OCR (optical character recognition) has gone mainstream in the world of document management. Accuracy has increased to near 90%, making it as accurate as manually-entered data (and often even more accurate). The productivity benefits are clear and the business case for the investment is a no-brainer, both in terms of hard cost savings (reduced labor costs) as well as all the intangibles associated with reduced errors, decreased cycle times, and happier employees and suppliers.

But are all OCR solutions created equally? Actually, no. It’s not just a case of scanning an invoice and all the data miraculously fills in the correct fields in your invoice automation system, ready for approval or further processing workflow (and ultimately) payment. A system may be able to identify and read a date on a form, but if it can’t tell if it’s the shipping date, the invoice date or the payment due date, it could lead to considerable challenges and delays in approval.

Getting this right involves highly sophisticated image processing, data extraction, and mapping capabilities, as well as ensuring that the technology vendor’s implementation team can configure a solution that works for your specific requirements.

Related: Easy, Accurate, Automated Invoice Data Extraction

Here are important things to consider when analysing OCR options to support your invoice automation process:

Where it fits

  • Deploying OCR technology should be an easy-to-use, seamless part of a broader AP automation tool, and not a tool in and of itself.  Implementing OCR into your invoice approval workflow should not require your finance team to be experts in image processing, document queue management, or even the details of OCR algorithms. Your vendor’s implementation team should set up your workflow based on your organisation’s rules, so you can just sit back and watch the results!

Related: Five Reasons Why Web Apps are Essential for Speedy Invoice Approvals

How it works

  • Know what data your OCR solution actually extracts, and how this supports and fits into your organisation’s invoice approval process. Not all invoice images are created the same, and extracting line item data from a 300-page phone bill is as useless as not extracting line item data from a PO invoice.
  • In order to future-proof your investment, the OCR system’s data extraction process should not have to rely on “template” driven algorithms. The solution should read all of the data on the invoice image, and intelligently decipher and correctly map this to the appropriate fields within your enterprise. Vendor invoice formats commonly change, and an inflexible solution could mean that your (lack of) quality is tied to an out-of-date format.
  • To ensure accuracy, the results provided by the OCR solution should be validated against your data. This doesn’t mean all fields, of course, but critical ones such as vendors, vendor addresses, people names, and purchase orders.  The data on the image should not just be what you end up with – it should be compared and validated against this data in your systems.
  • Your OCR data extraction solution should be available to a variety of invoice submission points.  For example, the ability to attach one or multiple images to an email sent into your AP automation application, uploading of various file types, and even the ability to receive and scan physical mail should all be supported options.

Related: Invoice Solutions that Finance Teams and Approvers Actually Love

Choosing a partner

  • To get the best results from your investment, OCR quality levels (the percentage of correct versus incorrect entries) should be understood. What quality level is being quoted by the vendor’s sales team? Is it the ability to correctly identify and extract certain fields on an invoice record, or the ability to correctly identify and extract all the fields on the invoice record? The latter is a more accurate way to measure the true quality of the tool, as it reflects your organisation’s real-life requirements. Ensure that you get clarification on this matter.
  • Understand how your own data quality plays a role in your OCR quality. Anyone who says your data quality is not important does not understand how OCR really works – make sure your OCR provider can explain to you what data is validated, and help you identify the major areas where clean-up would be beneficial.
  • Finally, while OCR is a highly effective solution for importing paper-based invoices into the approval workflow, it still isn’t as efficient as fully-electronic solutions where actual data files are transmitted between your suppliers and your AP processes. You should make sure that your OCR-enabled AP automation solution can support more than just OCR. While it continues to improve what it can do, it’s still no quality substitute for true electronic invoice data. Both approaches should be able to co-exist within the AP automation solution.

Related: Why the Manual Invoice Processing Model Is Broken (And How to Fix It)

Deploying OCR as part of your invoice automation solution can deliver huge efficiency and cost control benefits. However, make sure that you ask the right questions to ensure you get a solution that works for YOUR enterprise.  

Subscribe

Our choice of Chrome River EXPENSE was made in part due to the very user-friendly interface, easy configurability, and the clear commitment to impactful customer service – all aspects in which Chrome River was the clear winner. While Chrome River is not as large as some of the other vendors we considered, we found that to be a benefit and our due diligence showed that it could support us as well as any large players in the space, along with a personalized level of customer care. Sally Abella, Director of Corporate Travel Harman International
We are excited to be able to enforce much more stringent compliance to our expense guidelines and significantly enhance our expense reporting and analytics. By automating these processes, we will be able to free up AP time formerly spent on manual administrative tasks, and enhance the role by being much more strategic. Ben Zastrow Zelle