Layer 1 copy

Their learning is fast and their Sift Scores are reliable. We are confident that if Sift gives a high score, it is without a doubt fraudulent.

Gustavo TontiFraud Manager at Destinia

Overview

  • 1000s of daily transactions

  • 70% desktop and 30% mobile bookings (in 2016)

  • Experiencing exponential growth worldwide

Challenge

  • Card-not-present payment fraud

  • Manual review couldn’t scale

  • Chargebacks hit 2+ months post-transaction

Solution

  • Integration with 1 developer and 1 fraud analyst

  • Automation on Sift Scores

  • Data visualizations to block fraud rings

Results

  • Incredible accuracy right off the bat

  • Less friction for good customers

  • Better understanding of user base

Overview

Destinia is a rapidly-growing, Spain-based online travel agency (OTA) with offices in Madrid, Cairo, Dubai, and Tehran. There are more than 2 million global travelers using Destinia’s services in 90+ markets. Their website is accessible in over 30 languages, and offers over 500,000 hotels, 600 airlines, and all the travel-related services a traveler might need. The company is among the top 5 OTAs in Spain, and sees 70% of users booking through desktop and 30% on mobile.

Destinia is committed to a great experience for both their customers and their business partners, and has worked to stay competitive in an ever-changing market. With just a small team dedicated to managing fraud, the company has been agile and forward-focused.

Challenge

The online travel space has become increasingly crowded, as customers have access to countless new platforms and tools to book flights, tours, and accommodations. In order to stay ahead of the potential for fraud related to card-not-present transactions, Destinia turned to Gustavo Tonti to oversee a sustainable Risk & Fraud practice. Gustavo’s personal motto is: “Having the honor to serve our good customers has given me the experience to detect the ‘bad ones’.”

Given the global nature of Destinia’s offerings, payment fraud is at the forefront of their fraud challenges, followed closely by fraud rings and occasional friendly fraud. However, the nature of the product – quick access to flights, hotels, and other digital bookings – made manual review unscalable, since the team had such a narrow window to investigate hundreds of suspicious orders daily. When chargebacks did hit, it often took over 2 months for the fees to appear in Destinia’s books, throwing off numbers and affecting analytics. In order to prevent chargebacks, rather than simply respond to them, Gustavo felt that it was necessary the invest in a solution that required less hands-on maintenance and increase his team’s efficiency. To stay ahead of the inevitable surge in fraud, Destinia turned to machine learning.

“We didn’t have a serious fraud problem, but we wanted to reduce the friction for our good customers, improve the experience for users – we wanted to trust our clients, and we wanted to be confident in our visitors.”

Solution

Knowing that he wanted a machine learning-based solution was just the first step for Gustavo. After much research and deliberation, Destinia came across Sift Science and found that its products came highly recommended among other OTAs and businesses in the travel industry. Even better, Sift Science allowed Gustavo to try the product for free and test integration requirements in order to measure efficiency early and often.

Destinia committed 1 developer and 1 analyst to integrating Sift Science, and they found the APIs to be clearly documented and very user friendly. Once the solution was in place, Gustavo was pleasantly surprised with the speed and accuracy of the Sift Scores and data visualizations. This ratio of high accuracy to low time was essential in proving Sift Science’s value early on; a rules-based system would have required extensive analysis and continual creation of new rules in order to compete. Gustavo and his team can now visualize everything from the unified and intuitive Sift Console.

“Traditional rules-based solutions require that you are always reactive, instead of proactive. With machine learning, we’re more proactive and we’re staying ahead of the fraudsters.”

Results

Sift Science has delivered many benefits to Destinia. In terms of Gustavo’s fraud management workflow, he credits Sift Science with the sharp decrease in time needed for manual review of transactions, ability to have more reliable automated decisions, and fewer blind spots due to not needing to rely on rules. In terms of helping his team, Sift Science provides more intuitive and useful analytics and tools, plus the ability to see connections between potentially fraudulent users. However, the most important benefit of using Sift Science has been the ability to reduce friction for Destinia’s good customers.

Empowered by the efficiency of Sift Scores, Gustavo and his team are now working to streamline their workflows with Sift Science and getting ahead of fraud rings. The data visualizations allow Gustavo to more quickly spot trends and connected suspicious users, as well as pinpointing fraudulent signals to identify bad users before they check out. By leveraging the data and learnings from Sift Science, Destinia is gaining a better understanding of their users. Armed with this knowledge, Gustavo and his team can continue to slash Destinia’s chargeback rate, minimize friction for legitimate customers, and increase sales.

“The data visualizations allow us to detect more fraud patterns and crime rings. Overall, we have a better sense of control.”