Can the business model that’s kept the wheels of retail payments and commerce moving for the last six decades turn the whole notion of SMB trade finance on its head?
Formerly a managing director at Goldman Sachs, Paul Christensen, now co-founder of Previse, told Karen Webster that he’s convinced it will.
That retail payments credit card model – and how it works – is well-understood.
Buyer walks into a store to purchase something for $100 and uses a credit card to pay for it. Store gets $98 (more or less depending on their deal) and the buyer gets what she went in to buy, and probably some sort of reward back to the card she used to buy it – cash back, points, whatever. The remaining $2 goes to the bank that issued the card the buyer just used; they use some of it to pay the card network and acquirer.
The seller gets access to that money in a couple of days, maybe even quicker with new push payments and faster payments initiatives designed to accelerate the speed of settled sales to bank accounts.
The pushback, though, that one often hears is that the seller would pocket all $100 – and have it available immediately to use – if the buyer used cash instead. But increasingly, buyers don’t always want that, and for the seller, handling cash isn’t free. Accepting a credit card payment offers a convenient way for sellers to make a sale, and for buyers to pay them.
That’s far from the scenario that exists today when an SMB supplier is selling to a large corporate buyer.
There is no cash, and if payment via a virtual card is available and offered to a supplier, it’s often a costly option that does little to accelerate payment. Invoices still need to wind their way through the enterprise procurement and accounts payable departments, often taking 30, 60, 90 and sometimes as many as 120 days for the supplier to get paid.
At that point, taking a 2 or 3 percent haircut for a payment via a virtual card is acceptable, because suppliers don’t have any choice but to be paid at a discount months after an invoice has been sent.
This confluence of events is why late payments to SMB suppliers have been dubbed the silent killer of small business. Smoothing over the lumpy cash flow often requires tapping into working lines of capital or short-term loans from online lenders, both costly options and neither guaranteed to be available.
It’s why Christensen believes that exporting the credit card model to the accounts payable department at large enterprises will turn that two to three month (or longer) wait into a cash-on-delivery payment for the SMBs they plan to pay anyway.
When Payments Terms Are Instant
It’s a belief that Christensen has built a business around – Previse. The model looks very much like the retail credit card model, but is much cheaper. There is a buyer – the enterprise corporate. There is a seller – the supplier providing services to the enterprise corporate. There is a network that authenticates the transaction, sets the fees, and moves the invoice to a bank to authorize the transaction and make the payment to the seller, all instantly. The buyer then repays the bank once it has completed its normal invoice processing.
It works like this.
Bob the builder, the supplier, sends an invoice to Global Inc. for $100.
When the invoice hits Global’s ERP system, Previse the network sends it through its scoring model, which uses historic data from the buyer and seller’s relationship as well as other, similar types of payments that Global regularly makes to similar types of suppliers.
Previse then scores the likelihood that Global will approve and settle that invoice. If the score meets the threshold – say, 98 percent – that has been set by the bank, then the bank pushes the $98 to Bob the builder immediately, since the risk assessment model says that Global is more than 98 percent likely to approve it once it moves through the system two to three months hence.
The 1 percent fee – in this case, $1 – is paid by Bob in exchange for getting money deposited into his bank account instantly. Bob can also opt to get the full $100 and have his invoice go through the Global system as usual. Previse splits the $2 – in this example, between Global as the buyer and the bank paying the supplier.
When Bob the builder’s invoice hits Global’s AP department, they approve it and then do the account reconciliation.
The end result, Christensen summarized, is that Bob gets the equivalent of cash on delivery of services, without Global Inc. having to do anything differently, other than allocate a pool of money from which they’d like to make these instant payments.
ERP systems and internal AP processes run as usual. The big difference is that funds have been paid to the SMB supplier immediately upon receipt of the invoice.
Risk? What Risk?
To power this whole process, Previse applies machine learning to payables data from a large corporate entity. The algorithm looks at historical patterns, such as how often invoices of this type or from this supplier have caused problems or resulted in credit notes versus just being paid. It also looks at invoice features such as frequency, currency and category.
Christensen said the algorithm must determine the likelihood an invoice will be paid, taking into account factors such as how reliable Bob the builder is when it comes to delivering products or services on time at the buyer-approved quality – and only if the invoice is determined to be a good risk is Bob the Builder eligible for an instant payment.
Bob must also have a track record with Global – in this example, of at least a year, or roughly six invoices – so there is service quality and payment history from which to draw. According to Christensen, since large corporates change their suppliers so infrequently, many, if not most suppliers can see the benefit of this instant payment arrangement.
But suppose Bob the builder hires new people who don’t deliver – and instead of receiving the 100 black pens for which they sent an instant payment, they get 50 red ones. Christensen said it’s hardly a risk, because it’s in Bob’s best interest to make the company whole if there is an issue with an order – that is, if they want to stay a Global supplier. In that case, the system automatically generates a credit note that can be used against the next order.
Christensen said there are some categories in which this type of system is a “no brainer” – think the types of services that vary little month to month or year to year, like records management. The risk of making an instant payment is slim to none.
Process Change Optional – And Possible
When Christensen touts zero process change for buyers and sellers, what he means is that there is no technology integration or deployment needed to take advantage of the Previse platform. But if a company working with Previse wants to optimize its procurement, AP and invoicing process, he said the company’s machine learning can help them do that.
For example, instead of having 500 employees spend five minutes apiece approving 10,000 invoices, those employees can look at the risk rankings from Previse and focus their efforts on the riskiest invoices. Christensen said this will increase their ability to find fraud and duplicates, improving efficiency and perhaps leading the company to reallocate headcount away from manual review to other value-added areas of the enterprise.
But the biggest change is to the SMB who can reduce their reliance on costly trade finance options just because their current buyer’s process matches payment with approved invoice. By de-risking an instant payment, buyers not only help further the well-being of the SMB suppliers they rely on, but get paid each time they do.
The credit card model that has unlocked trillions of dollars of value for retail SMBs seems poised for an encore on the B2B side of payments.