PSD2 and You: Why Banks Are Turning to Artificial Intelligence

by Richard Harris, Head of International Operations

You know how the Chinese word for “crisis” is the same for “opportunity?” The bad news is that’s actually a myth. The good news is the metaphor still applies to PSD2. Whether banks like it or not, the second Payment Services Directive is coming in 2018. The European Commission wants to make way for emerging payment service providers, promote innovation in payments, reduce transaction costs, and introduce two new words into the payments lexicon: integration and aggregation.

After PSD2, customers will be able to link bank accounts directly with marketplaces, with new third parties acting as Payment Initiation Service Providers (PISPs). And they’ll be able to aggregate their accounts via APIs for a single view of all their accounts, with new third parties acting as Account Information Service Providers (AISPs).

Your world is changing

If you’re a bank, you’re probably concerned, and rightfully so. For one thing, you’re losing a valuable asset: exclusive access to customer data. For another thing, you’re being forced to open up your infrastructure to third parties. But banks can change a potential obstacle into an advantage by leveraging PSD2 to create a great customer experience. Because if you don’t use these next few months to build a promise connecting data and innovation, then another bank will.

One area where PSD2 has a magnified focus is on strong authentication. In 1880, the payment chain flowed under a wrought iron divider from a customer to a teller. Fast forward to 2018 and life after PSD2, there are many more pairs of hands between the customer and the bank, and the more layers we introduce, the greater the likelihood that this chain breaks. With more third party layers separating the customer from the bank, there’s a pressing need to verify that people, businesses, and providers are who they say they are.

From compliance to experience

As compelling the need for strong authentication, customers are increasingly resistant to it. Consider one common example of strong authentication, which is two-factor authentication, where you’re required to demonstrate your authentic identity through something you know (like a PIN), and something you have (like a credit card).

Nobody loves two-factor authentication. Customers want immediacy. They want whatever the opposite of friction is. They want to feel un-screened. And now, with PSD2 (along with GDPR), they’re enfranchised with the ownership of their transaction data in new and powerful ways.

So PSD2 is as much about customer service as it is about compliance, and banks need a unified strategy to solve for these two goals that are at complete odds with each other: user experience on one hand, risk on the other. There’s only one thing alive today that can solve for both these goals in one fell swoop, and that’s artificial intelligence. AI is augmenting our capacity for making sense of information. To solve for both compliance and experience, banks becoming PSD2-ready will need a risk engine that contains an AI brain. And not just any AI brain, but a brain that’s fast.

A focus on capabilities

Latency matters. We’ve known this for a while now. A decade ago, Googlers discovered that half a second in latency reduced traffic by 20%. At the same time, Amazon engineers found that after delaying pages in increments of 100 milliseconds, revenue drops were “substantial and costly.”

A decade later, customers have become even more demanding of immediacy. And post-PSD2, with multiple third parties entering into and out of the transaction workflow, banks will need risk engines that can accommodate more layers without creating more latency. Only a machine learning platform that can process ultra high volumes of data at ultra low latencies will do.

In addition to speed, a risk engine needs to have an AI brain that will ingest high volumes of varied data from different sources. Third parties will be seeding an already crowded payment ecosystem with new and siloed information inputs. Imagine a risk engine that can ingest all this diverse omnidata from all these new omnichannels to produce a single story of every transaction. AI can give you a complete view into every customer, so you can tell bad actors from good actors, and focus on the good. But a machine learning platform is only as powerful as the data it can understand. In order to give you accurate insight into every transaction in sub-100ms latencies, a risk engine needs to be able to instantly take in every input.

The banks who come out on top will be the ones that focus on PSD2 as an opportunity. This is your chance to woo your customer all over again.

It’s your chance to be proactive. The opening of APIs will break down siloes in the current landscape and enable more connected experiences and new fintech innovations. New marketplaces will emerge for P2P lending, digital wallets, and account aggregators, and new apps will be able to get complete views of a customer to create a service that’s devoid of any branding for the bank that’s actually behind the data.

“Innovate or die.” That’s probably too strong a phrase. But as we’ve seen with the Chinese-word-for-crisis misconception, a phrase doesn’t need to be true, to be true.

Richard Harris

Richard Harris

Head of International Operations at Feedzai
Richard Harris

Latest posts by Richard Harris (see all)