CNP Payments – How to Reduce Costs and Mitigate Risk Through Machine Learning
This interview was originally posted on the goEmerchant Blog
CEO Raymundo Leefmans has acquired substantial experience in the global e-payments industry working in key positions at Ingenico, First Data and Adyen, before co-founding Dimebox. This fintech company offers global Acquirers, Banks and Gateways a state-of-the-art Payment Intelligence Platform.
Shanty Elena: Hi Raymundo, thank you for your interview. Latest studies indicate that 8% -10% of IT budgets are spent on compliance. We see how the e-payments industry is moving from rule-based to data-driven ‘Due Diligence’. In which ways do data-driven frameworks reduce compliance costs?
Raymundo: Great question! Traditionally, financial institutions work with rigid rule-based systems to transfer existing rule-books into IT platforms as part of their risk management protocols. This is very inefficient and expensive and includes very high maintenance costs, because each time rules change, the IT platform requires modification. This process is not flexible at all and costly to maintain and this system cannot always catch up with fraudsters and cyber criminals who continuously update their criminal schemes. Smart-, real-time-, data-driven frameworks incorporate technology which is agile enough to adapt to each new compliance rule, AML measure and to new fraud schemes.
Q: How does ‘Machine Learning’ (ML) increase the speed with which new compliance rules and regulations can be integrated into a company’s existing Risk Assessment protocol?
Raymundo: That’s always an interesting topic. Traditional acquirers or financial institutions have a very static perspective on compliance rules and how they should be applied to secure their business. Machine learning enables you to assess your company’s business based on historical data and actual data running through your system and your payment platform. Assessments which verify at a granular level if all processes are compliant with rules and regulations. Machine Learning helps to analyze and investigate data. The system is flexible enough to adapt and with the knowledge retrieved from the data, the decision-making process can be improved in real-time. On a payment platform which processes enormous amounts of transactions it is crucial to adjust and set parameters and boundaries based on historic and real-time data in order to mitigate financial risk.
Q: Are there other benefits, besides Machine Learning’s ability to help reduce financial risk?
Raymundo: Machine Learning significantly reduces financial risk and operational costs. First of all, your company reduces cost because you need less employees in an operational capacity with manage Machine Learning. Extensive knowledge of compliance rules, fraud and cyber-crime schemes is a fundamental requirement. Secondly, Machine Learning is capable of detecting and mitigating financial risk based on historical and on real-time transactions flowing through the payment system, offering you full control, allowing you to adapt the parameters according to the ‘risk appetite’ of the financial institution (which varies).
Q: In which ways do Business Intelligence (BI) modules help merchant acquirers gain deeper insight into the millions of payment transactions they process?
Raymundo: That is actually a question that touches upon the very fundamental way of looking at transaction processing and merchant acquiring. In the traditional payments value chain each stakeholder takes responsibility for his specific task in the payment process, leaving one to solve unsolved issues. For example, merchants don’t engage directly with acquirers, they engage with their PSP which processes their transactions. The transaction is processed by the PSP and ends up with the acquiring bank which runs financial and reputational risk. So why don’t we try to set up payment platforms which are capable to predict risk before fraudsters attack? At the end of the day it’s the card issuer who gives the authorization, but if you have built a payment platform that is agile and smart enough to incorporate and anticipate card issuers’ requirements, questions concerning specific merchant types, amount thresholds, authentication, etc. specific risk which acquirers face can be reduced or eliminated. By including predictive analysis into your payment platform you reduce costs and ensure higher conversion.
Q: How important are integrated intuitive reporting tools and what developments do you see in that area from a user perspective?
Raymundo: That is an ongoing development in the payments landscape. All financial institutions try to provide their clients with the best information available within their capabilities, whether these clients are customers, merchants or payment facilitators. If you look at the traditional payment value chain, the PSP’s have become more market captive and actually engage with the merchants, providing relevant data and reporting to merchants. The PSP also shares data with the acquirer and vice-versa, which often results in an 85%/15% on average approval/decline rate, depending on the Merchants’ business type. This leads to what I call “Excel-reporting”, that holds no value anymore other than a simple calculation. Merchants increasingly want to understand what is actually happening to payment transactions at a detailed, granular level because this enables them to gain crucial insights, allowing them to understand their decline rates and help them scrape off percentages to increase conversion and resulting profit.
Q: Fintech startups use the latest technologies, integrated in white-label SaaS solutions. What makes Software-as-a-Service so interesting for Acquiring Banks?
Raymundo: Traditionally, IT departments at financial institutions depended on hardware, set in their data-centers or basements. With the current state of technology, this processing power is also available in the Cloud. This has many advantages; increased flexibility in total cost of ownership, which is significantly lower than when you have to manage and maintain your data-centers on your own premises including staffing. Furthermore, the agility when you develop new features, implement road-maps, is far superior when you provide those services from the cloud. This has a tremendous impact on flexibility as whole and on performance metrics around up-time, availability and scalability.
Q: You are mentioning the positive aspects on the SaaS solution in the cloud, but aren’t there any potential pitfalls in terms of security.
Raymundo: Due to the fact that Financial Institutions are regulated by (local) governance bodies, they have to comply with a specific set of rules and well-defined security standards. This explains traditional banks’ reluctance to move their operations to the cloud, but I personally believe that SaaS solutions can improve security and particularly the response to a threat as and when this occurs.
Q: Fintech startups are shaking up the financial industry. What developments and trends do you see in the near future in your specific market?
Raymundo: It’s no surprise that Fintech is booming and startups indeed shake up the FS industry. They’re going after the heart of the whole global financial industry and specifically after traditional banks. From micro-lending, to KYC to fraud analysis; fintech startups that specialize in a very specific area which is traditionally managed by a bank are perceived as disruptive to bigger banks, which often lack the flexibility to adapt their infrastructure, processes and procedures within a short timeframe to keep up with the real world and meet actual market demands. Small innovative startups haven’t reached the critical mass yet, but banks know that they have to reinvent themselves to reclaim their position in the whole payments value chain and therefore they are forced to evolve and innovate as well. These developments offer huge opportunities for fintech startups such as Dimebox. While we see a lot of startups tackling specific issues in the FS industry, Dimebox tackles a variety of technological and risk assessment issues, while dissecting the card schemes. We do not focus on one specific aspect in the value chain, we offer end-to-end solutions and we call this full-stack processing.