The pandemic has changed a number of industries, including online banking and e-commerce. Some eighty six percent of Americans now use online banking, and e-commerce in the U.S. This number has exploded by over 44 percent from last year.
Unfortunately, fraudsters have seized on this spike in online activity to prey on businesses and consumers. Credit card losses in the U.S. are estimated to hit about $11 billion at the end of 2020 and global losses from payment fraud has skyrocketed from $9.84 billion in 2011 to $32.39 trillion in 2020. In addition, payment fraud is expected to continue its upward trajectory and cost $40.62 billion in 2027, a twenty five percent hike from 2020.
It’s no wonder that fraud protection is top of mind for businesses today.
The Business Perspective
Fraud is expensive and its costs often reach beyond what many organizations even consider. For example, if you have too many chargebacks, credit card companies may raise your interchange fees. They might even yank your ability to accept credit cards if they decide you have too many chargebacks or other fraudulent activities.
Then, there’s the fraud investigation costs, reconciliation expenses and the blow to your reputation if you have too many “false positives” that wrongly accuse innocent customers or “false negatives” that allow fraudulent use to continue.
These are all solid reasons to be tackling fraud head on. But there is another, more important reason to address it: It can help you grow your business.
Specifically, if you can better understand your customers fraud risk profiles, you will be able to personalize customer offerings tailored specifically your customers individual needs and win a greater share of the market. This will enable you meet your strategic objectives such as meeting your revenue, profitability and growth targets, boost customer satisfaction and reduce customer churn.
Still, the biggest boost can come from setting your business apart from competitors. By using fraud risk analytics to your advantage, you can engage customers in every interaction in their customer experience.
Let me illustrate this with a real-world use case example in the credit card industry. In July 1994, Signet Bank, a small bank in Richmond, Va. , decided to spin off its credit card division as an independent company. This new credit card company begin life with approximately four million customers and $6.6 billion in credit card loans outstanding, very small by industry standards
They had goals of growing into a larger company, and their CEO decided to focus intensely on using data analytics to better understand customers so they could tailor personalized, attractive offers to their customers.
For example, if they had a cardholder who was deemed “high quality” and someone they really wanted as a customer, they would craft an offer that included a lower fee, a lower interest rate and a possibly additional benefits through their loyalty program or co-branded partners. Their objective was two-fold: offer a card so attractive that the customer would sign up for it and also use it regularly over their other payment options.
On the other hand, if they were contacted by a less desirable customer that they didn’t necessarily value as highly, they would craft an offer that would entice them to sign up, but would have higher fees and interest rates to ensure that the customer would remain profitable despite the perceived risks.
Their intense focus on data analytics-based marketing worked tremendously, and their credit card portfolio today boasts over 62 million credit card customers and $107 billion in credit card loans outstanding.
You may know this company by the new name they announced in October 1994: Capital One!
Today, Capital One remains intensely focused on data, and claims they perform more than 80,000 analytical experiments per year to test potential offerings.
When chip credit cards made it very difficult for criminals to use them fraudulently, the bad actors turned to easier targets such as identity fraud and card-not-present (CNP) transactions, usually found in either online or phone purchases. In 2018, $24.26 billion was lost in payment credit card fraud worldwide. The U.S – the most fraud-prone country – experienced 38.6 percent of those losses.
The growing sophistication of criminals has also led companies to develop more sophisticated fraud solutions. While the earliest systems to detect fraud were rule-based, more sophisticated solutions have been developed that use artificial intelligence (AI) and machine learning (ML), which have proved to be more accurate and lead to better results. Typically, AI/ML-based systems use about 10 to 100 attributes to look at over large data sets. And the most sophisticated companies are even moving beyond AI/ML to develop fraud solutions based upon neural network and deep learning.
For example, PayPal has begun using neural nets and deep learning to identify and combat fraud. These systems can analyze between 1 million to 10 million attributes, which must be processed very quickly (<30 ms) in order to provide a real-time purchase experience. Neural nets/deep learning solutions have shown themselves to be up to 30 percent more effective than traditional AI/ML-based systems.
If that weren’t enough, there’s another, new area for improving fraud and analytical systems called ‘Explainable AI’. This is sort of an expert system – on top of an expert system – that helps explain how the AI/ML system is actually working and making decisions. This is useful because the better you understand how the ‘black box’ works, they better you can make more meaningful tweaks to improve the system, and you will also be better able to explain to customers how their data is being used. These ‘explanations’ are also very important for customers to comply with data privacy regulations such as GDPR and CCPA, which require that you be able to tell customers how you use their data and how you are reaching decisions such as why you may have declined them for a credit card.
It’s critical to remember that no matter where you are in your journey – whether you’re currently using a rule-based system or moving into AI/ML-based solutions – you will require a technology solution that provides the ability for best-in-class performance, e.g., the ability to process data extremely quickly (<20ms) across extremely large data sets (Tbytes, PBytes), and be able to do this with a reduced TCO. This is vital to ensure your business can grow today but also provides a pathway to the future.
The last year has seen an explosion of growth in the ecommerce and online banking arena, but online fraud is also proliferating in tandem. One of the most effective ways that businesses can adapt to this new reality is to utilize sophisticated fraud analytics running on a modern technology platform to provide top performance. This gives businesses a way to not only thwart fraudsters, but better understand the customer, provide more targeted services, improve revenue and stand out from the competition.