×

blog

Bots and Humans Team Up to Fight Fraud


With technology changing and cybercriminals becoming savvier, preventing fraud-related issues is a continuous process for brands—one that demands a mix of human intervention and technology tools. At the same time, consumers want a platform they can trust to conduct commerce/secure transactions but don’t want to endure extra hassles for the privilege. This puts an additional burden on businesses: To increase security without complicating the customer journey.

The challenge gets more complex as digital channels increase. The advent of omnichannel provides customers with many ways to reach out to a brand if they have questions or wish to make a purchase. Financial institutions, online retailers and other companies across various verticals have started offering omnichannel support to provide faster and timely resolutions to their customers. Similarly, fraudsters are also getting sophisticated, finding new loopholes in disconnected systems or unchecked processes. This is why it is critical to create and maintain a continuous fraud prevention strategy that leverages the power of both technology and people across the business as the market evolves.

Technology should be any company’s first line of defence when it comes to battling fraud. Fraud identification filters, algorithms, and knowledge management databases power fraud detection bots that can react quickly – in less than a second – and alert companies to millions of attacks and suspicious activities. But bots alone won’t get the job done. They’re great at detecting fraudulent activity, but they are not perfect.

Automation may inadvertently flag legitimate transactions as suspicious, and vice versa; and there are many cases that are too complex, new, or unwieldly for bots to accurately handle. Therefore, a team of people dedicated to fraud prevention, detection, and mitigation is needed.

Additionally, learnings and trends observed by front-line agents can be analysed to study new patterns of fraud attempts, and these patterns are then included in automation algorithms to improve efficiency.

Fraud prevention’s impact on the customer experience

It is very important to have harmony between security and customer centricity. A genuine transaction, if cancelled, can lead to customer dissatisfaction and impacts brand name. And if a fraudulent transaction is approved, it can lead to financial losses by means of chargebacks. Hence, it becomes imperative to have appropriate measures in place across service channels that leverage automation as well as human intervention. Here are examples of the human/AI balance.

Automation: Whenever a user adds a payment instrument (credit card, bank account) to their profile, they will receive a one-time password to verify this change. If the user is genuine, it will not take more than five seconds to verify the payment instrument and his/her transactions will be completely secured. However, if the user is a fraudster, they will not be able to provide the one-time password and the account will remain safe.

Human: Most systems are set up to block users whose IP locations change from their usual location or if their device has changed. However, most of the times this happens when a genuine user might be traveling or using a friend’s device. At this juncture, the system flags such cases to the agents working on the case. The agent uses multiple resources (social networking sites, history of old IPs and devices, reservation history, etc.) to determine if the user is genuine or not. In such instances, human intervention is critical for context and analysis.

At TTEC India, fraud analysts review every case using static and dynamic indicators. They examine evidence from many sources: user-supplied information, proactive information search on the parties involved, digital fingerprints detected behind all user activity, profile building through Internet searches, and history on the platform, just to name a few. They focus on inconsistencies and abnormalities in information. The process gives the analysts valuable insights that are passed along to the companies, with the goal of detecting newer trends and evolving the risk mitigation infrastructure.

Our analysts perform a complex story-building exercise with each case that they review every day, using both static and dynamic identifiers. They sift through evidence from a wide range of sources – user-supplied information across different stages of the transaction life-cycle, proactive information search on the actors involved – digital fingerprints detected behind all user activity, profile building through internet searches, history on the platform, and collaborating with connected businesses in the ecosystem. They focus on inconsistencies and abnormalities in information, and deftly navigate their world where it is crucial to differentiate between the naive, the sophisticated, and the authentic.

Our goal is to detect trends, help update detection and prevention algorithms, constantly anticipate the next counteracting move from the bad actor, and detect and remove design weaknesses, to sustain and evolve the risk mitigation infrastructure.

Fraud prevention is not only necessary, but mandatory for better customer experience. With the right technology to support the strategy, it is easier to avoid fraud and minimize losses. However, human intervention in a process flow is important to protect customers from fraud as fraud trends are evolving and to analyse such trends human intervention is required to understand the patterns and translate such trends in terms of building the right frictions in AI / automation.