Optical Character Reading (OCR) has become widely adopted within the accounts function over recent years.
Rather than manually rekeying information – such as invoice data – OCR offers a chance to capture information automatically, speeding up the accounts payable process.
However, OCR has limitations and while the technology has improved in the last few years, accounting has changed. With so many companies now generating and emailing digital invoices direct from their accounting software (it is standard functionality), OCR is becoming an increasingly outdated method of data capture within accounts payable.
In this blog, we look at why relying purely on OCR for accounts payable automation is being increasingly marginalised and analyse the technologies which are replacing OCR to perform the data extraction tasks faster, more accurately, without human intervention and at a much lower cost per invoice processed.
Cost of Inaccurate Accounts Payable
Optical Character Reading (OCR) is the process of using technology to identify printed or handwritten text inside digital images or physical documents – such as scanned invoices. It works by examining the text of a document and translating the characters into code for data processing.
So far so good.
The problem is that OCR is simply not accurate enough – with most systems experiencing error rates up to 10%, even after finance teams have endured the laborious process of ‘teaching’ the OCR how to work. The result is every invoice must be visually checked, often taking 50% of the time to enter the information, before making corrections.
Common problems include mixing 0 and O; 5 and S; while poor quality or low-resolution scans reduce accuracy even further. Plus, of course, most invoices now arrive via email as a PDF and it is hugely inefficient to have to print out these invoices to scan them back into a system to be read, something that 23% of finance teams are still doing.
Just consider what that means for a moment: 10% of invoices are incorrectly submitted to your accounting systems. That means without undertaking manual verification of each one, errors can lead to potential under or over payment, cash flow issues and credit control problems. It means time spent on the phone to suppliers to resolve problems – including suppliers refusing to deliver until the correct payment is made. What may appear to be an ‘acceptable’ degree of inaccuracy can have very serious business ramifications.
Whilst there will always be paper invoices, so OCR still has a role to play, the future of digital invoice data capture has already moved on. Most finance departments have already migrated to sending their sales invoices via email as a PDF. It is fully automated and arrives instantly 100% of the time. Compare this approach to printing them out and posting them and the benefits are obvious (and green).
The critical difference is these emailed PDF documents contain all the data that is used to render the human readable copy. Instead of trying to convert readable characters into data for every single invoice processed by OCR, the new technologies identify where the relevant data is within the supplier’s invoice and once mapped, the data extraction for that supplier is fully automated thereafter. Extracting data removes all the inherent weaknesses of trying to create data with OCR.
This migration has been actively embraced by a few forward-thinking accounts payable (AP) automation providers, leveraging digital technologies in order to reduce costs, avoid errors and free up valuable time for the finance team. As a result of the government’s Making Tax Digital policy, even the smallest of SMEs are now using online software to generate and email their invoices, with most businesses receiving around 90% of their invoices as rich PDFs via email.
Digital invoice capture is not just accurate: it is automated, faster and cheaper. Every invoice is processed automatically and all the information on the data-rich PDF – including supplier bank details – is pulled and delivered directly into the AP automation application invoice register, ready for GL coding and approval. Also, a PDF of the invoice, the original email and any other associated documents are attached.
Using this approach also reduces the risk of fraud. Every invoice is validated against the supplier’s email address and instances where the invoice data is not formatted as expected are highlighted. It also validates and cross-checks both header and line-level values, and even validates the supplier bank information against that held within the integrated accounting system.
With immediate and accurate invoice data, finance teams can push forward with AP automation. Workflow approvals can be fully automated, using multiple criteria such as invoice value and supplier, to determine the right approval process. Direct integration with the accounting software ensures up-to-date budget information and a real-time view of spend. For the finance team, the real-time view of invoice status from the moment it is received ensures any exceptions that require active management can be highlighted and tracked.
Real time, accurate invoice data is an essential component of successful AP automation. It supports better cash flow management; enables effective supplier spend analysis; and facilitates the zero-touch AP processes that reduce both time and cost.
Any data inaccuracy compromises AP automation – and creates additional business cost and risk. In an increasingly digital world, there will be no place for paper invoices and therefore no place for OCR. It has become resource-intensive, slow, inaccurate and expensive. Isn’t it time to step into the next decade?
Explore our Compleat Guide to Accounts Payable Automation and find out how AP automation can deliver benefits for both AP and business managers, saving money, improving cash management and transforming procurement.