Building a Holistic View of Financial Crime and Fraud Risk

In the United States, enforcement agencies such as the Financial Crimes Enforcement Network (FinCEN) are encouraging banks to adopt more innovative approaches to addressing financial crime, including the use of AI, to meet banks’ obligations under the Bank Secrecy Act.  

The Hong Kong Monetary Authority has actively shared case studies of how technologies such as network analytics have been applied in AML and fraud cases. In 2022, the region’s banks using network analytics solutions were able to generate 319% more suspicious transaction reports compared with 2021; in addition, the banks saw a 113% increase in criminal proceeds that were restrained or confiscated. 

And enforcement agencies themselves are even starting to use AI and data analytics tools to successfully drive enforcements, as did France’s Autorite de controle prudentiel et de resolution (ACPR), which fined a bank division €1.5 million for poor due diligence and transaction-monitoring processes. ACPR used an AI solution to analyze hundreds of millions of the bank’s transactions and other data, finding that the bank failed to investigate common red flags for financial crime. 

Banks are implementing Enterprise Risk Management (ERM) models more broadly so that all risks – financial and non-financial – are managed under a Chief Risk and Compliance Officer. This is yet another step toward the single view of customers and counterparties that can allow analysts to create fuller and richer portraits of transactions and their impact.  

Collaboration to fight financial crime and fraud becomes much more possible when banks take an integrated approach: 

Between functions

Where banks drive collaboration between the risk functions in their organization, with some going as far as creating hybrid teams or financial crime centers of excellence.

---

Between banks

Via alliances such as the FrauDfense in Spain, where three banks have committed to exchange relevant and useful information to fight fraud.

---

Between banks and regulatory agencies

Such as the UK’s Joint Money Laundering Intelligence Task Force (JMLIT), formed of over 40 financial institutions, along with the UK’s financial regulatory authority and five law enforcement agencies.

Today, technologies such as entity resolution are giving banks a single view of customers far more effectively than traditional name matching. Advancements and innovation in AI, data management, and graph analytics have improved the efficiency and effectiveness of financial crime programs in many ways. Click below to discover how:

Analysts and investigators must change how they think and talk about financial crime. The old way was to pose a query such as, “Is John Smith a money launderer?” The new way is to ask, “Do I really understand my customers and counterparties? Have I got a clear and complete view to make good decisions about what I see?” 

The new way to fight financial crime does not start with an outcome, such as searching for money laundering or fraud. It starts with the customer, their network, and their associated counterparties. Banks have realized that counterparties expose them to even more risks, including non-customer counterparties, requiring moving beyond the narrow view of customers only.  

Here’s how to get one: 

1. Gain a single contextual view of customers and counterparties
   
   Knowing the customer is critical, but so is knowing which counterparties customers are interacting with. For example, criminals involved in authorized push payment (APP) fraud, or recruiting money mules, may be coercing customers into criminal activity, which wouldn’t be visible without a wider perspective on who is engaging with your customers. 
   
   With a single data foundation, banks gain a complete and single view of customers and counterparty activity. The process allows banks to build a full network of relationships from different perspectives, removing noise and revealing patterns more easily. 

2. Reuse data foundation for many needs
   
   Once data is connected, then money laundering, fraud, and KYC become lenses applied to the same data foundation to detect, monitor, and investigate financial crime and fraud more effectively. Your data foundation becomes a trusted, renewable, and reusable resource that can be used across the bank to expand insights.  
   
   This approach helps identify previously hidden connections between entities and greater risk. For example: you can more clearly see the interplay between a predicate crime (e.g., fraud) and attempts to launder such illicit funds.  

3. Use a dynamic approach to detection in place of rules-based models.
   
   Using predictive detection models with a broader range of inputs and indicators helps to identify risk more accurately and efficiently than use of individual rules in isolation – as is typical in traditional transaction monitoring approaches. Customers can be clustered in a dynamic way that evolves as they transact over time. Investigators can identify where and when a customer is acting abnormally in a particular cluster compared with their peers, or compared with how they’ve behaved previously. 

The value of contextual views is that they allow the detection, monitoring, and investigation of risks that otherwise would not have been found – or even been sought – by analysts. For example, by sharing known lists of bad actors or up-to-date KYC profiles between one line of business and another, risk teams get new information that can reveal new issues. 

Results come from detecting and investigating real risks, as opposed to endlessly responding to alerts that aren’t material to financial crimes. By proactively addressing real risk, teams can respond more quickly and mitigate exposure, since they’re no longer stuck in a cycle of reactive activity. Teams can also augment decision making with the context and insight needed to make informed decisions. 

