Purchase to pay / Invoice recognition - brief / Invoice recognition

Invoice recognition

Manually entering invoices into an invoice processing system has given way to scanning and recognizing digital invoices with OCR software. With e-invoicing, invoice recognition is completely taken out of your hands.

Scanning and recognition

In the Netherlands today, about eighty percent of all invoices are still sent as PDF invoices. Previously, the information on incoming invoices had to be manually entered into a financial system. With the advent of Optical Character Recognition (OCR) software, the information on a pdf invoice can be automatically converted to data.

The information on an image is thus recognized by OCR software. Paper invoices, of course, must first be scanned before the scan can be transferred to data using OCR software. For so-called machine-readable PDFs, the invoice information is immediately available as data.

OCR requires the receiving party to perform an additional operation. While this operation is faster than manually transcribing into a system, it is still an operation that requires labor and can lead to additional errors. E-invoicing offers an alternative.

E-invoicing

E-invoicing stands for sending and receiving invoices in the form of an e-invoice. An e-invoice is typically an XML data file with a data structure standardized for invoices. The advantage of e-invoicing is that it brings the invoice directly and automatically into an invoice processing application.

Adoption of e-invoicing is slow, mainly due to the lack of a data structure standard. Connection to an e-invoicing network offers an alternative, as conversion is taken care of by a third party. Suppliers can send any data format, as well as pdf invoices, to a network address, after which the invoices are converted to a standard that works for the receiving party.

Read all about e-invoicing.

Billing dates

Converting information on an image to data can be limited to header-level information or complete by converting line-level information as well. Today, automatic invoice matching has become commonplace. At least, matching header-level invoices with underlying purchase orders and receipts has become standard functionality. Both ERP systems and best-of-breed p2p systems offer this by default.

Head level recognition

But matching at the header level alone has not proved to be a watertight method in practice either.

This makes automation parties continuously look for ways to make matching more intelligent, such as matching invoices at line level. In other words, the software compares invoice lines with order and/or receipt lines, using certain tolerances, rematch periods, validation rules and matching scenarios. And it certainly doesn't stop there.

In practice, when the information is extracted only at the header level, this means an average of 15 fields must be recognized and validated. But if the information is also extracted at line level, then the challenge is much greater.

Matching becomes difficult when the customer and supplier have different practices. For example, when the customer orders item A, but the supplier lists its own item code X on the invoice. Or when different payment terms are used, as the invoice above shows. 

Line level invoice recognition

Automation players are constantly looking for ways to make invoice matching more intelligent. The ability to automatically match invoices at the header level has been around for a while. Matching invoices at line level (i.e., at the level of line items) is more recent. This enables partial matching of invoices and allows follow-up actions to be automated.

The advantages of sourcing invoice information at the line level are obvious. When the information on invoices on the one hand and the information on orders and receipts on the other are compared at the detail level, it is possible to identify with certainty where the discrepancies are.

With partial matching, only the discrepant part of the invoice is submitted for further manual processing. Therefore, the accounts payable department does not have to solve a puzzle to match the discrepancies.

This increases labor productivity and reduces the risk of errors because employees only have to focus on the mismatched lines. Also, any price differences can be assigned to the appropriate (order) line rather than the entire order. This increases the reliability of information in the various systems.

Problems also arise when different units are used on the invoice than on the order. Thus, if 100 units of item B are ordered, but the supplier delivers 1 box with item code Y - which may contain those 100 units of item B - a discrepancy will be noted unnecessarily. This is a common problem. The invoice in the image above shows something similar, but in numbers. The number of pieces on the invoice does not match the number of pieces ordered. If the system is not set up in advance here, this will be a stumbling block for automatic processing.

Invoice recognition software

• Basware Network is the world's largest open network for any-to-any conversion of invoices and orders. Over a million suppliers and buyers send and receive invoices and orders in the exact format they want.
• Learn more about invoice processing.
• With Kofax, information from scanned paper invoices and pdf invoices is read and this information automatically enters the appropriate fields. These include invoice date, invoice number, IBAN number, VAT number and net and gross amounts.