In the post titled Time to Reimagine Watchlist Screening, we discussed how legacy technology and approach used in watchlist screening produce suboptimal outcomes, generate too many false alerts, and result in case backlogs and growing cost of compliance. Financial institutions (FIs) are now seeking to leverage new technologies to improve screening performance in three critical ways.

  • Efficiency gains: Improve resource utilization by applying intelligent automation, improving match rates, automating workflows, and optimizing infrastructure usage.
  • Effectiveness enhancements: Reduce voluminous false positives, hone in on true criminal actors and activities, and future-proof systems and processes against continuous regulatory changes.
  • Ease of implementation and maintenance: Simplify technical complexities, ease implementation challenges, streamline constant system maintenance and upgrades, unify fragmented architecture, and reduce dependency on legacy technology.

Modern technology-powered solutions can transform screening methodology, analytics, case management, and infrastructure usage to help achieve these goals.

Improving match rates with contextual screening

The screening process can be vastly improved by the application of intelligent automation in data processing and matching phases. Improving the quality of source data is critical for optimal screening performance. New technologies such as intelligent character recognition, robotic process automation, fuzzy matching, and natural language processing techniques can greatly help in data cleansing, normalizing, reconciliation, and unifying customer records and watchlists. Modern technology-powered solutions can intelligently handle data issues such as misspellings, transpositions, missing information, nicknames, ordering, and acronyms. Similarly, they can contextually assess aliases, nicknames, and initials; distinguish names of natural persons from those of organizations; and so on.

In transaction screening, different payment types such as NACHA, SWIFT, and FedWire contain different fields and various types of information such as name, location, transaction date, amount, as well as free text information. Contextual analysis of the fields is essential so that the screening engine knows which fields to screen. Solutions that screen all or irrelevant fields inevitably generate too many alerts, which is a root cause of the false-positive problem. Therefore, the screening engine must first assess the message types and identify relevant fields and pertinent information for screening; this itself can significantly improve match rates.

Modern solutions offer the ability to fine-tune the matching process according to specific entities or subfields, such as matching nicknames for individuals but not organizations, neutral words (e.g., Ltd or LLC) for organizations, and so on. They also allow setting matching rules and tolerance levels for each field of comparison. Culture-specific language processing is another key feature of modern solutions that can significantly boost screening performance by assessing cultural factors such as naming conventions, ordering, common names, phonetic properties, and so on.

Artificial Intelligence (AI) and natural language processing (NLP) techniques can further enhance performance by automating text analysis, including native language analysis, and providing confidence scores for accuracy of the match. They can help in entity resolution by analyzing multiple data sources and identifying relations among seemingly disparate data points. AI and graph analytics such as link and network analysis can automate the case investigation process and improve its efficiency manifold.

Adopting cloud for infrastructure and cost optimization

The cloud model has the potential to transform watchlist screening in several ways. Financial institutions’ attitude toward cloud has radically changed in the last three years, with incumbent institutions gradually migrating workloads to the cloud while digital banks and fintech startups take a cloud-first or even cloud-native approach to technology implementation.

Cloud’s elastic scalability can be helpful in screening because of the need for high computing power in short bursts. Cloud enables FIs to easily scale up or down in a cost-effective way and minimize unused idle capacity. Cloud-based solutions can significantly lower the total cost of ownership (TCO) as they eliminate FIs’ need to invest in and maintain complex technology stacks. Software as a Service (SaaS) solutions facilitate rapid time to market, enabling FIs to set up and run new systems in a matter of days or weeks compared to on-premise deployments that can take months. The SaaS model also allows FIs to quickly adopt the latest best practices in the industry, especially helping mid-sized and smaller institutions leapfrog the technology advancement curve that legacy institutions have traversed over years.

