Is Straight Through Processing the silver bullet for reducing false positives?

“How can we reduce false positives?” is the million-dollar question facing the banking industry. Costly and potentially harmful from a customer service perspective when a legitimate account is frozen unnecessarily, the silver bullet is yet to be found.

But this problem lies in the very nature of fraud detection engines used during the KYC process – the technology is designed to surface every single false positive hit to show that the proper compliance processes were followed and, therefore, avoid allowing any possible criminal attempts to slip through the net, as missing any alerts can lead to substantial regulatory fines. Hence, the reduction in false positives is very difficult to achieve.

However, the route to real efficiency gains within financial crime could be achieved via Straight Through Processing (STP) of these false positives.

The benefits of STP

Many financial institutions have hundreds, if not thousands, of individuals working on a variety of financial crime applications simultaneously. But over time, these applications have grown in functional vertical isolation leading to risk blind spots, meaning banks are unable to truly create a holistic view of risk.

But by utilising strong case management with business rules built into the underlying platform, combined with AI, banks can bring together the data from their multiple Financial Investigation Units (FIU) and finally break down historic departmental silos. As a result, banks can realise an aggregated overview of client risk, bringing with it an ability to prioritise false positive alerts based on risk types and historical patterns of customer behaviour, as well as predict likely outcomes.  

Having this more detailed understanding of their customers’ “normal” patterns of behaviour places financial institutions in the unique position to make more informed decisions when analysing fraud hits from detection systems. They can also gain a clearer insight into the risk patterns of all parties involved. This will allow institutions to straight through process a large number of alerts that would otherwise require investigation.  

This is a truly practical application of AI within the area of AML & KYC for financial institutions and some banks are already reaping the rewards. Using this approach, one top-tier bank is now able to process 20,000 case reviews with just fifteen analysts for one department. By utilising automation technologies, they have been able to bring together all the underlying streams of financial crime - transaction monitoring, payment screening, adverse media and watchlists. As a result, their teams can leverage historical activity to spot unusual patterns before they turn into risk and have significantly reduced the amount of alerts their team has to review on a daily basis.

Furthermore, institutions could utilise “whitelists” to support sanction compliance efforts by identifying and saving repetitive matches that materialised as a false positive. So, as the “hit” is identified as a false positive, the matching criteria is saved on the “whitelist” so future hits can be screened against it. 

How to put STP into practice

First - an approach already adopted by one domestic bank in mass market - is to start by shortlisting your golden sources of data, i.e. identify what can be defined as a “true” source of your client data and create a pooled resource that can be used throughout the organisation. This pooled resource can then be accessed by your case management system and cases can be STP’d to a “closed” state with the right rules in place. If this is not possible, banks should at least attempt to quickly link to those underlying golden sources of data whether they be external or internal.

The next step is to implement “Process Optimisation”, to determine what customer information really needs to be captured at the onboarding stage to assess their risk level. To realise meaningful change, try to reduce the standard 100 questions you normally send to the new client, to the really critical 20 questions to gain the insight required to assess their risk categorisation.

Another consideration is to design your Case Management system to not only capture the risk level associated with a customer, but also assess it against the bank’s risk policies. This information can then feed into an AI that can automatically decide the best path to process the alert in the most efficient way.

Overcoming barriers to change

Granted, the ideas above require resources and careful planning upfront, but the outcome is real tangible cost and time benefit. Once designed and tested properly, banks can almost fully automate the KYC periodic review processes. Think of the cost savings in remediation exercises alone!

While it may be easier to achieve in the retail banking space, it is much harder to do in corporate and investment banking (CIB); however, it is feasible. Even if you are able to achieve 15% STP, your organisation will have more time for the more complex investigations we typically see within CIB.

Time will tell as to whether Straight Through Processing is the panacea for reducing false positives. What is for certain is that by taking advantage of technology and tools such as Artificial Intelligence and Case Management you can achieve a large degree of automation while removing internal silos. Thus, achieving efficiency gains while managing your risk exposure.