Urgent need for cash puts pressure on accounts receivable teams


Peter Lee
Published on:

Machine learning powers new software that can help companies chase down overdue invoices and failed payments without driving customers away.

head_clocks_time-340.jpgThe head of commercial banking at a large European lender gives Euromoney an unusual example of how it has been helping companies through the lockdown – not just extending overdrafts, granting repayment holidays or cutting account charges.

“We are a big company ourselves with lots of suppliers. We are adopting payment upon presentation of invoices. We need to keep these suppliers alive.”

Are the bank’s corporate customers doing the same? “This is bringing out some of the best behaviour and some of the worst that I have seen,” says the banker. “A number of big companies are accelerating payments to suppliers, but some tell us they are extending payment as long as they can to conserve their own cash, even if doing so hurts smaller companies.”

For suppliers, chasing overdue accounts receivable is now a fraught business, even for essential – but sometimes taken-for-granted – items such as software, IT and security services. 

We are astounded by how much pain there still is in the world of payments 
 - Duncan Barrigan, GoCardless

Wholesalers and distributors today operate on shrunken volumes and thin margins. They need to pay for goods, but when they sell them on buyers may partially defer payment.

Manufacturers that need to buy raw materials can face an even worse negative cash-flow trap if purchasers of finished goods buy on credit, leaving them even more dependent on working capital from banks.

Companies need to chase late-payers and reduce days sales outstanding (DSO), the key metric in accounts receivable, without risking driving away key customers.


Anthony Venus, co-founder and chief executive of YayPay, an accounts receivable platform that uses machine learning to predict likely late payers and the best ways to chase overdue payments, tells Euromoney: “This has all been such a shock that payments certainly slowed down, even from accounts that vendors could previously take for granted.

“It is urgent to get money in the door. What we are finding right now is that how you collect payment can have a big impact on customer satisfaction. Cash is certainly king. And companies operating on thin margins are concerned to improve collections and reduce bad debts.”


Anthony Venus,

He adds: “But accounts receivable is not just about cash management. It is also about managing important and sometimes delicate customer relationships. Do that well and you might put yourself at the front of the queue to get paid.”

YayPay starts with an automated system for handling the back and forth of email communications between vendors and their customers. It uses years of data on millions of buyers to identify behaviours that may indicate a likely late payer.

It also reads the language of their incoming emails – which may include complaints of non-delivery – and suggests optimum responses and tactics to reduce DSO.

Venus says: “Our system can suggest the appropriate tone of message to send to a late payer, the right time to send an initial message to motivate prompt settlement and the most productive frequency for follow-up messages.

“And it spits those messages out automatically – so that AR teams don’t have to compose them manually – and manages all this at scale.”

YayPay claims a 94% accuracy rate in predicting customer behaviour on payments – who will likely pay on schedule, who will probably be late but pay once a certain number of days overdue and who to really be wary of. It also claims an average 30% reduction for customers in DSO and big improvements in efficiency.

That allows companies to do a better job of internal communications.

Venus says: “You have the classic situation where once an invoice is more than 90 days past due and classed as delinquent, finance asks the sales team to help chase it and threatens to withhold commission. The sales person comes back and says: ‘Why didn’t you tell me this customer wasn’t paying? I have been selling more to them all this time.’

“Greater efficiency in accounts receivable can head this off and help the whole company engage better with customers.”

It sounds like the kind of service banks themselves might provide. Fifth Third Bank is a big investor in YayPay along with venture capital backers.

“One or two banks are beginning to distribute our software and integrate it into their payments offerings,” says Venus.


GoCardless takes a similar approach to recurring payments. It sets up a variant of automated direct debit schemes integrated into the accounting software of customers such as newspapers to automate subscription payments, gyms to take membership fees, accountants to take invoice payments and lenders to take repayment instalments.

It handles $15 billion of payments a year for more than 50,000 companies across 30 countries.


Duncan Barrigan,

“We are astounded by how much pain there still is in the world of payments,” Duncan Barrigan, chief product officer at GoCardless, tells Euromoney.

“E-commerce has transformed how generations of merchants accept customer payments online. But that same level of innovation has not applied to recurring payments for which card networks are not such a great use case.”

He adds: “The best way for merchants to handle recurring payments is not to wait for customers to decide to pay them, but rather to take agreed payments from their bank accounts on set dates. We provide a single global network to handle direct debits which has a 97.5% success rate.”

But failed payments can still be painful, especially today.

The company has been testing a new product, Success+, with 2,000 businesses to improve handling of failed payments, which can be time consuming for businesses to chase.

Aggressive handling of failed payments can increase customer churn.

“There is something intensely personal about a business that has provided you with a good service suddenly ringing you up to chase down a failed payment,” says Barrigan. “That can damage a previously good relationship.

“Often, regular customers are willing to pay but may not be able to. We are using machine learning to identify the best approach to retrying failed payments, for example retrying when it is most likely customers will have money in their accounts.”

He adds: “So, the next phase of the product development would be to identify when retries are unlikely to succeed at all. That may be the moment to stop retrying and offer certain customers the chance to pay by instalment instead.”

Of course, businesses don’t want to offer that instalment option automatically on every failed payment because, especially right now, every customer would likely seize it.