Harnessing AI to help business customers get paid

Many corporations still have manual steps in the way they process outstanding invoices. But Rodney Gardner of Bank of America Merrill Lynch — and other bankers — are starting to offer their corporate clients a way to automate the handling of incoming payments.

The service is proving to be valuable both for the banks and the corporations.

Gardner invites clients to give him a ledger of a month's worth of the money owed to them by their customers. He then runs the information through a system of cloud-based software that BofA calls Intelligent Receivables.

The system scours the river of data that flows to these corporations from their customers via myriad channels, from emails to online payment sites. Then it matches the data to money that comes in.

To do this, it harnesses high-powered computing developed by HighRadius, a fintech firm that has a partnership with the nation's third-largest bank by assets.

By using artificial intelligence to intuit information — the automated analysis compares customer names and dollar amounts on assorted documents — the system speeds the processing of receivables from invoicing through forecasting, relays the information to the corporation's accounting system and directs attention to such matters as shortfalls, disputes and delays.

ABM060118-b2b-payments.jpg

"I am able to figure out up to 95% of the time what this payment was for and who paid it, and send you a file, just like back in the lockbox days," said Gardner, who heads the global receivables business in BofA's investment banking unit. "I've not touched it. The client has not touched it. The algorithm gets smarter as it goes, and that's how the straight-through rate improves over time."

Think of it as machine learning meets the lockbox.

In an era of paper-based payments, the lockbox shaved days from the time it took companies to receive remittances, thanks to banks placing collection points in proximity to postal hubs.

Now as paper gives way to digitization for business-to-business payments, data science is allowing companies to collect cash they are owed faster and refining their ability to predict when they will be paid.

It is the same technology that is also picking the portfolios we invest in and suggesting the products we buy.

Though business payments by check continue to endure, they are trending lower. This year, 70% of business-to-business payments in the United States will take place electronically, a figure that is expected to climb to 80% by 2024, according to a recent report by the research firm Celent.

"We think paper's minority status is going to motivate more buyers and sellers to say this whole paper invoice and check processing thing is for the birds, it's time we get off this," said Bob Meara, a senior analyst at Celent who co-authored the report.

Dealing with the side effects of digital payments

Electronic payments represent a boon for buyers but generate a jumble of information, formats and sources for sellers.

Paper aside, the algorithms that power the latest generation of receivables platforms also promise to improve on the use of software that reads information about remittances from forms that accompany them.

If anything, mismatches between the data captured from such forms and remittances received electronically — a likelihood whenever someone varies the format or a field — have boosted the number of payments that companies must resolve manually.

"The state of the industry is dependent on this old, manual and error-prone templating technology that has a lot of problems with it," said Mike Dignen, who heads HighRadius' sales to banks.

Enter artificial intelligence, which avoids the errors by "essentially reading and interpreting the data much like a human would," he said.

The automation offers banks an opportunity to help their corporate clients improve interactions with their own customers, who increasingly expect commercial transactions to embody the convenience they experience at sites such as Amazon.

"All those trends in digital payments that consumers have experienced have drifted over to the business-to-business side," said Megan Kakani, who directs new partnerships for commercial payments at KeyCorp.

"Our clients' customers need that choice, but our clients need to be able to handle that complexity without drowning in it," she said.

The $135 billion-asset KeyCorp holds an equity stake in Billtrust, a fintech firm that automates accounts receivables. Software from Billtrust forms the backbone of the KeyTotal AR platform, which uses automation and machine learning to simplify both billing and collections.

According to Kakani, the value of teaming up with Billtrust is that it gives KeyCorp the ability to provide such services holistically to clients who already may be using its processing, billing and lockbox services.

"Wholesalers, distributors and manufacturers have the most pain because they have the widest range of customers," said Kakani. "Part of the value is being able to bring all that automation under one roof, so our clients don't have to manage multiple partners."

The convenience that Kakani alludes to may be an imperative for banks. Besides seeing a decline in revenue from lockboxes, financial institutions face the prospect of fintech firms coming between them and their corporate clients.

A new way for banks to become part of the workflow

As Meara sees it, the transition gives banks an opportunity to reassert their primacy at the center of their clients' financial workflow amid what he anticipates will be acceleration in the falloff of revenue from lockboxes. Still, lowering costs for corporate clients while boosting revenue and profitability for the bank may prove to be easier said than done.

"It sounds like a no-brainer that holds the potential to leave banks and customers better off; the problem is how most banks are organized," said Meara.

But he noted that the shift demands a willingness to forgo some of the $20 billion in revenue that banks earn annually from treasury services tied to cleaning up the data that accompanies remittances.

Teaming with fintech firms offers a way forward. Such arrangements can marry the business relationships of banks with the know-how of fintechs, which have experience selling systems that automate receivables to companies and their customers.

Fintech firms that started out by offering companies a way to cut costs and lower the fees they paid to banks have come to see that teaming up with banks makes sense, said Meara.

"Now there are a bunch of onetime rivals who are very happy to partner with banks and to provide a solution along with robust sales and service support," he added. "You just have to want to do it. The heavy lifting has already been done."

That seems to be the experience of Royal Bank of Canada, which has partnered with VersaPay, a Toronto-based company, to automate receivables for the bank's business customers with the goal of accelerating the flow of payments.

The joining of Canada's biggest bank by assets with a fintech firm in its backyard comes amid a move by Canada to modernize its payments ecosystem with a series of initiatives that, among other things, aim to promote efficiencies for businesses — such as straight-through processing. (Straight-through processing refers to transactions passing through the system without the need for manual intervention.)

"We have talked to corporate customers, from small businesses to multinationals, across the country," said Stephen Miller, RBC's senior director for working capital management solutions. "We see a light turn on around the opportunity this has to transform their accounts receivables capability."

For reprint and licensing requests for this article, click here.
Electronic invoicing Payment processing Digital payments Artificial intelligence Machine learning Bank of America Merrill Lynch KeyCorp
MORE FROM AMERICAN BANKER