Every kid knows that wind blows from high pressure areas to low pressure areas. But it takes much more to become a meteorologist. It takes expertise, devices, and… understanding.
Each and every organization that handles money these days knows very well what KYC is. Just in case the acronym doesn’t ring a bell, KYC stands for Know Your Customer – the process of a business identifying and verifying the identity of its clients. Or at least entities willing to become clients. Some organizations, e.g. banks, are obliged by law to proceed in accordance with certain KYC standards in order to halt money laundering or providing financial services to terrorist/criminal organizations. Others implement KYC procedures and solutions for their own and their customers’ security: to prevent fraud of various kinds. I believe that in either case a fundamental shift of paradigm is necessary for organizations to meet the challenges of the future.
From well-informed to well-understanding
Since KYC is all about being well-informed, the amounts of data collected and processed by business entities for this purpose are often quite impressive – especially when sales are carried out online, where verifying whether the person on the other end of the metaphorical wire is who they claim to be is more difficult than in a brick-and-mortar branch. Hence, it requires more information. Why not use this data also for other purposes? What if we could enrich, augment, and extend these thousands of data points with even more relevant data and move from just protecting the bottom line of an organization to proactively boosting its top line? What if we could broaden our perspective to take a super close look at all users, not just those who have already proceeded to the checkout or filled out the sign-up form, and reach beyond “knowing” to actually “understanding”? These questions have led the Nethone team to where we are today. From military-grade fraud prevention, real-time adaptive customer segmentation and retention tools, through account takeover detection based on advanced behavioural biometrics, up to solutions designed to protect satellites from hijacking, we help organizations make right, data-driven decisions.
Are we maverick business people? Not in the slightest. More and more companies are forsaking the path of hit-and-miss intuitive business decisions in favour of the data-driven approach and, as a result, data now plays a pivotal role in bolstering up customer intelligence in enterprises across the globe. In fact, having a virtual storehouse of information has become part and parcel of running an online business – any business. It no longer comes as a surprise that LMVH, the world leader in luxury goods, is one of the key exhibitors at the VivaTech Conference in Paris, showing off its customer intelligence solutions. L’Oréal, a global beauty company, has just appointed Thales’ Patrice Caine as a director for a four-year term in their efforts to join in this revolution. So, as you see, holistically speaking, data transforms businesses in terms of where they intend to head product- and strategy-wise. It impacts companies and their resources from the bottom to the top. Literally!
A qualitative breakthrough?
In the face of all the challenges emerging in front of progressively-thinking enterprises, I propose a shift from the traditionally defined KYC to a brand-new approach, which I shall call KYU – Know Your Users. By KYU I understand a comprehensive, exercised constantly, in-depth, multidimensional profiling of all users who interact with given online services. The aim of such profiling is to fuel an entire ecosystem of machine learning models trained to solve diverse problems and constantly improve their performance through mutual reinforcement, regular quality inspections and fine-tuning carried out by the world’s top data scientists. In effect, an organization which applies such an approach can not only identify and verify the identity of its clients (enforce its KYC policy), but also instantly understand any online user it interacts with and automatically predict whatever it needs to predict regarding a given individual with enormous accuracy. I think the fundamental qualitative difference is crystal clear.
See more, delve deeper, predict accurately
As explained earlier, thanks to KYU, an organization can effectively execute its KYC, but also always be one step ahead of whoever it interacts with online. In effect, it can finally decipher the intentions and motivation of new users at their very first contact with the service and adapt its properties in real time to what it “understands” as desired by every single one of them. Simply put, it can become for every single user what they want it to be, such as they want it to be, and the way they expect it to be – even if their needs are not “conscious” yet (!). Furthermore, the tools for effective KYU are already at hand, available to any organization ready for the shift of paradigm.
So, how to KYU?
One of Nethone’s core technologies – a cutting-edge profiling tool capable of extracting rich information about users’ software, hardware, network environments and behaviour (biometrics and patterns), is at the heart of the KYU approach. However, it is not enough to enact the idea presented in the previous paragraph. In-depth, multidimensional profiling requires much more data – information relevant to each single problem that the organization aims to solve. Fortunately, the lion’s share of the information needed is often already available for the organization – be it in its databases created for various purposes different than profiling or right on its customer-facing sites – just waiting to be collected. Another element necessary for effective KYU is what I shall call a cognitive apparatus. It consists in IT infrastructure, data science expertise and experience required to create and maintain top-performing AI – the core component in terms of decision making. Nethone’s CPO, Aleksander Kijek, and I, have been for quite a while now explaining the process of machine learning modelling for fraud prevention purposes, so please feel free to check our earlier posts for more information about it. This one, for example. However, to embrace the idea of KYU in this regard, one needs to understand that models trained for specific purposes can actually reinforce each other by leveraging the outcomes of analyses conducted in the past or at the very same time, but in different contexts for new analyses. Just to illustrate the idea; a model identifying prospective frequent shoppers can generate predictions to be used as an additional source of data for a new model, responsible for, say, detection of users likely to churn within 3 months to make their predictions more accurate, and so on and so forth. Creating an ecosystem of models is quite a challenge and requires the involvement of some of the brightest minds – people who understand all the business nuances as well as technical requirements. But it’s much more than just a theory. It is reality, an approach just waiting to be internalised by organizations, a necessity for future-minded businesses, government agencies and NGOs.
Just to preempt your questions about GDPR compliance, if you need to start with KYC, asking your customers for profiling permission brings less friction than forcing them to provide detailed information through, say, a verification form. Secondly, in most cases, effective KYU can be based on non-PII information or executed using anonymised data. Machine learning models usually do not learn about Mr. Smith or Mrs. Doe. They learn about users with certain features. If they learn that a set of features is likely to imply certain action, they issue relevant recommendation for the system the KYU solution is integrated with.