You'll drive greater efficiency 

Teams can reduce the time spent on clearing high false-positive alert rates and submitting overly defensive SARs or STRs. They can also reduce manual information-gathering once data is automatically curated into a single place. With the time they save, analysts and investigators can shift time toward responding to true risk. The process creates a force multiplier effect, where the value of the financial crime and fraud program increases to more than the sum of its parts. 

You'll improve the customer experience 

Many of the same data investments that improve understanding of the customer for risk management can be deployed to aid revenue-generating functions of the bank. This allows financial crime risk management to be a value-add to the bottom line, rather than merely a cost. Similarly, more effective identification of fraud risk means less customer friction from fraud controls, and less risk of loss to the bank or customer.

Know Your Customer (KYC) 

As different risks converge, successful detection of financial crime comes down to knowing your customer. The complete view of a customer is fundamental to this process: While one customer account may not appear risky, fraudulent activity could occur in an account owned by the same customer in another part of the bank.  

The single view of customers elevates the role of KYC from a box-checking exercise to a function that powers an understanding of customers. This is in line with the wider move to perpetual KYC (pKYC), which is being adopted by innovative banks to realize a data-driven and automation-first approach to understanding customers.   

Anti-Money Laundering (AML) 

AML functions have traditionally relied on static rules-based transaction monitoring, which triggers alerts based on limited information and single events. Such static transaction monitoring leads to false positives and poor risk identification.  

A contextual monitoring approach transforms the role played by AML functions. Contextual monitoring focuses on holistic relationships, rather than transaction values and volumes. By gaining the full context about a transaction – including the customer, counterparties, and related parties – AML investigators can narrow down behavior and interactions that are indicative of true risk. 

Fraud 

Fraudsters rarely stop at a single fraud; organized criminals often work in groups synthesizing IDs, colluding, and leveraging multiple businesses across similar operations. Contextual monitoring offers powerful analytic capabilities to alert against complex fraud typologies in real time. Banks are empowered to make trusted decisions in just seconds.  

In addition, entity resolution helps create dynamic networks in data, using links such as ownership structures, transactions with counterparties, and common KYC data. These networks and single views can be added to risk analytical models, monitoring, or reporting. 

Innovative banks have employed a decision intelligence approach to help them connect data, see the full context, and get a holistic view of financial crime. According to Gartner, decision intelligence improves decision-making by understanding and engineering how decisions are made, and how outcomes are evaluated, managed, and improved by feedback. 

Decision intelligence leverages innovative data management practices, advanced analytical techniques, and AI to turn large volumes of disparate data into actionable insights that drive critical business decisions.   

In a decision intelligence approach, data is ingested, connected, and contextualized using entity resolution and contextual analytics – connecting an entire organization’s previously siloed and scattered data sources to create one trusted and reusable resource. 

How Quantexa’s Decision Intelligence Platform enables a holistic approach to financial crime and fraud 

1. Unifying data  
   The first step to a unified financial crime and fraud approach is to build a strong data foundation. Quantexa’s intelligent data cleansing and enrichment processes help make sense of incomplete structured and unstructured data sources. This approach enables banks to aggregate and process internal and external data such as adverse media, watchlists, and transaction data without complex coding or specialized skills.  
   
   Quantexa’s category-leading Entity Resolution is powered by a transparent and tuneable predictive model to address data availability and quality issues without pre-resolved entity lists for training. Quantexa’s model creates a single view of entities dynamically across internal and external data sources, with an industry leading 99% accuracy match rate.   

2. Creating context  
   Quantexa provides pre-built, domain-specific proprietary models across fraud, AML, KYC, and other types of financial crime. These models leverage emerging techniques such as Contextual Monitoring that transform the view of risk for a clearer understanding of customers, counterparties, and relationships. 
   
   The Quantexa composite scoring and alerting models, built on a strong data foundation, are applied to processes such as Contextual Monitoring, leveraging graph analytics, contextual insight, and domain expertise. This approach minimizes false positives, provides explainable decisions, and increases confidence in critical decisions. With scoring models and methodologies, organizations can quickly deploy trusted solutions without technical expertise in-house for baseline scoring and alerting.  

3. Deciding and acting
   By enabling a human/AI partnership, Quantexa enhances customers’ ability to make critical business decisions, reduce the burden of repetitive tasks, and enable teams to be more productive:  

• Empower data-driven analysts with the data, context, and insights needed to make informed decisions  

• Reduce time spent cleansing, prepping, and analyzing data with data visualizations that support decisions  

• Explore large volumes of data and easily identify potentially significant information 

 To ensure customers can successfully understand how models perform, Quantexa prioritizes transparency and comprehension of AI and machine learning models. Network visualizations are user-friendly for technical and non-technical teams. Users can easily make sense of extensive data volumes. 

