Transaction Monitoring Optimization Transaction Monitoring Optimization

Transparent and effective detection of financial crime is a reality. KPMG firms are disrupting and transforming the future of transaction monitoring.

The current challenge for banks is a combination of ever increasing and evolving regulations and legislation creating a complex regulatory environment in which to operate and comply with. Getting compliance wrong is not an option.

The tuning of Transaction Monitoring ("TM") systems is a complex task, intensified by heightened expectations from regulators. Financial Institutions often perform tuning on a periodic basis, relying on simplistic 'above and below the line' testing. Inefficient systems result in the production of large volumes of alerts. Worldwide banks manually review millions of financial crime monitoring alerts per month with almost 95% of the alerts raised being 'non-suspicious'. Current alert review processes frequently rely on extensive teams of personnel, often spread over disparate geographical locations. Many compliance professionals interviewed expected to see further increases in headcount this year to maintain their current levels of compliance.

Our solutions

In the face of increasing and evolving regulation, increasing transactional volumes, and the identification of new money laundering typologies, KPMG has developed solutions that target both systems optimization and alert review processes. The goal is an increase in quality of monitoring systems performance and faster, more accurate and consistent alert review, helping Financial Institutions to meet their TM obligations with increased quality and at reduced cost.

Automated tuning optimization

Ongoing review and refinement of TM systems can result in production of lower volumes of better quality alerts that can be reviewed more easily, whether by humans or use of the TM alert classifier.

• Automated tuning optimization relies on use of machine learning to run thousands of 'what if' scenarios to determine the optimal state for customer segmentation, scenarios, rules and thresholds.
• These recommendations can be reviewed by skilled human operators to determine whether adjustments to the TM system should be made in line with the clients own risk appetite.
• The automated tuning optimization tool can also be used to model business changes to determine impact on current monitoring scenarios. This insight will enable informed decisions on business change and resource requirements to manage alert spikes.

Transaction Monitoring alert classifier

The TM alert classifier automates alert review decision making at the first level of review.

• Using advanced machine learning techniques, the tool automates the identification of alerts that are likely to require investigation (auto-escalation) and alerts that are non-suspicious (auto-closure) allowing analysts to focus on high risk activity.
• Each alert review decision has a confidence level and is supported by a human readable decision rationale. Clients can therefore tailor deployment to dictate coverage and accuracy rates that support their risk appetite.
• The use of supervised machine learning ensures the tool is transparent and can be independently reviewed by auditors and Regulators.

Management information and analytics

Management information and analytics provides insight into system performance, data quality and additional insights that can inform financial crime controls.

• Analytics packages can be developed to provide insight into payment activity and indicate areas of risk within transactional behaviour.
• Visually rich customisable dashboards present management information in a way that highlights key insights and enables data-driven decisions making.

The benefits

KPMG has combined experience in Financial Services, Transaction Monitoring and technology with world class Data Science professionals to create advanced technology solutions for Transaction Monitoring systems. Experience of working with clients around the globe has helped ensure KPMG technology solutions develop at pace to reflect regulator appetite whilst addressing client challenges.

• Risk based - Client segmentation can be more accurate and aligned to the clients risk-based approach.
• Effectiveness and efficiency - AI tests thousands of different "what if" scenarios, providing recommendations for optimal tuning of customer segmentation, rules, scenarios and thresholds. The optimized system decreases the rates of 'non-suspicious' alerts.
• Cost reduction - Optimized TM systems are likely to produce fewer 'poor quality' alerts requiring time consuming manual review. Automated L1 review minimises the requirement for 4-eye human review, dependent on risk appetite.
• Regulatory compliance - Automated detection of 'Suspicious' alerts enables faster escalation of high risk cases to compliance. Documented, transparent and auditable systems testing and alert review processes demonstrate regulatory compliance.
• Insightful - Management information and analytics identifies links, patterns and behaviour to deliver real insight.

Let's do this.