Cloud provides a conducive environment for easier interaction among the players in the financial services ecosystem—such as integrating a patchwork of internal and vended solutions, consuming watchlist data from external providers, collaborating with dev-ops teams spread across different functions and locations, and so on. With growing adoption of the open banking model, cloud’s common framework and interoperability can foster innovation and accelerate technology advancements. Ecosystem interaction can be further enhanced by the use of application programming interfaces (APIs) where independent applications—such as in KYC, transaction processing, risk scoring, and case management—can automatically communicate with each other and be tied more tightly with the screening process.

Equally important, by outsourcing the management of software and technical environment to expert providers, the cloud model allows FIs to stay up to date with best practices without having to spend much resources on their maintenance.

Watchlist screening is a prime use case for cloud adoption because in addition to optimizing cost and infrastructure, cloud offers the benefits of multi-tenancy where a group of user FIs can mutualize common resources. Across all types and sizes of FIs, there is a strong preference for cloud to implement new solutions or replace solutions that are coming to their end of life. Refreshment of screening technology across the industry therefore should see greater cloud adoption in the near future.

Optimizing performance with distributed framework

Screening is essentially comparing two datasets, each containing large volumes of information. Analysis of very large datasets can be greatly improved by using a distributed framework instead of the brute force approach of comparing every item on one list with every item on the other. In the distributed approach, search functions such as crawling, data mining, indexing, and query processing are distributed among several computers in a decentralized manner instead of being performed by a central supercomputer.

The distributed elastic search framework offers high scalability for in-memory processing on a network of distributed nodes, thereby vastly improving the speed of searches. Customer names and other information are first tokenized and indexed, such as splitting a full name into first name, middle name, and family name. Subsets of the data are then sent to a cluster of computers with similar characteristics. The records are then compared, and results returned along with indexes to reconstruct the final match scores. In this way, comparisons are only done for a subset of cases that are chosen intelligently and run in parallel, which boosts computing performance manifold.

From remediation to disruption

Beyond the improvements in efficiency and effectiveness, modern technology-powered solutions can bring about a paradigm change in screening. For example, in response to the increasing demand for digital financial services, modern solutions powered by intelligent automation and API technologies and supported by low-latency infrastructure can carry out near real-time screening and decisioning.

Another paradigm change could be brought about by converging transaction screening with fraud analysis. The industry has long talked about the need for unifying fraud and AML, but progress has been limited. A key challenge is that fraud must be resolved quickly, whereas AML transaction monitoring analysis happens post transaction over days and months. This has resulted in disparate processes that are siloed and difficult to unify. Transaction screening and fraud detection are similar in that decisioning—whether to pass or block a transaction—needs to be prompt in both cases. Furthermore, fraud analysis increasingly involves analyzing the intent of transactions rather than just ascertaining whether the activity is an account takeover. Therefore, information and analytics gleaned from transactions—such as beneficiary vectors, device ID, IP, transaction amount, and derived values such as risk scores—can be used in both fraud and screening. Converging the two will streamline processes and strengthen compliance.

Thriving in a hyper-digitalized post-pandemic world will necessitate reimagining and disrupting screening operations. FIs must invest in next-generation tools and technology to improve screening performance and regulatory compliance. The transformation journey will look different for different FIs. Large FIs, because of the deadweight of entrenched legacy systems, usually begin experimentation with new technology as a complementary or challenger solution to their existing screening engines. Their key objective is to minimize risks of transformation, so they start at the edges and gradually replace the core with modern technology. Smaller and more agile FIs, and fintechs, unencumbered by complex legacy architecture, can take a more direct approach to core transformation by directly adopting new solutions.

Subscribe

For further insight into the payments and banking industries, subscribe now and stay up-to-date on the latest tips, trends, and topics. You can also check out The Payments Podcast, where experts engage each other on the real-world factors impacting your industry.

Posted by Arin Ray

Arin Ray is a Senior Analyst with Celent based in the firm's New York office with over a decade’s experience in tracking business, regulatory, and technology trends in financial services. He has published extensively on technology trends in financial crime compliance operations with specific focus on the role of next generation technology in combatting financial crime. His advisory engagements have included developing technology strategy, market entry and expansion strategy, and benchmarking studies for financial institutions and technology providers.