Can Cognitive Transform KYC?

Picture this: A young professional sitting in front of a group of monitors — four all stacked together into a grid. On each monitor they’re doing a different search on one person, trying to find everything they can about their financial history. How many accounts do they have? Where are they? Do they use any other names? Have they ever committed a financial crime? From each screen, they are trying to find pertinent information to pull into one comprehensive report, but there is so much out there that it takes forever. They swivel between screens, trying to read everything, get what they need, and move on. This process can take days, and while they want to get it right, they also want to get it done. It’s long, tedious, and pretty boring.

This is the first line of defense against financial crimes.

Know Your Customer, or KYC, starts this way in every bank. A low-level analyst gets the often tedious job of data aggregation — the first round of research into every entity. Considering that these searches are used later in the Anti-Money Laundering (AML) process, they are crucial. And yet right now, experts estimate that roughly 80% of analysts are doing the work incorrectly, or inconsistently. Daniel Gotlieb of IBM says this, “I’ve spent time in the field, and I’ve seen how analysts work across the board. Right now, the process of gathering data is manual, and there aren’t good data collection processes across banks or the industry. Most every analyst does their job differently, which takes a lot of time and yields inconsistent results. In the end it’s a huge problem for banks.”

KYC alone costs banks between $60 and $500 million a year, according to a report by Thompson Reuters. And it’s just the first step of the process that works money and the world safe. Given the time, effort, and importance of these processes, banks need to find better solutions. Add in the burgeoning expense, risk and compliance challenges required by Bank Secrecy Act / Anti-Money Laundering (BSA/AML) regulations — and help is absolutely necessary. Mike Andrud, Big Data Lead for IBM financial services, says this, “Regulators have exposed many banks incomplete or ineffective risk management processes, presenting them with thousands of Matters Requiring Attention (MRA’s) Consent Orders, not to mention levying large fines upon the worst offenders… It’s not good, and the regulations will only continue the charge. This coupled with the growing amount of data will make it harder and harder for banks to keep AML under control if they don’t find a solution now.”

Luckily, Recent advancements in technologies such as Cloud, Cognitive Computing and Machine Learning are showing promise in cutting manual activities in the process while reducing risk. In fact, these capabilities may make it possible to cut the manual efforts of Know Your Customer (KYC) compliance by at least 50% while improving banks’ overall ability to reduce the risk of Money Laundering.

Banks are looking for cost relief, efficiency, and accuracy that will allow them to conduct compliant KYC at all levels from the more basic Customer Due Diligence (CDD), through the more intensive Enhanced Due Diligence (EDD) processes. A combination of Cognitive computing and Machine learning capabilities can help in these areas.

Cognitive Computing + Machine Learning is the answer

Cognitive Computing combined with Machine Learning can build a better KYC process. Analysts need a better way to aggregate and analyze data, which is what cognitive engines are so good at. If data aggregation were easier and more consistent, it would free up time and manpower to actually investigate legitimate alerts, which should be the most important focus of KYC/AML activities. Because of the amount of data being created each and every day, it’s going to become harder and harder for analysts to pull all of the information they need to report on an entity in a timely fashion.

Cognitive computing capabilities are quickly becoming more adept at searching through data sources and pointing out the most relevant information to the KYC analyst as well as identifying a lack of information, which may help identify a “shell” business set up to potentially launder money. In addition, Cognitive computing can identify business relationships (via link analysis) between related or associated parties displaying potentially suspicious combinations which would warrant further study by a more senior investigator.

The KYC analyst would be able to set up ongoing monitoring of these varied data sources to continuously monitor a customer’s risk profile. As significant changes are noticed by the Cognitive system, AML department would be alerted to legitimate issues. A positive benefit of this capability to better and more accurately risk rate a potential customer is the reduction of False Positive AML alerts — all of which must be reviewed or investigated. This would vastly reduce analyst workload.

Machine learning is also a promising capability in that it could be used to provide continuous monitoring of transactions and be able to better identify if a particular transaction is worthy of follow up investigation, given the systems analytics of historical transaction patterns and behaviors. A major advantage of machine learning is that the ability to monitor previous transactions will establish a pattern of customer behavior. Thus, the system will be able to flag truly new and unusual events or provide the AML investigator with an analytical context to understand the nature of the transaction. The net result of this is to focus investigator time on those money movement transactions most interesting from an AML risk perspective.

The Proof is in the Use

Cognitive Computing layered with Machine Learning has the potential to be an effective solution for many banks, and it’s even gone beyond the ideation phase. We now have use cases that prove that these technologies can change the game for KYC/AML, and save banks time and money.

A Major Canadian Bank reduced watch list checks from between 8 and 12 hours to less than 15 minutes, increased name checks from 2,500 to more than 40,000, reduced false positives by 75%, and realized ROI in 3 months. Likewise, a South American Bank improved efficiency by 60% by reducing administrative costs. They also reduced AML alerts by 90% which in turn increased accuracy by 60%.

Banks Need to Transform Now

Implementing cognitive solutions as the first line of defense against financial crimes is a must, and executives need to realized that it is one piece of a total transformation puzzle. Andrud points out, “The mitigation of KYC/AML risk starts with a clear and full understanding of money movement between associated parties. Today, due to the siloed nature of systems across bank product lines and the sheer volume of external data sources, this is a difficult process.”

In addition, understanding the potential relationship between parties is a difficult task, due to the lack of true customer master data management and a governed enterprise data lake in most banks.

In the future, however, it will be important for data to have more crossover — the same big data required for KYC/AML is also the same data needed for marketing and revenue growth analytics, such as next best offer, increasing account relationships per customer and customized pricing. In the absence of a strong enterprise data governance environment, compliance marketing teams, etc. are forced to operate in separate data and analytics environments, which drives up cost and decreases efficiency.

While there are many exciting advances on the table, Gotlieb cautions that new technology isn’t a catch- all solution. “We don’t see these capabilities as a replacement to traditional alert management systems used in the current KYC/AML processes,” He says, “However, these capabilities will significantly improve manual and data intensive processes and workflows while helping to mitigate ongoing KYC/AML risk.”

Over the next few years, we will see these capabilities mature even more. This has started in forward thinking banks that have the vision and technical acumen to understand and harness the power of these opportunities. All banks, however, need to embrace these updated processes at the risk of falling behind. Given the potential upside of fine avoidance, MRA’s, Consent Orders and Legal Actions while strengthening the overall banking system and lowering bank expense ratios, the time is now to apply these advanced technologies to one of banking’s toughest and most important issues.