OCR vs IDP: improve your document transformation process

To elaborate, it converts those formats into text files removing the need for double keying, and offers the promise of reduced costs and a reduction in errors.

However, for most, the reality is OCR is only effective for basic-level conversion. This is because the application is not intelligent and is simply looking at the file and extracting it.

Thus common reading errors such as numbers being transposed as letters and vice-versa remain a constant challenge.

In contrast, IDP (Intelligent Document Processing) uses AI and computer learning to interpret data from structured and unstructured content, leading to significantly higher levels of accuracy.

If you are looking to reduce the costs of double keying across your business processes, whether that is application forms, or inbound orders and invoices then what follows will ensure you make the right purchase decision and are not held-back in your document transformation process by legacy technology.

In this blog, we will cover…

At the heart of Intelligent Document Processing (IDP) are a mix of technologies and processes including artificial-intelligence-based and computer learning.

To expand, these technologies analyse and make intelligent predictions of content contained in structured, semi-structured, and unstructured documents.

So unlike legacy OCR that is looking at content, IDP is reading and making contextual choices about the content. Thus, making IDP applicable to a wider range of business process tasks.

IPD offers great flexibility of use than OCR and therefore a better choice of technology that can reach across many processes within a business.

An early win for businesses migrating to IDP is automation of the Accounts Payable process. Transalis own research*, supported by other researchers such as Billentis** highlight savings of 60% are possible by simply removing the need to double-key invoices.

Account Payable Automation from Transalis leverages the technological advances of IDP to the benefit of our customers in several crucial ways. Below are a few examples.

OCR technology relies on each supplier’s invoice specific invoice format to be mapped so that the technology “knows” where to look for certain fields such as a Purchase Order for example.

In fact, this process can take up to 2 days development work per supplier, resulting in a solution roll-out that could take months if not years.

On the other hand, IPD uses machine learning, built-up from processing millions of documents annually to immediately recognise formatting within a newly presented invoice.

Moreover, combined with AI that starts to make decisions on where to look for fields such as the Purchase Order but also the territory, currency, and mandated fields.

Hence, this reduces the onboarding of new suppliers from days to hours.

As mentioned earlier, OCR is looking and not reading a document creating challenges in differentiating words and numbers.

Also, this is true of format changes to a document. For example, a supplier decides to update their invoice and change the position of the Purchase Order number. With traditional OCR solutions, this would create a failure in the process or worse still process the document incorrectly.

Furthermore, as IDP is reading each new invoice and comparing it to what has been processed before these type of format changes, adjustments are easily spotted and flagged as an exception for review.

Additionally, to reduce the correction time for such exceptions, Transalis AP Automation will place the invoice in an exceptions queue and issue an alert email to the user.

Thus, the user can view the specific invoice via their online platform and use the exception tool to review the errors highlighted and make on the fly corrections.

The use of machine learning and AI to deploy an AP solution reduces the development resources typically required with OCR solutions.

Thus, this results in a reduction of cost and time during the implementation phase. Overall, meaning better value for our customers.

As in the example of invoice mapping, the machine-based learning derived from processing millions of different documents annually and AI technology used by Transalis results in a library of pre-trained documents that can be deployed at pace.

Transalis can deploy a solution that resolves many laborious and time-sensitive processes, whether that is scanning utility bills as part of a KYC process, insurance claim submission or expense receipts via a phone app,

Smart extraction of information will deliver significant savings to the business in both time and money.

Overall, our recommendation is to leverage these savings to directly integrate with your specific back-office systems.

In fact, Transalis offers a range of off-the-shelve API integrations into many ERP, CRM, accountancy and enterprise applications.