If you process a variety of invoice formats or, like many companies, have found that your vendors aren’t exactly excited about being forced into using your invoicing portal or joining a buying group, then an Optical Character Recognition (OCR) software platform designed specifically for AP processing sounds like the perfect solution doesn’t it? Technology that can “read” a written document and accurately populate the appropriate fields in your AP processing software with the appropriate values should make everyone’s life easier.
While we’ve used OCR processing since 2009, and it certainly does make our lives easier, even after all the improvements we’ve seen in the technology it isn’t perfect. Not even close.
Setting Reasonable Expectations for OCR Accuracy
Out of the box the best you can expect from an OCR product is an 80 – 90 percent accuracy rate. That isn’t usable for business processes, so the next phase of your project will be to teach the recognition software to fix its own errors.
You’ll need to assign one or more of your subject matter experts to use the auto-learn function to train the software how to read and organize data and how to handle exceptions. Provided that your vendors don’t send anything with handwritten values, you can expect to get about another 5 percent accuracy through this phase of your process.
Even if your remaining error rate is only 3 percent, and that is optimistic, that still leaves a lot of margin for error which can lead to over paying, under paying, paying the vendor twice, not paying the vendor at all, or paying the wrong vendor. Not only can these be expensive errors, they can also tarnish your reputation with your vendors which is something no company wants to risk.
Auto-learning has limitations too. Just because you teach the system that an invoice from Company X, for instance, needs to have the invoice number read from a certain section of the page does not mean that it will always read that invoice number accurately or that your vendor will never change invoice formats, making the previous AI learning invalid.
And even that level of accuracy assumes that what is being taught to the system is consistent and accurate. Which isn’t always the case since in most departments multiple people would be involved in training the recognition software.
There will always be a portion of your documents that can never be read by OCR because they arrive with some damage to the page or the print quality isn’t dark enough or clear enough to scan. But even for those invoices that are clear, easily scanned, and consistently formatted, it takes a great deal of human intervention to bring the accuracy rate to anything like acceptable.
What to Expect in Terms of Time, Money, and Human Resources
Of course OCR technology is a huge investment in terms of initial purchase, but you’ll also have yearly software licensing and maintenance fees, and usage fees for the number of connections you have and the volume of pages you’re running through the system. You’ll also have ongoing costs as upgrades become available.
If you choose to do all of the installation, set up, and training internally it will easily be a year or more before your OCR system is working optimally. If you engage professional services to help you get it ready to move into production you might have it up and running in four to six months.
You can set the software up on your network or you can run it from the cloud. Both will be extremely demanding for your IT department as well as time intensive for your internal experts – the people who understand what needs to be captured and how it needs to be captured. You will also need to have project managers dedicated to the process of getting the OCR system online, customized, and tested.
Even once you have it trained to maximum accuracy, OCR is never going to completely replace human judgment. Someone still has to manage and monitor the system. At the end of day there has to be a pair of human eyes doing quality assurance to make sure what was picked up off the page matches what was actually on the page.
What are the Advantages of Letting DataServ OCR Your Invoices?
You have AP experts on your staff. So do we. But we also have OCR experts on our staff so you don’t have to. We’ve been using OCR for more than eight years. We’re good at it, and we’re always getting better because we continually improve our technology and our best practices. That means you can send us the documents that will go through the OCR process with ease, but you can also send us the invoices from the local family-owned hardware store or office supply shop that won’t scan properly at all.
Plus, with our Single Stream approach to data capture, you can send us invoices in any format – whether they were received via email, EDI, or print. We’ll import the data into a single format and route it back to you with 99 percent accuracy.
When you have us managing your invoice scans it’s our technology. So you don’t have to worry about installing it, or maintaining it, or troubleshooting it, or training it, or upgrading it. You won’t have to worry about training your staff to use it, or training every person you add to your staff to use it. You get to let the technology be our problem.
There is also a layer of protection for you in having objective professionals handling the quality assurance. If you’re like most of our clients you have staff who occasionally skip steps and cut corners because they trust the knowledge in their heads more than they trust best practices. You can depend on our staff to follow best practices regardless.
Finally you’ll get the cost savings and technology advantage of adding your invoices to our OCR scan volume. For instance, outsourcing validation is not cost effective for most of our clients, not only because they don’t process enough documents, but they also don’t have someone in house who can be dedicated to managing the outsource team. Because of our document volume and our expert staff, that resource becomes cost-effective for all of our clients whether they’re processing a thousand documents or many thousand documents.
We’re also passing an economy of scale savings on to you for the cutting edge technology and the experts who manage it. We have a Document Imaging Architect on staff full time who is charged with making sure our OCR technology and processes are maintained at the highest level of efficiency, accuracy, and security. While that level of support isn’t cost-effective for most companies, even on an outsource basis, our volume justifies it and our commitment to our clients requires it.
So yes, OCR technology can and does make the lives of AP professionals easier. But you can save yourself and your staff a lot of headaches, and your company a lot of risk and expense, by letting us handle the technology while you handle the tasks that are the most productive and meaningful.
Ready to talk? Contact us to learn how we can help.