Using Machine Learning to Re-Think the Customer Experience Paradigm
With first-hand insight into the havoc wreaked by fraud, Card Analytics and Infrastructure SVP Youssef Lahrech has a personal stake in our efforts to fight fraud with ML.
Fraud is an extremely personal and often traumatizing event
In the fall of 2008, my parents moved from Morocco to Peru. My mom had gotten a new job as a diplomat and my dad had sold his medical practice to retire and focus on humanitarian work. After decades of living in the same place, it was a big change for them to pack up and move halfway around the world to a totally new country and new culture, but they were excited and looking forward to starting a new chapter in their lives.
Shortly after landing in Lima, however, my mother’s debit card was compromised, and her bank account was cleaned out. Years of savings — my parents’ safety net — was lost in an instant. It took several days to resolve the issue and get the bank to return the funds to my mother’s account, and she describes that waiting period as one of the most uncertain and stressful times in her life. That experience, especially at such a pivotal time in her life, shattered my mother’s trust in banks.
Fraud is a booming business
According to Nielsen, credit card fraud losses this year are expected to total upwards of $30 billion, while a CompareCards survey estimated that 33 million Americans — one in seven adults — were potential credit card fraud victims last year alone.
Like innovators in nearly every industry, fraudsters are relying more and more on technology. Specifically, they’re relying on big data, APIs, distributed computing and machine learning to harvest customer card numbers and identities, which they can turn around and sell on the dark web.
How machine learning fights fraud and builds trust
Unfortunately for fraudsters, we at Capital One are also turning to technology to revolutionize how we’re able to combat fraud. We’re leveraging powerful algorithms, machine learning-driven tools, huge swaths of data, and incredible talent. With these tools at our disposal, we challenged ourselves to improve the entire cycle of fraud detection and resolution for our customers. How could we take an experience that is often a source of fear and anxiety (like the one my mother went through) and turn it into something that engenders trust and confidence in us instead?
One example of how we are using machine-learning, data, and a customer-centric approach to fight fraud is through our proactive intelligent assistant, Eno. Eno automatically alerts customers to potential fraud in real-time, and if necessary, locks their card. What makes Eno unique is that it communicates all of the above proactively and intelligently, which means when customers are alerted about a potential fraud attempt, they can respond to us on their own terms and in their own words. When customers respond to our alerts to help us understand what’s fraudulent and what’s not, they also provide us with data about how they communicate. All of this data enhances our algorithms, which translates to less fraud and an overall experience that is quicker, more fluid, and more natural for customers.
If what happened to my mother’s debit card back in 2008 were to happen to her credit card today, the advances we’ve made in applying machine learning to fraud could help her to resolve her fraud issue in less than one minute with a couple of taps on her mobile device.
We believe that if we continue to design machine learning systems for more use cases that put the customer at the center, like those that inspired the fraud enhancements to Eno, then we can promote a cooperative endeavor between humans and machines and revolutionize the broader experience for our customers.
DISCLOSURE STATEMENT: © 2019 Capital One. Opinions are those of the individual author. Unless noted otherwise in this post, Capital One is not affiliated with, nor endorsed by, any of the companies mentioned. All trademarks and other intellectual property used or displayed are property of their respective owners.