Gaining efficiency in fraud detection

Even the best preventive measures can’t stop all fraud. And banks and other organizations need to be ready to expect the unexpected as fraud techniques evolve as new opportunities arise. An effective fraud detection system is an essential piece of any anti-fraud system. Efficient fraud detection helps prevent losses, creates better customer experience, and helps an organization build a better business.

Evolving fraud threats

There are almost countless known fraud schemes. Any business could fall victim to business email compromise or cyber security related fraud. Banks and financial institutions must be on the lookout for credit card fraud schemes, synthetic identity fraud, account takeovers, and more. Insurance companies must be vigilant for fraudulent claims.  

In addition to these many known fraud methods, fraud patterns and techniques are constantly evolving. Clever fraudsters find new opportunities all the time, meaning financial institutions and other organizations need an efficient and adaptable system to flag all kinds of suspicious activity. 

Covid-19 is an example of how quickly fraud threats can evolve. The onset of the global pandemic forced some major societal changes. Huge numbers of people began working from home, away from in-office security systems. And with more people staying home, more business and transactions went digital. Both of these rapid changes created ample opportunities for fraudsters willing to manipulate a crisis for their own benefit. 

At the same time, governments increased aid to citizens and businesses to prevent dire economic outcomes. Fraudsters were also quick to take advantage of such easy money. 

None of these opportunities for fraudsters could have been predicted before the pandemic hit. Prevention only goes so far in an unprecedented situation—or in day to day operations. It highlights the importance of having a reliable and solid fraud detection system to catch criminals before they can do too much damage.

Rules-based detection solutions

Since a lot of fraud involves repeated patterns, analyzing data to find those patterns is a focus for analysts. Setting up rules to automatically flag suspicious behavior based on known patterns saves time on detecting common types of fraud, since these systems can process huge amounts of data in a short amount of time.  

The downside to rules-based detection solutions is that they generate a high number of false positives. Rules-based detection solutions are also often poor at detecting sophisticated fraud techniques, complex networks of fraudsters, or previously unknown fraud patterns.

Understanding context for sophisticated fraud detection

Understanding the context around your data is essential for uncovering many fraud schemes. And context is exactly what’s missing from the analysis of many traditional rules-based solutions. They are good at connecting individual data points: a business to an address, a customer to a bank account, etc. But rules-based detection systems are unable to analyze the deeper connections and low-level signals to detect more complex fraudulent behavior or networks of fraudsters.

Graph analytics and fraud detection

illustration showing fraud detection using graph analyticsGraph analytics is a powerful tool for fraud detection. It solves for many of the limitations of rules-based detection systems. Structured as nodes (data points) and edges (the relationships between data points), a graph model enables you to explore not only your data, but also the connections within. This data model is particularly well suited to organize and analyze data where connections are as important as individual data points. 

With an investigation platform like Linkurious, based in graph analytics, you can detect complex fraud patterns that would have been difficult or impossible to uncover otherwise. 

Take the example of credit card skimming fraud. A credit card skimming device copies a credit card when it’s slid into a card reader at an ATM, cas pump, or other point of sale. The device owner can then use this information to make fraudulent transactions, such as online purchases. Once a bank receives reports of fraudulent transactions, a graph analytics solution can show if there are common points of sale among the various transactions. These connections can reveal when and where the credit cards were originally stolen. The skimming scheme can then be stopped.