Balancing preventing fraud while accepting good transactions – interview with Fraugster

When tackling fraud, there is always a high percentage of quality transactions that are falsely declined. How can merchants eliminate fraud without accidentally blocking good customers?

When it comes to approval rates, there is a limited amount of data points that merchants can use to base their decisions on. And if businesses don’t fully utilise the data, they could be rejecting legitimate transactions because their anti-fraud technology isn’t agile enough. Due to outdated technologies, merchants across industries are now seeing record losses because of fraudulent actions and the increase in false positives. It is more important than ever to ensure that your fraud prevention system understands the buyer’s true intent and can make considered decisions on each transaction attempted.

Fundamentally, this comes down to the information that needs to be collected. Data, such as IP addresses, email addresses, and payment methods, combine to paint a picture of your online transactions’ validity. However, more and more businesses find that these assets don’t always tell the whole story. Merchants require more adaptive approaches to differentiate between good and bad transactions. And this is where Artificial Intelligence can play a big part.

With Fraugster’s proprietary AI-technology, each transaction’s behavioural pattern is analysed in real time, leading to an accurate decision. This has the benefit of delivering a frictionless customer experience without letting bad transactions slip through.

How can Artificial Intelligence spot the difference between a potential fraudster and a legitimate customer in real time?

The AI performs at a rate comparable to thousands of excellent risk analysts. So, if a customer attempts a transaction, all of their data points are sent to the AI-Engine for inspection. Here, the AI enriches the data, investigating each piece of information and comparing it against others to determine whether it is suspicious. Does the username match the email? How many times has this credit card been used for a purchase on this site? Within what time frame?

On their own, individual data points can’t tell us the whole story. For example, one shipping address could be connected with two different credit cards. This doesn’t mean that it is fraudulent, it could simply mean that it’s a shared living space. But when the AI-Engine consolidates all of this information together, a better understanding of the story emerges. Like a detective building a case of evidence, the AI-Engine evaluates thousands of connections and attributes, looking for possible patterns, mismatches, or irregularities.

From here, the AI-Engine presents a Fraud Score, and a transparent and traceable determination is reached: ’Is this a good transaction or a bad one?’ or, ‘Is this a real customer or a fraudster?’ And because a team of real fraud analysts has taught the AI-Engine to mimic their thought process, the Engine can now perform this analysis on every single transaction in less than 15-milliseconds.

So, the merchants who work with an AI-based fraud prevention solution now have the accuracy of a human analyst and the speed and scalability of a computer, which they require to grow their business while protecting it from ongoing threats.

How can merchants maintain high approval rates in the PSD2 world?

While PSD2 attempts to provide a more secure experience for customers, the SCA (Strong Customer Authentication) requirements introduce a new set of challenges. While it does reduce the risk of fraud within the European Economic Area (EEA), it also creates a more complicated payment experience for the customer. Tests carried out by payments consultancy CMSPI suggest up to 30% of sales could be at risk because frustrated customers abandon baskets at checkout and because 3DS2 solutions are not working effectively. Customers want a safe, secure checkout experience but without additional friction or waiting times. But how can merchants present an acceptable balance of both, seemingly opposing, demands? It comes down to merchants understanding the regulation. And specifically, knowing what exemptions they can raise while remaining compliant. For instance, PSD2 states that exemptions can be raised for a variety of circumstances. These include recurring payments, low value, and low risk (often referred to as Transaction Risk Analysis). Coordination with PSPs and acquirers is the key to success here. Firstly, it is important to liaise with your PSP in advance to know their fraud level and understand their dependencies with their acquirer. Then analyse your current flow and forecast where you have the highest probability of raising exemptions. For instance, PSD2 allows exemptions up to EUR 100 if your fraud rate is under 0.13%. A potential best strategy for merchants could be: keep a low fraud rate to exempt low risk transactions.

The merchants who have this balance will see a dramatic improvement in their authorisation rates. And the best way to execute this strategy is a focused effort with an integrated fraud detection solution.

This editorial was published in the Fraud Prevention in Ecommerce Report 2020/2021, the go-to source in securing transactions while offering a frictionless customer journey.

About Max Laemmle

Max is the CEO and Founder of Fraugster, an AI-driven fraud prevention company. Prior to founding Fraugster, Max co-founded Better Payment, a German payment gateway service provider, and also acted as the product evangelist for the mPOS company SumUp. 

About Fraugster

Fraugster is a fraud prevention company, which helps online merchants maximise their revenues while reducing operational costs. Fraugster serves clients across industries: from online retail and marketplaces to gaming and travel. Fraugster operates globally and serves merchants both directly and through payment companies, minimising the integration effort.