Taking the Credit: Developing strategies to mitigate Fraud Risk

“Give credit where credit is due”

We have likely all encountered this renowned proverb at some point in our lives, and it’s never been more relevant than in my day-to-day work discussions about ‘credit.’

Wondering what this means? Let me explain.

At Tamara, we are driven by a mission to empower communities through innovative products and services that facilitate financial freedom. Our principal product, Split Payment, allows users to pay for their purchases in manageable instalments without incurring additional interest, essentially relying on our trust in their commitment to repay over time.

However, this trust-based model is not without its challenges. A minority of users may exploit the system with no intention of repayment, presenting significant financial and operational risks. Such behaviour is particularly concerning in a business model predicated on mutual trust and responsibility.

The growing BNPL market and fraud risks that follow

Consider the global Buy Now, Pay Later (BNPL) market: valued at $6.13 billion in 2022, it is projected to grow at an annual rate of about 26% through 2030.

Furthermore, a 2023 report by Juniper Research, forecasts global payment fraud costs to increase to $358.5 billion in 2023, up from $338.4 billion in 2022 — a 6% rise. These figures underscore the critical need for robust and proactive fraud prevention strategies within rapidly growing businesses like ours.

Understanding the different faces of fraud

When it comes to the forms of fraud we face, it broadly falls into one of three categories:

1. First-party fraud arises when a customer intentionally defrauds a business by making false claims, misrepresenting information, or engaging in other deceptive practices. For instance, a customer might return an item they never purchased or file a claim for a product they never received.
2. Second-party fraud occurs when an intermediary, such as a merchant or payment processor, defrauds a business. For example, a merchant might inflate the price of goods or services or process payments for unauthorized transactions.
3. Third-party fraud occurs when an unrelated individual defrauds a business. For instance, a hacker might steal credit card information or create fake accounts to make fraudulent purchases.

Addressing first-party fraud is complex due to the challenge in discerning honest customer behaviour. In contrast, third-party fraud often allows for more straightforward detection, such as noticing account takeovers or unusual changes like password resets or new device usage.

Moreover, our vigilance extends to monitoring the merchants on our platform, ensuring they too adhere to ethical practices.

Balancing User Experience with Fraud Prevention

Our customer-centric approach, prioritizing seamless user experiences, presents a unique challenge in fraud prevention. Risk decisions must be made almost instantaneously at checkout to maintain customer satisfaction and minimize friction, preventing prolonged user wait times. This requires a delicate balance between protecting the business from fraudulent transactions and ensuring a smooth and uninterrupted customer journey. This means we have to make decisions very quickly once we get user details and try to move detection processes before the checkout process.

In combating fraud, especially within the BNPL sector, we employ several key strategies:

1. Customer due diligence during onboarding — The most crucial step in combating fraud lies in the user onboarding process. For BNPL providers, ease of use and a streamlined signup process are key value propositions. Therefore, we must strike a balance between frictionless onboarding and robust fraud checks. This can be achieved by not only a comprehensive KYC process but also collecting static data points about the user during signup and using these to improve confidence about the behaviour of the user. These details also facilitate the identification of potential Sanctions or PEP (Politically Exposed Person) matches, prompting more stringent checks for these users. Information gathered during signup can be used to develop network graphs of the user base to discover fraud trends, and customer risk scores and even have different onboarding flows based on the transaction risk.
2. Rule-based Risk Assessment — Post onboarding, we need to establish rules to address potential red flags relevant to our business model comprehensively. The optimal approach to developing alerts and controls stems from either the risk assessment, specific business requirements, or data-driven analysis. Subsequently, drawing upon regional and international guidance from regulatory bodies helps us identify additional red flags applicable to our business model.
3. Detection and Mmonitoring — An essential aspect of fraud prevention is establishing robust detection and monitoring capabilities. This entails implementing dashboards and alerts to track crucial metrics such as fraud rates, decline rates, false positives, order volumes, etc. This enables the early detection of fraud attacks and provides a feedback loop to incorporate this information into rule development and risk assessments. Additionally, fraud monitoring necessitates a team of specialized fraud agents who can conduct retrospective customer reviews to gain a deeper comprehension of behavioral patterns and share valuable insights.
4. Ongoing enhanced due diligence — As part of our risk-based coverage, we should also implement ongoing risk-based due diligence. This could involve a periodic review of our customer cohorts based on risk or trigger-based review. This helps us understand user behavior post-purchase as well to make better decisions earlier in the funnel.
5. Machine learning models — As we accumulate more data from our rules and gain a deeper understanding of fraud patterns, we should explore the development of ML features to extend our coverage beyond the scope of our rules. While rule-based risk controls offer the benefits of predictability and explainability, unsupervised or supervised ML models enable us to identify trends before our rules can capture them. Despite their inherent unpredictability, ML models should complement our risk-based rules as we strive toward achieving comprehensive coverage.

The evolving BNPL landscape brings forth a dynamic risk environment, necessitating a multifaceted approach to fraud prevention. The ever-evolving nature of fraud has led to a rise in social engineering tactics that affect countless users on a daily basis. News outlets, social media platforms, and even messaging groups are inundated with stories of unsuspecting individuals falling victim to scams or witnessing large-scale attacks on businesses perpetrated for either data or financial gain.

At Tamara, we’re tackling this head-on with a great team that combines smart tech, data-driven insights, and industry expertise to stay ahead of the curve and effectively combat fraud.

Keeping fraud rates low is our goal and we are happy to take the credit for that.