Machine learning, deep learning, and natural language processing in the onboarding process for SMEs and corporate clients
Corporate banking is a highly formalized and difficult area of banking services due to several reasons such as the type of entities serviced, and specific products offered (mainly credit and related services). Many entities, including start-ups and early-stage inventors, wishing to obtain additional financing or guarantees have to undergo a long and burdensome procedure called ‘onboarding and Know-Your-Customer – KYC process’. This process is usually painful for both the bank and the customer due to specific internal requirements and operational challenges.
Many documents that are required during this process are not in a unified form and digital format, they are mostly in the .pdf format or ‘hidden’ in the public registers. As a result, the analyst must read all and whole documents to extract certain data that is required according to the specific rules or procedures. In addition, many other actors may be engaged at this moment, including legal advisors and risk analysts, to do – don’t be surprised – the same. This is not only a burdensome and boring task but also vulnerable to human errors and costly in terms of the operational budget.
Most of the documents that are subject to this process are quite similar or typical as legal and regulatory requirements are usually the same for the same entities. Therefore, this repetitive task may be automated and facilitated bringing savings for the bank and the client as one ‘collective’ request can be made.
This can be achieved by using natural language processing (NLP) and machine learning. The NLP tools can not only gather the required information listed for example in the onboarding procedure but also catch ‘risky’ areas that should be further investigated by a human. NLP with its high effectiveness and ‘context thinking may help to grasp a full view and situation of the future customer in seconds not hours. Many regulators and supervisors within the financial sector, including the European Banking Authority, are not only aware of the potential of so-called RegTech tools but they are creating opportunities for such solutions to be adopted widely.
In addition, the information gathered may be used in other procedures and processes, such as anti-money laundering and risk management. For example, the data from the financial statements and internal resolutions may not only be used for the onboarding but also serve as a base for the creditworthiness and risk assessment and further periodical reviews (if automated – the whole process may be frictionless for the client).
The implementation and the ‘execution’ of the NLP tools do not have to be cumbersome, even though, legal and regulatory requirements will have to be met. The operationalization of the process will require some activity from the institution (so-called ‘feedback’) but will inevitably be a game-changer for the institution and people within the organization. Personalization will be a key factor. Don’t miss it.
