Combating Payments Fraud in a Big Data World
In today’s hyper-connected world, the fight against payments fraud, money laundering and other cyber threats is challenging for banks and other financial institutions around the world.
While new social and cloud technologies empower businesses and consumers, they also have the potential to be used as smokescreens for criminals. According to the Association for Financial Professionals’ 2013 Payments Fraud and Control Survey, 61% of organizations have experienced attempted or actual payments fraud. In addition, more than a quarter of respondents saw those threats increase last year.
This is happening at a time when our digital society creates and shares more data every two days than existed from the dawn of modern civilization until 2003. And by 2020, IDC estimates there will be 450 billion online business transactions each day. As the volume, variety and velocity of those transactions increase, so do the complexities associated with identifying and preventing payments fraud.
To combat fraudulent activity, organizations are turning to big data and predictive analytics technologies that can distill massive volumes of diverse and seemingly unrelated information into actionable insight. Just as criminals seek to cover their tracks in this ocean of data, C-Suite and line-of-business leaders are recognizing that predictive analytics can help prevent, identify and eliminate fraud across the organization.
Turning the Tables
To prevent fraud, organizations must re-think and re-design how they store, manage and interconnect data across multiple siloes. With today’s evolving threats, time is of the essence. If data must be extracted from numerous, unconnected warehouses and then manually compiled and analyzed, then the battle is already lost. There is a high probability that key connections will be missed as a result of human error, or result in inadequate entity resolution due to the high volume and disparate types of structured and unstructured data living in the system.
Data is only valuable if the right people have access to it at the right time – both to drive the core business and to identify and eliminate fraud. Many organizations have found the most successful way to detect and prevent fraud is by creating a central data analysis environment that applies advanced statistical, entity resolution and link analysis that can spot trends, patterns and anomalies that could be potential fraud indicators.
This “central nervous system” of data allows banks and other businesses to make connections and identify threats they would otherwise miss. Instead of a “one-size-fits-all” warehouse where all information must be migrated to a single location, organizations should seek out scalable and secure solutions that build on the existing infrastructure to interconnect and integrate relevant data sources into a single repository for analysis. This is made possible by the cloud and the new extreme data transfer capabilities from companies that can reduce the time it takes to move a 24 gigabyte around the world from hours to seconds.
The right tools enable business leaders to pull and analyze the information they need, when they need it, without applying too many resources on consolidating everything into a single database. This can be achieved through a combination of cloud and on-premise solutions that securely aggregate and analyze encrypted customer payment data from all online, mobile and physical transactions. This enables banks and other financial institutions to more quickly uncover fraudulent activity and ensure legitimate transactions are processed without delay.
Once the data is aggregated into a centralized analytics environment, the next step is to perform statistical risk analysis. The process defines who is talking to whom and identifies the key entities in a transaction, and analyzes unusual or anomalous activity. As fraudulent patterns are detected, a red flag is raised to the business to differentiate explicit threats from run of the mill payment errors.
This critical concept is called “minimizing the zone of ignorance” or reducing the gulf between the volume of internal and external data available to an organization and what it can actually process and analyze. The combination of big data platforms and predictive analytics help organizations close that gap by understanding the nuances and interrelationships within structured and unstructured data.
Visualizing the Threat
When employing a big data and analytics approach, it is also important to keep next-generation technology in mind since it can help maximize an organization’s fight against fraud. Visualization is such a technology being applied to a range of complex fraud challenges to rapidly collapse them into a manageable scope, identify and prioritize threats, develop critical intelligence and make quick informed decisions.
Differentiating between fraudulent and legitimate activities is difficult when vast amounts of information are buried within rows of numbers. Visual representation can illustrate the story behind the data, and in cases of potential fraud, can demonstrate linkages that are not obvious between people, places and things, such as individuals, fraudulent information, financial records, suspicious locations, phone calls and any number of items.
For example, a financial institution that manages millions of card transactions daily must be able to understand the legacy threats that are derived from billions of terabytes of historical data and rapidly compare them to evolving threats. Visualization of this activity is critical for financial institutions to quickly plug vulnerabilities and prevent potential instances of payment fraud. It would be impossible to react to such scenarios if analysts had to manually delve into a statistical report that could comprise reams of paper. In the aftermath of any fraudulent transactions, visualization also allows organizations to rapidly communicate the course of events to law enforcement authorities so they can take action.
Banks can take steps to recognize whether any anomalies result from user error, such as accidentally running a credit card transaction more than once, or from intentional misuse. By applying advanced analytics, organizations can quickly identify fraudulent activities while continuing to process legitimate transactions and keep business moving forward. In addition, this type of analytic modeling can help to determine, when fraud is identified, whether this type of fraud is the work of an individual or criminal organization.
Today, solutions exist to help organizations tackle the relevant data from the mountains of existing data sources into a single environment where analytics can be applied easily and effectively to share intelligence and prevent payments fraud.