Fraudsters are getting more and more organized and have started more sophisticated and targeted attacks on financial systems over the last couple of years. With the overall payment ecosystem getting more complex, fraud has become increasingly difficult to detect and prevent. On the other hand, computing power is cheaper than ever and technologies such as machine learning have made great strides over the last couple of years. In a world where customers, who are much more informed than before, communicate openly with their peers, fraud and the associated reputational damage has become an increasingly sensitive subject.

Machine learning and AI techniques are being adopted in the market for the simple reason that machines are much better at repetitive tasks and managing large pools of data compared to humans. When writing rules to capture fraud, a person would most likely be looking at traditional payment attributes such as country codes or merchant category codes to asses where fraud is most likely going to happen. But in a continuously evolving world, the fraud may already have moved elsewhere which indicates that more sophisticated pattern recognition is required in order to be successful.
Machines are able to detect and recognize thousands of features on a user’s purchasing journey instead of the few that can be captured by creating rules. Contrary to humans, machines are able to see deep into the data and make very concrete decisions based on large volumes of transactional data. Implementing machine learning generates an obvious gain on three axis:

  • Speed: only machine learning techniques enable us to achieve decisions in microseconds with the level of confidence needed to approve or decline a transaction.
  • Scale: rules-based programming and machine learning approaches have an inverse relationship with the size of data sets. Rules become less effective while machine learning approaches get better with larger datasets. So simply put, looking at the cost of maintaining a fraud detection system as a merchant’s customer base gets larger, machine learning is the only logical choice.
  • Efficiency: Machines love repetitive tasks, human hate them. Algorithms are there to do the heavy lifting and only escalate decisions to people when their input adds insight.

Putting payments processors in control

It is key for a payments processor to be in control of their fraud prevention solution. When purely relying on a vendor, there is a dependency on the implementation capability of this vendor to roll out new rules fast enough to still be relevant in today’s fraud landscape. Payment processors should be looking at self-service tools for cheaper and faster to roll out , allowing them to stay in tune with reality. There is no one size fits all for when it comes to fraud prevention for payments, so a payments process needs to have the power to adjust quickly and remain in the driver seat.
However, these things should not be only be about who controls the buttons. At BPC, we offer tools for automated rules generation that have proven a tremendous benefit when it comes to identifying where fraud is coming from and proactively adjusting the rules that are applied in detecting potential new fraud cases.
Customers want to be proactive and empowered – Less artificial, more intelligent
It is not just about putting the right technology in place to better detect and prevent fraud, there is also an interest to give customers tools to manage their payment instruments based on their risk appetite. By equipping consumers with the tools to define when, where and how their cards can be used, customers are empowered to take control and banks benefit from lower fraud costs and false positive rates while at the same time increasing the card usage and customer retention rates.
Allowing consumers to control their card usage is not simply about building customer confidence, reducing fraud or increasing revenue by driving higher card use. It is a tool to empower customers and business owners to take charge of their financials and create a much deeper relation between the bank and the customer. Technologies such as Artificial Intelligence and Machine Learning are merely tools to increase speed and efficiency and reducing fraud rates but combining these powerful new technologies with a human touch is what is really creating most value for processors and customers alike.