South Africa’s fintech sector is growing exponentially, with 71% of consumers now using fintech to perform five or more everyday financial tasks. Consequently, a number of regulations to govern the industry are currently being developed, amended, or discussed.
“According to local law firm Bowmans, one of the main regulatory compliance areas presenting issues for South African fintechs is anti-money laundering (AML) and financial crime. AML technology has therefore become increasingly integral to the governance of these companies,” says Bradley Smith, co-founder of RelyComply, a platform for KYC (know your customer), AML and CTF (combating the financing of terrorism).
“Traditionally, investing in AI-powered AML capabilities was not something on the radar for fintechs looking to take the market by storm,” he explains. “However, over the past 15 years, the tide has turned on this sort of thinking, with savvy companies recognising the importance of prioritising AML investment.”
Smith, who is also the founder of several fintechs, including an AI solutions business catering to banks and insurers and an on-demand insurance company, offers three tips for how fintechs should handle AML to meet the challenges of regulatory compliance, while also remaining relevant in an industry that is only getting more competitive:
Next-generation KYC
“Fintech startups are often somewhat strapped for cash, especially when seed or series A capital raises have been small. In such cases, the focus is largely on solving the customers’ financially-related problems. KYC and AML compliance are usually seen as hurdles to be overcome with the least spend,” he points out.
“The result is often either a minimal and hastily constructed in-house solution or the procurement of a basic suite. Yet, with business traction and product analytics comes insights, one of which is the customer drop-off rate associated with mediocre KYC tooling at sign up. The customer experience is hampered and, in turn, with drop off comes a wane in word-of-mouth referrals. For those users who stick it out, confidence in the product dips and customer churn is far more likely.”
Smith adds that there are a number of other AML tooling red flags, prime amongst which is inadequate identity validation where processes such as extracting information from documents, facial verification and liveness detection are not of sufficient calibre.
“Other customer onboarding deterrents include simplistic watchlist screening with high false positive rates and poorly constructed customer risk assessment (CRA) criteria that result in the unnecessary classification of people as high risk. The ensuing due diligence will do little to buoy customer satisfaction, as no one enjoys sourcing and submitting obscure forms to prove their worth.
“Leveraging state-of-the-art AI, however, can help to straighten out early-stage sign ups,” he suggests. “This also ensures that onboarding is swift and hassle-free, allowing for the best possible user experience. CRA solutions should also provide what I call ‘sane defaults’ that are optimised for acceptable business risk, but also allow for easy configuration to adapt the settings as needed.”
AML at scale
“With growth in product and market complexity comes the need for expanded AML solution capabilities. Where in-house solutions have been devised, these need to be extended, but such development work is often difficult to prioritise in a backlog,” says Smith.
“In the case of simplistic vendor solutions, functionality may be limited or lacking, resulting in ‘hacks’ or ‘duct tape’ being applied to work around this. Sometimes supplementary systems are procured, which serve to increase system complexity and risk, as multiple AML systems need to be integrated into one another.
“Additionally, many of these systems are not configurable through a ‘DevOps’ approach, resulting in a challenge when it comes to pushing changes, as releases need to be carefully coordinated on a manual basis.”
He advises fintechs to look for end-to-end solutions that can grow with a business, alleviating the need for those hacks and duct tape. “This should work with a fintech’s modern technology stack and must be easily integrated and customised accordingly.”
Growth and alert noise
Smith shares that gaining traction and drawing in hordes of new customers is on every fintech’s to-do list. “Rising numbers, however, mean surging alert volumes, which can become overwhelming for AML teams to manage. The result is often significant backlogs in handling alerts and less time for compliance teams to adapt to new and ever-evolving patterns of money laundering.
“There is therefore a need for a system to handle transaction monitoring at enormous scale to detect money laundering based on well-defined rules that search for particular patterns. Machine learning is integral here, adroitly discerning false positives with speed and ease. Moreover, complementary forms of machine learning can pinpoint anomalous behaviour, flagging money laundering where rules have not yet been configured. For fintechs, this means scaling transaction monitoring efficiently and effectively as the business scales.
“It’s all a matter of priorities. Certainly, early-stage fintechs are juggling many balls, but prioritising high-quality AML early on paves the way for optimising customer experience and long-term business value,” concludes Smith.