OCR Does Not Work, I Will Never Try It Again.

In over two decades of my professional experience, I never came across a ‘happy’ OCR customer. I would go as far as saying that there are no happy OCR customers!

In fact, one of the reasons we started Infrrd was to change that. As a part of our sales process, we regularly meet prospects who have burnt their fingers in the past with OCR technology. Some of them are so disappointed that they do not want to give it another try… ever!

We have turned more than 30 OCR detractors into IDP believers. But there are so many others out there that are still against trying it again. If you are one of them, read on, let me make a case for why it is time to give it another look.

But first things first, before we go any further, an apology is in order. An apology on behalf of all the OCR salespeople from 10-15 years ago that got carried away making promises that their technology simply could not deliver at the time. Back in the early 2000s, most of these companies had a technology that could convert printed text into digital text. There were no means of understanding text back then, but the businesses really needed it.

I assume that inadvertently a few salespeople said, “Yes, we can do that but it will need some customization” and put their customers through long customization cycles with huge professional services bills that were never delivered. This was back when there were no neural networks and no access to all the algorithms for Natural Language Processing or Computer Vision needed for understanding documents.

But a lot has changed since then.

Why is it different this time?

The first step towards overcoming your past nightmares of using the OCR system may be comparing the features of OCR from when you last used it to today. Let us compare:

As you can see, modern IDP systems are reliable, can handle a lot of complexity, and are very easy to use.

Why should you care?

AnyToday’s OCR is not just OCR, but it is an ecosystem of Natural Language Processing, Computer Vision, Predictive Analytics rolled into Intelligent Document Processing (IDP). Machine learning and AI capabilities have grown leaps and bounds and filled the gaps OCR was not designed to address. If you were to look at an extreme example of machine learning and AI technologies at work, you should look at self-driving cars. These cars use AI and ML to make critical decisions by looking at a huge number of signals that tell them when to stop or take a left or right turn.

Similarly, IDP solutions of today look at several signals fed into several machine learning algorithms to read and understand documents. This gives it the capabilities to extract data that could not be extracted in the past. It also gives it the ability to handle millions of variations of documents without creating templates.

There was a time when ATMs were a differentiator for a bank, but today you cannot run a bank and not provide ATMs to your customers. Similarly, you will not have the choice of not offering automation capabilities to your customers in the future. It is going to become a matter of hygiene. Your customers will demand you do things cheaper, faster, and better using automation technologies. So, even if you have had a bad experience with OCR in the past, you need to figure out a way to give IDP another chance.

What can you do to avoid heartbreak this time?

Before you commit yourself to try this technology again, you want to make sure that you are not headed for another heartbreak and burning another hole in your budget. Start with your toughest and complex documents when you pick IDP technology. Your complex documents could be ones with hundreds and thousands of variations or extremely complex tables or completely unstructured documents. Try these with the IDP technology, let the technology earn your trust again. See it learn and improve itself over time, and then add more documents and scale it out.

Have faith and give it another chance

Modern IDP solutions offer cognitive automation. These technologies have human-like capabilities in understanding and analyzing situations and tasks. For example, Infrrd’s IDP solution uses OCR systems with cognitive technologies, such as artificial intelligence, machine learning, and neural networks-based deep learning, to meaningfully extract and classify documents with higher accuracy, faster processing, and minimal or no errors.

If traditional OCR was focused on digitizing physical copies of documents by recognizing the text characters, its more evolved form in IDP solutions can recognize the information’s context, structure, and presentation. Even if deep learning OCR systems have let you down in the past, have faith and give IDP a spin. You and your customers will be pleasantly surprised.