Everyone wants to capture more information, faster, while spending less time to process it manually. That’s a tall order, but document capture technology helps to achieve this quest for digital transformation and efficiency, some of which is much older than you realize.
Case in point: optical character recognition (OCR). OCR has been used widely on a commercial basis for over three decades, and is often central to an accurate and efficient capture process. While you can’t “just OCR it,” the technology has evolved significantly and, when combined with more recent technologies, greatly streamlines business processes, enhances customer service levels, and adds more to your bottom line.
Here’s what you should know about today’s OCR and how it helps solve business problems today.
The Evolution of OCR
OCR translates pixels on scanned documents into searchable data. With its analog roots dating back to before World War I, digital OCR was developed by Ray Kurzweil as part of a reading machine for the blind in 1974. OCR really took off commercially in the 90s when document scanners became more widely adopted. OCR engines rapidly evolved in the 2000s and OCR now has become a mature technology, and greatly refined.
Automatic Document Recognition (ADR) was developed in the 2000s and married with OCR in document capture applications like Kofax TotalAgility so that users can scan a wide variety of documents without having to pre-sort them, which takes longer than scanning itself. ADR software classifies them and, based on processing rules for each document type, OCR then extracts data over the entire document to enable full-text search or capture information in document zones where specific fields reside.
Robotic Process Automation (RPA) is a technology that has recently emerged and uses software “robots” to automate specific processes. RPA can be coupled with OCR to automate the extraction portion of the process.
Common OCR Questions
How Accurate Is OCR Software?
Everyone wants the highest accuracy but improvements in OCR technology have rendered accuracy almost irrelevant as a point of comparing software OCR engines. All OCR software on the market has an accuracy rate of over 90%.
In the 2000s, some were drawn to choosing OCR engine A with 94.5% accuracy vs. OCR engine B with 95.6% accuracy. However, this is not really a demonstrable difference because the comparison was not based on exactly the same documents scanned in exactly the same condition, and the difference was within the margin of error of the testing itself.
How Much Data Can Accurately Be Extracted by OCR?
While OCR technology can create dramatic efficiency gains, it’s not perfect and will not eliminate manual processing and quality control. However, it can reduce manual document processing time by over 90% compared with an entirely manual process. It can also dramatically reduce mistakes.
Whether templates need to be created depends on the business case for how many documents need to be developed and how much labor is involved in doing so vs. the option of investing in higher-end, advanced capture software that conducts full-text OCR without the need for templates. In either case, OCR works essentially the same way, so it comes down to the capabilities of the advanced capture software.
Does OCR Self-Learn or Does It Use Artificial Intelligence (AI)?
The short answer is neither: OCR is not self-learning, nor is it considered AI.
While you have to manually “teach” OCR-using software new rules based on new documents or conditions, this is often a quick process that is needed less and less over time.
More Questions?
Like any decision relating to business process improvement, the capabilities of the technology have to be considered against the desired business outcome. The question becomes less about what the technology can do and more about what you are trying to achieve. OCR is a technology component that is almost entirely wrapped within advanced capture software, and not a standalone solution in itself.
Combined, OCR and advanced capture technology can streamline most business processes by up to 90%. While this means there is still at 10% manual labor component, the time savings and quality improvements can be a major win. RPA software like Kofax RPA may get you closer to 100% but requires a fair bit of planning and testing to determine fit.
At MetaSource, we design solutions that blend OCR and other state-of-the-art recognition technologies with customized business process solutions that fit what you are trying to do. Contact us and we’ll help you get on the right track.