The significant uptick in fraud naturally correlates to much costlier ways to mitigate it. To put things in perspective, every $1 lost to fraud now costs United States financial services firms $4.23, a 16.2% increase since 2020. In comparing today's environment with pre-pandemic numbers, the cost of fraud for U.S. financial services and lending firms has increased by as much as 9.9%, despite 71% of respondents increasing their spending on fraud prevention year-over-year. Other areas of fraud to consider include hidden costs, legal repercussions, regulatory fines, reputational damage and loss of clients.

Overcoming Obstacles to Counter the Threat

Perhaps not surprisingly, respondents reported that the main barrier to defense is insufficient, ineffective or lack of automation. Forty-six percent of respondents cited a greater need for automation as the most common barrier to being prepared to combat fraud, followed by an absence of dedicated teams for fraud prevention (41%) and the inability to adapt to new threats (39%).

So, how can credit unions bear the cost of more effectively mitigating fraud without sacrificing the quality of the member experience? Check fraud, which has surged 84%, can be effectively detected by leveraging the latest iterations of machine learning – specifically, probabilistic computation processes that use extensive mathematical and machine learning methods to analyze both images and transactional data to identify potential suspicious items and corresponding risk exposure. Machine learning permits risk scoring of multiple attributes to be escalated and displayed on a centralized interface for suspect adjudication workflow and case management for those items requiring additional research.

Traditionally, detecting more fraud required credit unions to explore a more comprehensive array of solutions, which often led to an increased load on the analysts and a net loss to the credit union. Today's technology permits credit unions to improve accuracy by combining automated decisioning with predictive analytics, allowing more "true suspect" items to be analyzed, fewer false positives returned and a decreased analyst alert queue load.

The combination of transaction and image analysis functionality drives detection rates even higher, nearing 95%. At the same time, digital capabilities allow credit unions to reduce false positives from the typical 1-2% of transaction volume down to 0.6%. Scoring and decisioning inputs into machine learning algorithms also significantly improves detection accuracy, reducing labor costs tied to reviewing suspect items. More importantly, this pairing of increased automated functionality for detection with decreased human intervention enables credit unions to establish a more robust fraud posture without significantly increasing operational costs.

Fraudsters will always be cultivating new techniques that will put credit unions and their members at risk. Leveraging machine learning technology helps credit unions level the playing field. Those prioritizing the modernization of fraud risk mitigation through automation will be better positioned to protect the member experience – as well as their bottom line.

Todd Robertson

Todd Robertson is SVP of Business Development for ARGO, a Richardson, Texas-based technology and fraud solutions company serving the financial services and health care industries.