The upside of credit card fraud

When it comes to using artificial intelligence (AI) for credit-card fraud management, the benefits are more than meets the eye. Yes, AI-powered detection systems spot fraud more quickly and help contain losses. But the systems also put a new spin on fraud for credit card providers: Real-time fraud detection helps inspire customer confidence and boost loyalty.

Call it the upside of fraud. According to at least one study, organizations with a formalized program in place to monitor and prevent fraud dramatically improve their customer satisfaction and retention rates.

Defending consumers from bad actors trying to infiltrate credit card systems is  a chance for financial services providers to forge better relationships with customers, the majority of whom point to security as a top concern in online transactions, according to research by Experian. (Coming in a distant second at 18% is convenience.)

Using AI to proactively disrupt fraudulent transactions makes for an even greater opportunity to strengthen customer relationships. Brands forge stronger bonds with customers; cardholders feel comfortable about continuing to charge purchases. We’ve seen this in action: The real-time fraud analytics platform we created for a leading financial services provider not only slashed the company’s fraud losses by $60 million annually but also boosted customer spend by $480 million.

Where Anti-Fraud Controls Typically Fall Short

But for most card providers, real-time analytics aren’t the norm. Banks’ traditional anti-fraud controls come up short in several ways. For one thing, old-school anti-fraud controls are no match for the volume of data – videos, emails, images, text files – that can now be put to use to effectively detect fraud.

What’s more, the false positives that traditional systems generate are costly. Imagine the annoyance of a U.S. resident whose card is declined in Cambodia because the location doesn’t match. In one study, only one in five fraud predictions were found to be correct. In addition to being an inconvenience to cardholders, these errors can cost a bank $118 billion in lost revenue, according to the study, when customers refrain from using the declined credit card. For customers, an abundance of false positives is a like a system crying wolf: Providers lose their credibility.

AI is an important ally for banks in reducing false positives. With its superior ability to quickly detect patterns in large data volumes, AI-powered systems can go far beyond assessing a single data point like “location.” They’re better equipped to analyze a wider variety of data points generated by a traveling cardholder and distinguish these from someone using a stolen card overseas. Using its real-time fraud platform, our financial services client reduced its false positive by 450,000.

Giving Customers More Protection and More Control 

We often say the best approach for AI is an iterative one that starts small and keeps improving. This is especially apt for fraud detection because the risk of error is so high for cardholders. We recommend a graduated path of AI adoption that gives your organization a sure footing and cardholders greater protection.

1. Use supervised machine learning (ML) to train models that spot fraud faster. With supervised ML, banks use a limited amount of data to teach the system to detect fraud. Fraud analysts continuously monitor both the data that’s fed to the ML algorithms and what the algorithms do with it, ultimately training them to act autonomously and become capable of detecting fraud more quickly and accurately than human analysts can. Starting with supervised ML offers organizations the opportunity to understand the algorithms’ logic, maintain control over their decisions and continue tweaking the algorithm’s parameters to optimize outcomes.

2. Implement adaptive fraud analytics. By combining supervised ML with unsupervised ML in their fraud detection algorithms, credit-card providers can realize a big advance in outcomes, particularly in reducing false positives. With unsupervised ML, algorithms learn on their own without the guidance of a fraud analyst. Struggling for a more accurate way to detect payment fraud, a global bank we partnered with implemented a machine learning solution that uses both types of ML analytics. The system can score 5,000 transactions per second and has resulted in an 85% drop in false positives.

3. Improve personalization by consolidating a multitude of data and transactions. AI-enabled fraud detection gains traction when it brings in data from an array of channels because this additional information enables the system to better understand customers at the individual level through their behavioral and spending patterns. Leading banks already use custom-built AI fraud detection systems to enable hyper-personalization. As more standardized AI-powered systems come on the market, any size bank can begin benefiting from AI-driven hyper-personalization.

4. Make fraud less disruptive for customers. Once your AI-enabled system has become adept at identifying fraud, the next step is to use AI’s real-time capabilities to improve customer interactions. Card providers can send alerts on possible fraud events directly to customers, along with a simple way to respond, such as texting YES or NO. They can also provide easy-to-use ways of remotely turning off credit and debit cards that are lost, temporarily misplaced or simply unused. All these steps empower customers with greater control over protecting their personal information and card use.

While no credit-card provider would welcome fraudulent activities into the services they provide, it’s also true that there’s no way to completely eliminate the presence of fraud. By putting AI-enabled detection systems in place, banks can at least minimize the impact of fraud while also establishing a trusted relationship with their customers.