"Know Your Business" is costing financial institutions millions of dollars per year and automation has the potential to revolutionize how these organizations approach customer due diligence. But can KYB be truly conducted without any human intervention?
Failing to comply with KYB requirements and Anti-Money Laundering (AML) regulations can have dire consequences for financial institutions, as Santander just found out. The bank was fined $132M by a UK watchdog for not having properly checked 560,000 business customers. More than $5B in fines were issued for non-compliance in 2021 alone, illustrating how financial institutions struggle to keep up with increasingly demanding regulations such as the EU’s AML5 directive.
KYB is still a mostly manual, lengthy and resource intensive process that is prone to mishaps. The average bank will spend north of $60M a year on KYC or KYB procedures, and this figure keeps creeping up. Banks could cut the annual cost of KYC by potentially slash those costs by investing in technology. Automating KYB has become a top-priority project for financial institutions. However, automating such complex processes comes with a lot of challenges that we will explore in this blog post.
Know Your Business (KYB) is a verification process performed by regulated organizations on other companies they want to do business with. This framework, also called “KYC for companies”, is used to comply with Anti-Money Laundering (AML) obligations and other regulations.
The term “KYB” was mainly introduced in 2016 after the Panama Papers case, which revealed a global network of companies used to launder money and hide assets. Soon after the scandal, the US Financial Crimes Enforcement Network (FinCEN) updated their Customer Due Diligence Requirements for Financial Institutions, strengthening their business-related rules. The EU followed suit with the 4th Anti-Money Laundering Directive.
Regulated businesses in the US and EU must perform KYB checks. This includes not only banks, but also fintechs and financial services companies, following the 5th AML Directive in the EU and the latest updates of the US’s CDD Final Rule. A lot of other countries are also extending their AML regulations like the Hong Kong Monetary Authority (HKMA) that revised their guidelines in September 2020.
Even if KYB requirements vary by country, they often follow the same structure: identifying the legal entity, its representatives and ultimate beneficial owners (UBO), then performing identity verification as well as sanctions & PEP screenings.
One of the most common challenge when automating Know Your Business (KYB) processes is the lack of unified data sources. This can make it difficult to obtain the necessary information to identify the legal entity and its representatives. Finding the Ultimate Beneficial Owners (UBO) of a company can be tough when dealing with complex ownership structures.
The most challenging situations occur when they are several layers of corporate ownership, including holding companies based in different countries that do not have the same level of transparency regarding business owners. Tracing the ownership chain may thus require the integration and orchestration of many data sources such as private vendors or national business registers.
Regulators are not making things easier. The Beneficial Ownership Registers Interconnection System (BORIS) promised more transparency at the EU level as it went live by end of 2022. But a few weeks later, the Court of Justice of the European Union (CJEU) invalidated a provision of the 5th AML Directive that guaranteed public access to this information. The Court stated that this provision was in direct conflict with the fundamental rights to respect for private life and to the protection of personal data. This marked a significant step-back in ownership transparency and performing such verifications will remain arduous for the foreseeable future.
Collecting information and documents from customers is a necessary step in the onboarding process, but it can also present challenges when it comes to automating KYB.
Institutions are pushing the adoption of more modern technologies such as video-based identity checks, open banking or qualified e-signature, but less tech-savvy customers can be reluctant to use newer verification methods when other might be concerned about privacy. It is therefore important to support different verification methods and have fallbacks when one proves ineffective or the company runs the risk to seriously damage conversion rates and activation.
Backs-and-forth with customers about erroneous documents are also a frequent cause of headaches, with some businesses reporting as many as 70% of KYB processes containing at least one document that cannot be validated on the first attempt or requiring additional information. On top of requiring additional time and effort from compliance officers, this makes the process of automating all the chain of customer interactions even more complex.
On top of this, companies with an international footprint need to juggle with different sets of requirements, familiarize themselves with the preferred verification methods, the most common documents and deliver perfectly localized instructions that will guide the user through their onboarding journey.
This is by far the biggest challenge that companies that are striving to automate KYB need to address.
Optical Character Recognition (OCR) solutions or AI-driven techniques have proved effective at approving common documents such as driving licences with good accuracy levels. Unfortunately manual review remains unavoidable for some unstructured documents that cannot yet be read by a machine. This is the case for Articles of Incorporation, cap tables, proof of funds and many more that are useful as part pf the KYB process.
Human validation will inevitably lead to errors, and is especially challenging at scale when hundreds of people, sometimes contractors are involved. They need to receive appropriate training on the intricacies of each document. Their productivity needs to be closely watched to control costs, their decisions audited and “double-blind reviews” implemented on the most sensitive documents.
The most companies that will be the most successful in automating KYB will be on the lookout for alternatives to manual reviews whenever possible and try to use it only as a fallback when all other methods have failed or on edge cases or risky investigations.
After talking with industry leaders in AI and document processing and Fortune 500 corporations with massive document review needs, we remains convinced that verification will remain hybrid for some time. Therefore, we have worked on a range of techniques to augment the human reviewer and allow him to perform is job faster with even more accuracy.
In a common KYB exemple, you are using a private business data provider like Kompany to obtain beneficial ownership information, verify the identity of each beneficial owner, screen each of them for Sanctions, Warnings and PEP, discard any false positive and match the information with collected documents. Depending on the results, the decision might be escalated to a supervisor or require additional checks, for instance if the client’s company country of incorporation is flagged for poor financial transparency, if the company was created recently or if its activity is considered risky.
To make things even more tricky, KYB should not be a fixed point in time bur rather seen as a continuum. This is why clients need to be monitored continuously for suspicious events, such as a change of control or a newly reported court action. Each of these flag potentially impacting the score, the status or triggering more checks on the customer.
Chaining checks and building different decision scenarios to address various risk situations is very costly in terms of engineering resources. Very few companies can afford to build a custom decisions engine, where rules can be designed and implemented without coding. Yet, it’s the only path towards a more automated and intelligent process.
Automating KYB has been historically challenging for many reasons. Various providers are required at different stages of the KYB chain, either to get exhaustive data or to run checks. This comes with technical challenges to integrate, maintain and orchestrate these solutions but also frictions in contracting and negotiating with each vendor. In addition, risk profile and geographical regulatory specificities add another dimension of complexity. On top of that, there still are situations requiring human intervention, like managing back-and-forth with customers and review unstructured documents.
A modern solution like Dotfile can help regulated institutions increase automation across their compliance process to approve more than 90% of customers in less than a minute without requiring any human intervention. This results in a more seamless user experience and reduction in operational complexity and costs.