By investing in Quantexa’s Decision Intelligence Platform, you can enable the following capabilities across your financial crime and fraud programs:  

Contextual Monitoring

• Go beyond transaction monitoring and begin to continuously monitor your customer data, focusing on holistic relationships rather than transaction values and volumes.

• Harness the power of networks while maintaining the transactional risk analysis to transform the overall view of risk to a clearer understanding of customers, counterparties and their relationships in real time.

• Uncover hidden risk quickly and more accurately, accelerating efficiencies and effectiveness.

Detection

• Shift from isolated risk indicators to a typology-based scorecard approach with entity resolution, network generation, and anomaly detection

• Effectively identify risk with fewer false positives and efficient triage and prioritization

• Aggregate and map indicators against financial crime typologies

• Build multiple layers of typologies into a single structured scorecard

• Deploy Composite AI, which uses a combination of subject matter expertise and domain knowledge and machine learning, natural language processing, and deep learning techniques to detect new typologies

Intelligence-led investigations

• Transcend traditional investigative approaches that profile only a narrow scope of activity when searching for “known known” risks

• Opt for an intelligence-led investigation approach that automatically provides a single resolved view of each entity across internal and external data for directly related and associated entities

• Eliminate screen hopping and data gathering by overlaying a scoring framework that focuses on particular types of risk that need a higher profile

Banks don’t need to rip and replace systems. Think more along the lines of incremental change. Banks are finding success by starting in one area (e.g., risk function or business line) and building out, steadily gaining more value as they go. Click on each to learn more.

Set strategy and vision

As with any financial crime transformation effort, it’s vital to define what you’re trying to achieve, the expertise needed to drive success, and potential impact on any existing functions. Banks must set common goals across teams to work effectively together – and acknowledge that teams may have different goals. For example, AML and KYC teams are focused on compliance, whereas fraud teams have financial goals to meet.

---

Build out your data foundation

Create a consolidated view of your customers and counterparties by stitching together internal and external data to build a strong data foundation.  Invest in a technology partner that has a proven record of building solid data foundations that support an integrated approach to financial crime. Work closely with the partner to successfully deploy data solutions. 

---

Prove value in one area, and expand incrementally

Start by focusing on one geography, line of business, or risk area to prove value. Use that success as an accelerator to expand to other areas, generating more value and expanding your view of customer activity (and risk) as you go. 

Meeting resistance? Or think building up a holistic view of financial crime is too difficult? We’ve guided banks through this kind of continuous improvement effort, so we understand the challenges involved and how to overcome them.

You don’t need to embark on a massive transformation project to begin improving data quality before you start connecting it together. Technology has evolved to the point that data can be connected and used downstream without normalizing, cleaning, or placing it in a single data lake. Even poor-quality data is usable. This means you can see value much earlier and with less effort than in the past.  

While larger banks have led the way in terms of financial crime innovation, banks don’t need to be large global institutions to gain value from this approach. In fact, it can be easier for smaller institutions to carry out this data transformation because their operations are simpler (i.e., fewer systems and teams). In addition, smaller banks’ compliance budgets can be larger as a proportion of total budget, so they may find themselves under greater pressure to innovate and see results. 

Leveraging a Decision Intelligence platform can also reduce costs. It’s expensive for banks to build their own cross-program platform, whereas leveraging a DI platform can accelerate the journey. If the platform includes industry leading analytics and machine learning models out of the box too, then the bank doesn’t need a team of data scientists to build them from scratch. 

While some institutions are putting hybrid teams in place – such as FRAML teams – this operating model isn’t required to benefit from the more holistic view of risk. When specialized teams all have the same holistic view across connected data, they all make better decisions, even when working separately. 

To detect and act on potential criminal activity with AI, a bank needs to be able to consistently explain to regulators, auditors, and other stakeholders how decisions are made. And when there’s confidence within the bank about decision-making, adoption of technology is more rapid, scalable, and drives better regulatory compliance.  

 Look to adopt a “white box” approach to AI – meaning one that promotes transparency – so you can easily understand decisions and account for them to meet governance and compliance requirements. By adopting solutions with intuitive user interfaces, technical and non-technical analysts understand the outputs of AI-powered decisions to tune and improve performance. 

Banks should measure outcomes early and at key milestones, allowing more robust business cases for expansion.