Enhanced Transaction Surveillance Capability for a leading Financial Services Company
Transaction surveillance or transaction monitoring has become a pivotal aspect in anti-money laundering procedures. With the rapid increase in number of transactions, banks and major financial institutes are moving from manual to automated transaction surveillance systems. This automated surveillance system is important to oversee suspicious transaction activities at a much larger scale than manual surveillance system.
A leading financial services provider and one of the largest credit card brands in the US partnered with Mu Sigma to enhance its surveillance system efficiency and develop a workbench to automate its transaction monitoring process and highlight fraudulent entries.
The client’s anti-money laundering segment (AML) team is responsible for managing and mitigating the risk embedded in banking activities. These activities centre around Transaction Monitoring.
The client’s existing transaction monitoring and alert generation process was inefficient that resulted in the generation of thousands of alerts monthly.
However, reviewing these thousands of alerts itself was a difficult and gigantic task. Moreover, analysts spent most of their time (40%) gathering information to investigate an alert. This was due to a disconnect between the alert generation engine and data sources that resulted in the entire process being time-consuming, heuristic, and susceptible to manual errors.
The team approached us with an objective to automate their transaction monitoring process and develop an accurate alert mechanism to flag off fraudulent activities.
The Mu Sigma Approach
We began by interacting with various stakeholders from the client’s end to get a deep understanding of their analysts’ alert review process and leveraged our problem-solving frameworks to map the complexity of the problem space, thus developing a comprehensive roadmap for problem-solving.
This allowed us to uncover different data sources that helped with better transaction monitoring that the bank had not considered using previously. This also led to redefining business rules and making them more robust.
We followed the below methodology to enable automation and software-based pre-processing, reducing manual effort:
• We focused on understanding the actual review process, and listed down all the different elements involved in the context of a transaction setting up SLAs against the industry benchmark
• Key transaction elements were identified and reviewed, such as internal EDW information, location information, and transaction channels
• Data that was previously perceived and used in siloed form from different data sources such as banks, third party was replaced and integrated into one data source by mapping problem complexity
• Rules to identify a transaction as fraudulent were set based on the learning from the analyst review process and discussion with other key stakeholders
• Historical data was used to test the efficiency of different models to correctly identify fraudulent activities and drive the most relevant alert mechanism. A rich KYC and other reports were rolled out to share with business stakeholders. The alert mechanism and report generation module together formed AML surveillance workbench
The AML surveillance workbench was used as the crucial single-screen review framework for the entire transaction monitoring process. The visualization of this workbench captured every aspect of the analyst investigation process.
A tool providing detailed context on each transaction was built. Users had the flexibility to obtain information across transactions, historical flows etc. The tool was hosted on a rich featured workbench designed for scalability and high usability.
• The solution reduced the investigation time by 40-50%.
• The productivity of an analyst boosted from 3-4 reviews/day to 6 reviews/day.