Research indicates that many businesses are reluctant to embrace a data-driven culture, but anyone who fails to do so will miss out on the benefits of data-led 'know your customer' controls. HNW individuals do not have static identities and risk profiles, so the data-led approach is the way forward.

Firms are only embracing data science for KYC purposes at a slow pace. This can be attributed to a misplaced anxiety about whether new software packages are compliant, to confusion over the volume of software programmes available on the market and to fear of the unknown.

Despite widespread hesitancy, the solution is simple and readily available: strengthen the records you keep of your customers by using large sets of cross-industry data attributes (including screening records) and compare them with consolidated internal data attributes for each customer. If you keep those attributes dynamically enriched, you will know about any changes to identity and risk immediately.

This will enable your firm to identify duplicate accounts in its own records quickly, trace lost customers and update files to ensure that its view of its customers is accurate.

This sophisticated data-matching capability is available now. It is being embraced by some, but many more firms continue to use less effective, more costly methods.
 
Overcoming ‘explainability’

Artificial intelligence or AI, in the broadest sense, is simple: it is the use of computers to perform tasks that could otherwise be performed by humans. More sophisticated forms of AI, such as machine learning, use algorithms to analyse large sets of data and thereby spot patterns that lead to specific undesirable things (e.g. fraud/money laundering/payment default/corruption etc), all without human intervention. The more data that one feeds into the machine-learning programme, the more sophisticated the pattern analysis can become and the less easy it is for people to perform the same analysis.

At the most complex end of the analytic process, ‘deep learning’ can use either structured or unstructured data to analyse patterns on many layers in order to spot events that might have otherwise been impossible to predict or identify. A bank can use the same thing to 'reverse-engineer' some analysis from an event, thereby spotting the attributes that could have indicated that event (e.g. fraud or money laundering). Such complex, multi-layered analysis is helping firms to fix problems in a broad range of industries – from medicine to aeronautical design and manufacturing.
 
Despite its considerable successes, the financial services sector has so far failed to reap the benefits of this approach. This is largely because ‘deep learning’ struggles to explain exactly how it arrived at a particular conclusion. This jars with financial institutions, which must be able to explain how and why they reached this-or-that decision to their regulators. Before long, the industry will have to decide whether it is preferable to have ‘explainability’ or to have more effective, game-changing risk-management IT.

In the meantime, some banks are using simple forms of data-led statistical modelling and machine learning AI software that ‘comply’ with regulations fully to find patterns and associations in large data sets. They are doing this most successfully and already improving their traditional control models no end.
 
To understand the possibilities

In an increasingly data-led world, it is crucial for organisations to understand the value of data (when used correctly) and all its possibilities. By acquiring the right technology to manage the vast amounts of data that they need to help them make decisions, banks can truly reach their full potential. Knowing one's customer is no longer an option; instead, it is a pre-requisite for the survival of one's financial business.

It is clear that richer data sets are capable of improving both the experience of customers and the insights that their banks have about risks. They can also reduce the cost of data-related errors, duplicate accounts and inefficient decision-making processes. With that in mind, banks would do well to consider the competitive advantages they could gain by taking up such technology.