Why e-invoicing needs machine learning to speed up invoice financing
In the past, most procurement organizations admitted that they did a generally poor job of relating the purchasing processes to the actual receipt of invoices, the invoice approval process, and subsequent payment to suppliers.
There have been many reasons for this, not least of which was the limited coordination between âpurchasingâ systems, including vendors, people and processes, and the accounts payable function.
But in recent years, companies have moved to the cloud for exchanging documents and data around their source-to-pay processes, due to factors such as the rise of platform-based technologies that promote efficiency and effectiveness in the areas of procurement and accounts payable, as well as government tax regulations in Europe and Latin America.
Business case for electronic invoicing
The business case for electronic invoicing has never been better. But there is still paper and PDFs and a lot. Companies receive invoices from their supplier ecosystems through different methods, and most invoices are not electronic to electronic (i.e. EDI, XML, PO Flip). Many invoices still arrive in PDF and paper format and require some form of automatic recognition.
A typical business will have hundreds, if not thousands of suppliers, and will have multiple sources of supplier types, each with their own nuances:
- Direct commodity or material suppliers – a large portion of the expense, and often suppliers are connected via EDI, XML, or some other electronic link).
- Freight / Logistics – typically one of the largest expense categories, and typically handled outside of accounts payable in global logistics or the supply chain.
- Indirect categories such as telecommunications, IT, legal, etc. which are important expenditure categories and generally have national contracts.
- Thousands of small indirect suppliers providing goods and services around the world, who use most of the paper and PDFs.
- Spending without purchase orders – small vendors, one-time expenses, P card expenses, etc.
Now, new technology allows businesses to capture data directly from paper and PDF invoices, even unstructured ones. The ability to extract data from unstructured invoice formats and use the data instantly for compliance, taxes, accounts payable, etc. Not only is it becoming a huge labor saving for large companies, it is also a game changer for invoice financing.
How machine learning is a game-changer for prepayment financing
When you use supply chain finance and integrate with a platform like Ariba or Taulia, invoices are automatically sent and scheduled for payment through the API. But suppliers’ non-platform invoices – which represent the vast majority of their invoices – are not visible through a platform.
Off-platform financing means exactly what it means: financing invoices that aren’t on a cloud-based provider portal. Off-platform lending allows vendors to access financing by providing invoices to a third-party lender in any format; and through data extraction, the machine can automatically read all data with 100% accuracy.
Now, with machine learning, scanned document delivery and automatic retrieval provide a way to make instant credit decisions for off-platform financing. This is true invoice financing as opposed to supply chain financing pending a buyer’s scheduled payment file.
Many P2P providers use optical character recognition (OCR) and smart models to analyze invoices, but this is not 100% reliable. Although there have been significant advances with systems capable of producing a high degree of recognition accuracy for a variety of digital image file format inputs, the accuracy of OCR is 80% or 85 % at best when processing invoices. Some other vendors claim to have 100% invoice automation, but that was only after forcing a buyer’s vendors to deliver invoices in a structured invoice template. Good luck in changing behavior.
And this is where machine learning and invoices provide a great opportunity for invoice financing.
Advance financing represents an option for providers to access ad hoc capital. It is a choice they have. But when suppliers have to connect to one P2P portal to do e-invoicing for one buyer, another portal for another, and yet another to do discounts, etc., it becomes a headache. This is the problem that suppliers face when dealing with OEMs and large buyers using many different systems. Simply put: if they want access to ad hoc capital, you need to access multiple systems.
Off-platform loans go a long way in solving this problem, if not more, as they include invoices that are not part of any digital P2P platform.
Many robotic process automation (RPA) companies have plans around machine learning and invoicing, but not around finance. Machine learning models can be inserted into RPA workflows to perform machine perception tasks, such as image recognition.
It is only a matter of time before unstructured invoices are read by machines without human intervention. Then the real fun can begin.