Welcome back to Matrix AML Academy. In case you missed our previous post, this is the 2nd part of an 8-part educational program on how to improve tour AML program and systems

Last week, in Chapter 1, we focused on Transaction Monitoring Implementation Best Practices. However, no matter how well you implement your systems, if the data that is being fed into them is incomplete or inaccurate you are likely to receive erroneous results, unnecessarily straining your Investigators. In this chapter and the two chapters to come, we will dive deeper into a subject that is painful for most financial institutions – Data Quality, and how to improve it.

One of the key challenges financial institutions face today is in depicting a complete view of customer relationships to better assess customer risk while monitoring customer transactions. Without a holistic 360-degree customer view, AML and fraud alert processing takes into account only limited account-based information of the customer. For example, account-based monitoring may not flag suspicious transactions amounting to $5,000/month in a certain customer account, yet customers may have (or control) 5, 10, or 50 other accounts that the surveillance models may not be linking together.  

Data quality plays a key role in Entity Resolution and Account Linking. Good quality data is critical for high-confidence matching, as your firm connects the dots from data across the enterprise. Advancements in artificial intelligence and automation offer promising potential in uncovering relationships and expanding the traditional scope of data sources available to the compliance officer. Nevertheless, foundational data quality controls, such as validation, normalization, and standardization processes, are necessary to fully utilize these advanced capabilities.

The Effects of Poor Quality Data

Accurate and complete data drives our ability to function as an effective line of defense against Financial Crime. We rely on data to initiate effective prevention, whether in the AML or Fraud arenas. Poor data affects our ability to have a real sense of the customer’s interactions with the institution. Without good quality data:

• A 360-degree view of a customer cannot be fully achieved
• AML systems generate false positive alerts
• Duplication of investigative efforts due to inability to connect between similar/ related entities

Challenges with New Technologies

Given the vast amounts of data sets within firms, advanced technologies such as Artificial Intelligence (AI) and Machine learning, cannot be useful without being fed with reliable data. For that reason, more attention should be paid to data collection, storage, and transformation practices before data analysis take place. Moreover, without the ability of senior management to measure and track the improvement of data quality, it will be harder to show good returns from investments made to replace augment traditional rule-based analysis with innovative technology.

Monitor Everything

To indeed be able to evaluate data quality and work towards measurable governance programs, firms should strive to trace the data from the upstream points and into their end state. To truly qualify the state of the data, one needs to understand the use of the data downstream and the differences between one technology solution to the next, and between one business line usage of the data to the other.

Why Automate Data Quality?

In recent decades, firms focused on getting the primary financial crime and compliance programs up and running under the duress of stringent timelines and with tremendous pressures by management and regulators – who expected to see systems operational as soon as possible, at the expense of good data quality. This has led to what we deem as the “Day 2 DQ problem” – systems that were initiated with incomplete data and had never since been cleaned-up. The only way to assure improvement is to add a programmatic data quality automation layer above the golden source and to run these procedures frequently and consistently.

More on data quality automation in our on-demand webinar below.