FinTechTalk: Future proofing your KYC processes

On 28 November, FintechTalk host Charles Orton-Jones was joined by Payel Sarkar, Program Head - KYC Compliance Transformation, Société Générale; and Howard Wimpory, KYC Transformation Director, Encompass.

Views on news

Globally, banks took on average 95 days to complete a KYC review in 2023, up meaningfully from 84 days in 2022. This trend is particularly stark in the UK, where firms take on average 17 more days to complete a KYC review this year than in 2022. Nearly half (48 per cent) of banks globally said they have lost clients due to slow or inefficient onboarding processes, a figure that falls slightly to 39 per cent for banks operating in the UK. Switching from manual to automated KYC processes is not easy and there is a certain amount of inertia too involved in banks not adopting new ways for doing their KYC. Manual processes tick the box for compliance, but that doesn’t mean that companies aren’t keen to improve their processes in the medium term. While technology available now can make the jump to automation much easier, record keeping and data availability continue to be a challenge. 

How can you modernise your KYC processes?

The transformation should be gradual, there is no big bang solution. First, break down the processes you want to automate into 3-4 categories, one of them being data. Identify which data drives KYC risk identification and make sure you have the data and nothing but the data, otherwise you’ll be automating white noise. Then introduce the components that’ll make your KYC ecosystem work, such as a data searching and beneficiary ownership discovery component, a workflow system (CLM platform), plus a data entity resolution.

There are lots of data sources that financial institutions can use including social media. The AI and blockchain solutions available now off-the-shelf are designed to reduce false positives. BNPL providers often use as many as 100 different data sources – some of them quite arcane. What banks are looking out for is disconnects in data on the beneficial owner, handovers that haven’t been properly documented or out-of-date data. At onboarding a client, the sources that banks can rely on for KYC are limited, and shell companies are easy to set up across the globe. The real efficacy of a bank’s KYC procedures comes into play when the client starts transacting. The client can present itself as a bona fide organisation at the start, but it’s the organisation’s behaviour as a client that will show the actual risk a client presents, as well as any unexpected changes in behaviour. 

Reviews that have taken up to 5 years done manually can now take place in near real time. The changes to look out for are country, address or the entire documentation and once these changes are detected, a targeted review of the change will be conducted instead of a full KYC. Red flags can be found by comparing missing or out of date data with the same data featuring in other sources. Data connectivity works with some countries but not with others, where API connectivity doesn’t exist, in which case you need to download complete documents and read them to extract the relevant information. In the latter case, technology can help tremendously with data extraction. To get internal data in order is one of the most complex processes of KYC. Technology can speed up these processes, but it’s key that people trust the technology that automates processes for them. Explainability of technology to regulators is a central factor in this. 

Panels’s advice

Strip data you have back right to the barebones and then build it back to avoid having superfluous data. 
Biometrics are going to be next big thing in KYC, although data sharing is constrained by personal data protection regulations. 

• There should be a combination of buying solutions and developing them in-house.
•  Don’t fall for one-stop solution vendors. There is no such thing in this area. Although the dial is moving to “buy”, it will never be just one or the other. 
• Make sure your data analysts are analysing and not just collecting data.