creditcontrol.co.uk
​
​
UBS wins FICO Decisions Award
UBS Card Center, which processes roughly 25% of the credit cards in Switzerland, has cut the amount of fraud write-offs per compromised card, stopped more fraudulent transactions and reduced false-positive rates using the FICO® Falcon® Platform (www.fico.com/en/products/fico-falcon-platform). The card processor stopped 74% more fraudulent transactions in 2016 than in 2015, and saved potential losses of CHF172 million (about $172 million). UBS Card Center achieved these gains by leveraging the latest artificial intelligence and machine learning capabilities of the Falcon Platform, including the use of FICO® Fraud Predictor with Merchant Profiles and adaptive analytics.
“In times of constant cost optimization, we need to keep fraud write-offs to the very minimum,” said Marcel Drescher, director of fraud services at UBS Card Center. “We were facing new business volumes, but did not have the ability to increase staff numbers to absorb the new volumes. Proactivity in combatting fraud, paired with very high standards on customer experience and satisfaction, are major challenges we're facing on a day-to-day basis.” To solve this challenge, fraud experts at UBS Card Center used the free-form rule writing within FICO Falcon Fraud Manager to create complex rules that deployed the AI-based FICO® Falcon® Fraud Score and Falcon’s innovative adaptive analytics, which use the results of fraud investigations to automatically fine-tune the neural network models in Falcon. “These custom rules, combined with the advanced analytics, was the only way to improve false-positives and account detection rates,” Drescher said.
UBS Card Center also took a new approach – for the first time, it began to directly decline transactions that appeared to be fraudulent, rather than creating a Falcon Fraud Manager alert to be reviewed with the customer. “This saved manpower – only with further fraudulent activity on a card is an alert created and customer contact established,” Drescher said. “Using this approach, we have been able to manage a 42% increase in fraud cases without the need to bring in new staff resources.”