What happens when you miss a letter in a name? Not much you’d think.
In 2009, a government agency recorded information stating that Taylor & Sons Ltd – a 124-year-old Welsh engineering firm employing more than 250 people – had been wound up.
What happened next was that Taylor & Sons’ business shattered: orders were canceled, contracts were lost and credit from suppliers was withdrawn and the company was dissolved.
The agency wasn’t entirely (?) wrong; they had made a small (?) error. The problem was the agency wanted to disclose Taylor & Son Ltd had gone into liquidation and not Taylor & Sons Ltd.
The agency reported that it was a silly (?) mistake and corrected it three days later. But, the damage was done. Taylor & Sons Ltd. sued and the company’s claim was valued at £8.8m.
That’s close to £9m for an alphabet ‘s’ and one manual error.
Imagine that happening to multiple documents that organizations process every day. And it could likely be happening with the OCR tools.
While traditional OCR tools can scan documents and convert them into a legible form, it cannot really understand your data.
But, Intelligent Document Processing aka IDP helps you extract data from pdf, images, & other file formats, and understand the context of the data that OCR misses out on.
For example, when you’re processing an invoice, OCR will be able to ascertain the digits 2 0 2 2 as individual digits but IDP will tell you that 2022 is in fact a part of the date of invoice issued and will identify all such dates across documents, with uniformity.
IDP is much more than a modern version of an OCR. And with accuracy levels above 99%, safe to assume, an IDP could likely save you a lot of money and effort if not £8.8m because it mistook 2 for Z.
Here are the following reasons to have Intelligent Document Processing in your workflow:-
- The accuracy of IDP is unmatched. IDP produces 99%+ field-level accuracy, and because IDP captures context-based data, the accuracy can further be increased with data validation.
- IDP adapts to different use cases which means you don’t need to train the data capture model for all the variations for similar document types.
- You don’t need to train the model every time there’s any change in the document you’re capturing data from. You can hire and grow your team for specialized tasks rather than data entry.
Summing up, if you want to automate data extraction from pdf, images, or other doctype formats, you should definitely explore Intelligent Document Processing. If you’re processing documents on the scale, this is your way forward.