Model Behavior: Banks See AI As A Customer Experience Tool

Given the roller coaster ride consumer finances have been on for the last 10 months, managing risk has become critical for financial institutions (FIs), both in terms of rising fraud counts and in terms of rising consumer delinquencies.

Mastercard’s Vice President, Global Head of Product for Artificial Intelligence (AI) Express and Credit Risk Amyn Dhala told Karen Webster in a discussion that technology can make that real-time risk management attainable. It can help banks reduce tens of millions of dollars in losses, which will get the attention of every financial services company on the planet.

But AI, he said, can provide a lot more than that in terms of protecting FIs from risk. It can also make it possible to preserve a delightful consumer experience while layering in more efficient protections.

“Imagine you take your grocery cart, and you go to the cashier, and your transaction gets declined at the grocery store,” Dhala said. “That’s probably the worst experience in terms of false positives today: the customer experiencing that embarrassment. But using a combination of modeling technology, additional data elements permissible by country regulations, and a continuum of intelligence, we are able to develop highly accurate models, which are able to predict the credit risk of customers in this environment.”

Driving Actionable Intelligence In Real Time

The power of AI, he said, is to take data from a variety of different streams and collate it into a single, personalized and actionable model for each customer and their standard behaviors. Using a combination of smart agent technology, customer intelligence and other data sources, Mastercard develops highly accurate models. And these models aren’t one-off events. The AI algorithm constantly updates its data. A consumer buys something online, and even that data is fed into the model and further influences its risk calculations.

“Banks can actually benefit from looking at how a customer’s behavior has been over a longer period of time, and then act accordingly rather than just at a single point of time,” Dhala said, explaining that over the last year, “single point in time” data has become more questionable.

It was particularly hard to get an accurate shot of a consumer’s credit risk during 2020, he said, because there were so many interacting factors in play, such as stimulus checks. Some consumers paid down debt and stayed up to date on bills that they otherwise wouldn’t have been able to. All good to be sure, Dhala and Webster agreed, but it presents a genuine challenge in evaluating credit risk. It can be hard to see if the consumer actually has funds of their own, or if they are leaning hard on government infusions of cash that may soon be running out.

The power of AI, he said, is that it can see all of that inconsistency over time and present a highly accurate credit risk model simply because it can account for the entire data set that surrounds each consumer.

Focusing On The Consumer And Building The AI

The consumer experience, particularly in the age of remote interaction, Dhala said, has taken on a new significance in a post-pandemic world. Businesses are beginning to see the ways in which AI can be useful to developing those new communications channels.

“I think increasingly, FIs realize that every opportunity to interact with a customer is an opportunity to engage the customer, to provide a better experience so that they can actually get a greater share of the customer’s financial activity, especially given the digital environment which we are in today,” he said.

What AI allows, he said, is for the kind of real-time decision making that is accurate, that doesn’t result in the false positives that cause consumers so much embarrassment at the point of sale (POS). And its use goes even further, helping a consumer mitigate a financial problem before it actually starts up.

For example, he said, AI modeling can often spot who is likely to become delinquent 70 to 80 days before they actually become delinquent. That insight provides the opportunity to start working with digital customers over a couple of statement cycles and trying out myriad treatments so they can provide the best possible customer experience. The AI might say the best intervention is no intervention because the consumer is already on the road to self-correcting the issue. Or it might be as simple as sending reminder payment push notifications.

“They could look at the continuum of the timeliness, giving them the flexibility to try out a range of treatments while ensuring a better customer experience and improving profitability as well,” Dhala said, noting that this is the true power of AI done correctly.

The FI reduces risk and loss to delinquencies without having to create painful experiences for consumers that involve false positives or harassing collections processes for consumers who do not need or want them.

And ultimately, he said, the tools are out there for any FI to use. There is no magic secret to setting up AI, just a simple process. FIs need to develop the roadmap for the issue they are looking to take on, test, find a partner to help them if necessary, deploy, and then start again with the next objective. It’s a pattern, he said, that more and more banks worldwide are adopting — because the last year has proven that these are the tools entities not only want but need.

“Given the severity of the circumstance which you’re facing today and the volatility in terms of significantly high delinquencies,” Dhala said, “it really requires a critical look in terms of how you manage credit risk and how you can improve customer experience.”