How One Airline Prevented Fraud from Taking Flight: An Interview with DataVisor’s Randall Maddern

It’s no secret that flights can be expensive. So when airlines offer great deals, it’s a tough opportunity to pass up. This isn’t just a head-turner for travelers, however; fraudsters are also taking advantage of low airfare deals and turning them into their own revenue streams. 

This was the case for a discount airline based in Malaysia, where fraudsters were using bots to reserve value-priced seats and reselling them via third-party channels to users for a tidy profit. This also kept the legitimate customer base from being able to secure the discounts.

The airline company was aware of the problem and had deployed various tools and technology to curb fraudulent activities. However, these reactive rather than proactive tools were unable to keep up with the variety and frequency of bot attacks. As a result, the airline lost large sums of revenue to fraudulent reservations.

In a recent interview, Randall Maddern, Enterprise Sales Director with DataVisor, sat down to deliver insights into how AI machine learning improved fraud detection and prevention for the airline.

Finding Favor with a Comprehensive Fraud Detection Solution

With the airline industry’s heavy reliance on technology, DataVisor’s machine learning solutions provided a perfect fit because they go beyond the immediate need for fighting ticketing fraud by offering strategic tools to encompass other use cases, such as loyalty program fraud and full-service travel booking. 

Randall noted:

“The airline itself is a conglomerate of businesses. Customers can use the platform to book their flights, hotel, transportation, and more, and get different incentives or discounts in doing so. The airline is expanding beyond the core business of what its initial problem was, and the company found that DataVisor can go beyond this use case and strengthen fraud detection across its entire brand.”

How Airline Fraud Looks and Functions

The airline company already knew that fraud was being committed in its ticketing processes, but upon implementing DataVisor’s machine learning fraud detection solutions, a few specific patterns emerged.

First, the fraud detection software was able to identify fraudsters based on their digital footprint, such as their IP addresses. Looking more closely, however, the fraudulent attacks also appeared to be targeting the same flight routes, specifically one that goes between Singapore and Kuala Lumpur. This was a high-impact, high-volume route for the airline and was likely chosen because fraudsters could more easily go undetected in that high traffic environment. 

Other patterns detected included similar user session durations (usually less than 100 seconds), area codes, and direct bookings (versus coming from a web referrer). What’s more, DataVisor found that the fraudsters were specifically creating bots for different data sets to avoid detection. These are hallmarks of a high-level, organized crime ring that can often evade fraud detection that doesn’t look at data holistically.

“In this specific case, there are a lot of micro signals that are in constant fluctuation, and the only way to detect them is through machine learning,” explains Randall. “The reason we leverage unsupervised machine learning (UML) is that we don’t know what the next attack is going to look like. That’s the benefit of UML that doesn’t rely on data training and previous pattern detection.”

How DataVisor Stops Fraud in Its Tracks

For this airline, Randall remarks that transactions or reservations initiated by bots aren’t confirmed. Once identified, the transaction is brought to a stop, which is a direct result of the automated decision-making provided by DataVisor. 

Randall explains:

“Something unique about our platform is that it doesn’t prevent potentially good transactions from being conducted. If an activity is deemed suspicious, the transaction is paused and flagged for human review. Auditors can manually approve or decline a reservation based on the data we provide.”

The airline deployed two DataVisor tools to improve their fraud detection: dVector and dEdge. Both DataVisor’s DVector and dEdge work to create a comprehensive fraud detection and prevention platform that can help companies mitigate the impact of fraud proactively rather than reactively.

Machine Learning–dVector

To take a proactive approach to fraud, the airline needed real-time insight into their transactions to identify fraud before the transaction is complete. With its ability to analyze all accounts and activities simultaneously in real time to find known and unknown attacks, dVector’s unsupervised machine learning delivers those real-time, actionable insights.

Device Intelligence–dEdge

The increasing popularity of transacting business via mobile devices opens up a whole new window of opportunity for fraudsters. DataVisor’s dEdge is designed to narrow that window considerably by helping companies to learn more about the devices their customers are using.

Randall explains:

“DataVisor’s dEdge helps companies identify whether a device is real or whether it’s being used as an emulator to mimic other devices. Fraudsters can use a device to mimic 100 phones at the same time to go to the airline’s website and make reservations.”

Randall explains that dEdge can verify a device’s authenticity much like a digital fingerprint and can return that information as part of the reservation.

DataVisor Impact: How Fraud Detection Has Improved for the Airline

Randall notes that the airline company looks not only at its bottom-line savings but also at its reduced revenue at risk since its engagement with DataVisor. 

DataVisor’s machine learning solutions also helped the airline to improve the customer experience during the booking process, which can now be simplified due to increased fraud detection. The company believes that providing a great customer experience from the start can also help it to promote its comprehensive booking services for hotel, transportation, and the loyalty program — revenue that otherwise would have been lost due to a complex booking process.

To date, the airline company enjoys an additional 53% of fraud captured beyond what its existing tools found with a 97% detection accuracy. This has also resulted in more than 113,000 hours of seat occupancy time saved per year. Read the complete airline fraud case study here and learn how DataVisor is helping to prevent fraud from taking flight.