk LatAm markets and helping companies to lower their payment fraud rates. 5 questions about e-commerce in Brazil

Currently, there are ~66.4M e-commerce users in Brazil. Until 2021 the number is going to raise by 28.2M. What is their preferred online payment method?

In Brazil, 69% of the online transactions are done with credit cards and 24% are done with “boleto bancário”, which is a type of a bank transfer that allows to process payments without having a credit card or even a bank account. We observed that many customers in Latin America still encounter difficulties to access even basic bank products, such as accounts. E-commerce plaftforms are trying to be flexible and provide new payment methods, but despite that credit cards remain number one.

O Estado de S. Paulo newspaper revealed that Brazil is suffering from 3,6 fraudulent purchases every minute. This number is significant - no wonder that total worth of online fraud and scams in Brazil reaches over BRL 70 billion yearly. The problem is more concerning with electronics, luxury and travel sectors. The goods provided by such companies are relatively pricey and at the same time, they are attractive and easy to resell in the Darknet.

What is the impact of this rate to the merchant?

Firstly, high chargeback rates. When they are soaring, the credit card issuer can increase the transaction commision or, worst case scenario, decide to stop cooperating with the merchant. Also, there are additional costs tied to the chargeback procedure itself.

Secondly, Customer Experience. The competition is high and all the legitimate customers, that for some reasons were treated as fraudsters can easily switch to another shop. Imagine the frustration of a genuine, recurring and loyal customer whose transaction was denied by mistake. There is a high risk that the loss of such customer will not be only one-time, but rather a loss of their lifetime shopping potential.

How do fraudsters target merchants?

In Brazil, the most common fraudulent attempts are preceded by credit card testing. In short, a fraudster obtains a credit card data illegally and shops online to test if it works. According to Visa’s study, this type of fraud covers even 45% of frauds in the country! They test the card with cheap online products, usually worth below BRL50 (around USD15). Once the transaction is approved, they begin to purchase (in this case - steal) more expensive goods through the same platform.

Another relevant type is friendly fraud, and we observe that merchants are suffering a lot from it in the fashion industry. Their customers go through the purchase process, receive ordered products and then claim a chargeback, but never return the goods. In such cases, when the claim has been accepted, the merchant loses total worth of the never-returned products, the amount of money that has been refunded and the additional costs of processing the chargeback.

How Machine Learning-models can mitigate this problem?

The first value of ML models is to look at all available attributes and distinguish the types of customers: the legitimate ones from fraudsters. Genuine customers can be redirected to a different, more smooth purchase process, dedicated to the most trustworthy customers. It provides better customer experience and therefore, encourages them to come back to the online store and buy more. These customers tend to leave positive reviews and recommend the shop to their friends which results in increasing revenues. For merchants the dedicated process also reduces the number of manual reviews, and operational costs associated with fraud prevention.

Another significant value for the merchants is the safe expansion of the business to the new markets. Going internationally, embraced with a solution that investigates user behaviour on your website or app and delivers instant insights, can help prevent fraud in the new environment - in contrast to rule-based systems, ML models easily adapt to new trends.

We must remember that carding is a full-time “job” and fraudsters are constantly seeking for more and more creative, sneaky ways to do the crime. That’s why it’s crucial to stay one step ahead of them and implement intelligent solutions which can protect both business and the customers.


Financed by: