When invoice images are captured, they are just a picture of the invoice, dots interpreted by the human brain as data and characters. This data can be manually entered into the system or there is an opportunity to make huge efficiency savings by utilising invoice recognition software. RPA OCR (Robotic Process Automation and Optical Character Recognition).
A little known fact is that full-page OCR extracts not only the characters on every page of the digital image but also their positions. Invoice recognition software uses the content and positional information, along with some clever artificial intelligence to locate and extract the relevant invoice data, such as invoice number, PO number, tax date, net amount, tax amounts, total amount and a list of the line items on the invoice. The latest software uses keyword searches and the position of the data to work out what the data extracted is, all without the use of a fixed positional template.
RPA takes over
The Workflow process (RPA) now takes over to match and confirm that the data extracted is validated against all the data you hold on file, whether that is Supplier data, Purchase Order data or third party data sources. This is why it is crucial that your Master Data is correct. Of Course anything that fails your business rules can also be flagged now for human checking.
It is important to note that no system exists that will extract invoice data 100% therefore some extracted information may need to be checked by a human, however, the goal is to save 80-90% of costs associated with reading, and typing in of invoice data, although lesser percentages will still make the investment worthwhile.
The next step of RPA is approvals and this will be covered in a future blog.