In their natural habitat, elephants rule because they’re smart, strong and fast. But they don’t dance. In today’s digital transformation, real-time payments environment, big banks are the elephants – large and formidable, but not especially known for being nimble.
But can elephants, in fact, dance? Modern business history contains numerous examples of large organizations that have successfully navigated radical changes. Here, achieving success often involves adopting technologies that let established firms connect their past with their future, in ways that catapult robust-but-inflexible legacy infrastructure into new markets and contexts.
In the world of fraud detection, radically flexible enterprise-class platforms give large institutions the agility to fight fraud in the real-time payment types they’re spinning up to compete with fintechs.
Clearing Legacy Hurdles
In speeding new products to market, banks face a number of challenges, many of which can be tied back to the legacy systems that remain at the center of the transaction universe. Upstart fintechs — the squirrels (sticking with our animal metaphor) of the financial kingdom, darting around the market to find sustenance, and sometimes even eking out an impressive share –– aren’t burdened by the inflexibility of legacy systems, or the stringent regulations banks face.
In this environment, speed to market is critical, placing banks in a triple bind between innovation, compliance and fraud. Banks must innovate to meet consumer demands, getting new products and new real-time payment offerings to market quickly on legacy technology foundations while maintaining compliance. However, fast time to market only accentuates fraud risk, because real-time payment systems are catnip to fraudsters.
High Fraud Losses at Launch
It’s well-known that fraudsters are quick to exploit new payment offerings, probing them to find their weaknesses. Banks quickly figure out fraud patterns and adjust their defenses accordingly. Real-time payments present heightened risk, though, because they are bombarded by fraudsters and it’s difficult to recognize fraud until it’s too late.
For example shortly after the launch of the real-time person-to-person (P2P) payment system Zelle in 2018, American Banker reported, “Banks have at times reported double-digit basis-point spikes in fraud after adopting such a system, and one banker recently said that soon after his firm launched Zelle, fraud skyrocketed so high he had to go into the office at midnight to shut down the system.”
Detecting Fraud Patterns Much Faster
Part of the challenge of banks introducing new payment products is that their fraud and compliance infrastructure is geared toward fixed legacy data feeds and APIs developed for other product lines. New real-time payments product lines will have new data, requiring fraud models to be completely flexible in order to ingest it. Many of these products are delivered through or with partners, necessitating the fraud models to quickly accommodate partner data and integrate it. In addition, detecting new fraud patterns for real-time payments requires rapid iteration of the fraud analytics, deploying and redeploying them quickly to take advantage of learnings.
To sum it up, banks need flexibility and agility on the systems back-end to maintain and sustain their digital transformation initiatives. In the real-time payments world, that means innovation in payment offerings, new products and new access methods.
The most advanced enterprise fraud solutions allow banks to infuse new capabilities and data into existing fraud detection infrastructure. These capabilities include open machine learning and X-dimensional profiling, enabling banks to detect fraud and financial crimes with unprecedented accuracy, across products and contexts.
In addition:
- Radically flexible data architecture allows banks to experiment with new data sources or partner-sourced information in the fight against fraud. There’s never a need to stop developing improved fraud defenses through the discovery of new signal in non-traditional data sets.
- Open machine learning allows banks to choose a mix of proven machine learning models, in-house developed models and imported third-party models. New tools are available to help fraud organizations develop, evaluate and deploy models using open libraries, their own data science and analytic assets, and proven techniques.
- X-dimensional profiling enables data from any source to be incorporated into fraud models, to create nearly limitless analytic models.
If you are at FICO World next week in New York City, there are numerous ways to learn more about what your organization can improve its dance moves with FICO® Falcon X, our radically flexible, self-learning solution to detect fraud and financial crime. Check out the Fraud and Compliance conference track, and see Falcon X in action in the FICO World Solution Center.
Follow me on Twitter @dougoclare.
