Digital businesses are facing increasing pressure to enhance their fraud prevention measures, particularly as AI-driven fraud tactics evolve. The emergence of synthetic identities and generated documents poses significant challenges to traditional verification methods. Automated attacks enable organized fraud groups to exploit vulnerabilities quickly.
Know Your Customer (KYC) checks are especially vulnerable, with seasoned fraudsters employing AI to imitate genuine customers. "Currently, KYC is the most targeted layer, where real-time document generation and deepfake face-swapping are becoming remarkably sophisticated," explains Volodymyr Todurov, CEO of Frogo, a fraud prevention and risk management firm.
However, the consequences of overly stringent fraud controls can be detrimental to business. While aggressive measures may mitigate suspicious transactions, they can also inadvertently block legitimate payments, slow down withdrawals, and drive away customers. This leads to a paradox where a business’s fraud rate decreases at the expense of valuable clientele.
Frogo, which integrates device intelligence, automated risk scoring, and real-time alerts, suggests that businesses reevaluate their approach to fraud prevention. "Operating with a ‘guilty until proven innocent’ mindset and relying on aggressive, automatic blocks," Todurov cautions, “Treating legitimate players like suspects completely destroys the customer experience.” This perspective highlights a crucial standard for measuring anti-fraud effectiveness: it must safeguard profitability while accounting for prevented losses, operating costs, and impacts on honest customers.
Measuring the commercial impact of fraud prevention requires a comprehensive evaluation. Todurov characterizes anti-fraud as a "revenue optimisation engine," which should be assessed through three key metrics: direct loss prevention, operational efficiency, and customer retention and lifetime value.
Direct loss prevention measures the monetary losses linked to fraud, such as fraudulent withdrawals, and whether these losses are decreasing. However, the resources needed to achieve this reduction are equally important. If routine checks necessitate continuous manual oversight, businesses may struggle to scale effectively without increasing their risk teams.
"Without a nuanced evaluation model, you risk blocking your highest-revenue players,” Todurov adds.
Another critical metric is the number of monthly active users that each analyst can handle, alongside the time taken to investigate cases. Streamlining routine decisions can ensure that specialists focus on more complex threats. The third metric emphasizes the treatment of legitimate customers. Todurov proposes that “automated withdrawals built on the presumption of innocence” can foster smoother transactions, thereby enhancing trust and long-term customer value. Delays or excessive checks can incur costs that might not appear in fraud statistics yet still affect customer satisfaction.
These metrics must be analyzed collectively. A system may effectively flag fraudulent actions yet overlook significant fraud elsewhere or achieve loss reduction through controls that repel legitimate users. Understanding what the system fails to catch and the costs of its interventions is vital for financial outcomes.
High-value customers highlight the risks associated with harsh controls. Their activity can differ significantly from average users; for example, frequent travelers may exhibit sudden location changes or large transactions that could trigger false fraud alerts. Without sufficient context, businesses risk rejecting some of their most valuable patrons.
Volodymyr Todurov points out, "Without a nuanced evaluation model, you risk blocking your highest-revenue players." Frogo believes that unusual behaviors should prompt informed assessments rather than immediate rejections. Specialists who comprehend the product and its customers must decide which signals require intervention, drawing on their expertise to determine where AI can enhance decisions and where established rules are more effective.
"AI isn’t a silver bullet," Todurov emphasizes. "Its success depends on specialists who know precisely where machine learning adds value and where tried-and-true, deterministic logic is still superior."
These specialists also require the right information to make informed conclusions. Frogo advocates collecting user data from their first visit, starting with device identification and building risk assessments as user activity progresses. "It all begins with device fingerprinting, where we analyze hardware and software parameters to verify user legitimacy and detect bots, spoofed devices, or masked origins," Todurov describes.
Frogo provides a standalone Device Fingerprinting tool that can augment existing fraud prevention systems, particularly for businesses with basic in-house rules but limited device recognition capabilities.
This tool clarifies the context of signals like sudden location changes, allowing analysts to determine whether the activity is consistent with a returning customer or if further investigation is warranted. "Anti-fraud tools should be viewed as core data engines," Todurov asserts.
The tool evaluates over 300 client-side signals, facilitating the creation of identifiers that serve to maintain continuity within a single domain and recognize devices across a brand portfolio. It also provides context such as IP risks, VPN usage, browser specifications, screen resolutions, and hardware characteristics. Other signals include time-zone mismatches, remote desktop tool usage, and virtual machine detection.
Within Frogo’s platform, actions such as transactions are examined against a business’s predefined risk rules, with system responses generated in under a second, detailing activated risk triggers.
The usefulness of this information extends beyond merely identifying threats; it can also assist in recognizing positive customer behavior. "Anti-fraud tools should be viewed as core data engines," Todurov reiterates. "Instead of just identifying bad behavior, you can define triggers that highlight high-value, ‘ideal’ player behavior."
Frogo’s real-time alert system routes negative risk indicators to analysts while positive behavioral alerts are sent to marketing and VIP teams, allowing different departments to respond to the same activity according to their responsibilities.
Applying this fraud prevention approach in new markets requires ongoing adjustments as customer behaviors, payment preferences, and attack strategies evolve. Businesses must revise their rules to adapt to varying circumstances.
The overarching goal is to identify genuine threats, minimize unnecessary workloads, and simplify the experience for legitimate customers as businesses grow. "The winners will combine a highly skilled fraud team with agile technology, viewing anti-fraud through a growth mindset," Todurov concludes.
