The economic crisis is driving finance departments to seek productivity gains like never before. In today’s context, the Procure-to-Pay process represents one of the main sources for improving efficiency. Digitalising the process allows to reduce costs and processing times by 6 or more, compared to a manual process…
But these performances are only provided by the best! Achieving such success is possible only under one condition: having maximum automation. Automation itself is based on one main factor: being able to leverage real-time, comprehensive invoice data, valid and consistent, whatever the source.
Concretely speaking, automated reading and validation of invoice data is THE key to performance. In that sense, every percentage of automation matters more than you might imagine.
Extracting accurate data: not so easy!
With most invoices received in paper or PDF formats, there is a real need to automate data extraction.
Before being able to measure the impact of automation, it is important to understand how it works.
What happens when robots automatically read a document such as a supplier invoice? Just like an accountant, their job is to analyse document content and identify useful information to extract, such as the supplier’s name, P.O. number, total amount, etc.
Sounds easy, right? Actually, the only thing more different from an invoice is another invoice! The location of information, syntax, and field labels can change radically from one invoice to another, and a supplier can even change their own invoices from one day to the next:
How do you identify a supplier when all you have is a logo on an invoice?
How can you distinguish the order date from the invoice date if labels are not clear?
How is it possible to identify the currency if it is not indicated on the invoice?
What might be easy for a person can sometimes be complicated for a machine. This is one reason why automation rates can vary from one solution to another and why that criterion must be watched closely.
Automation rate and “remaining workload”
How can you measure the automation rate’s impact on performance?
A first approach involves leveraging the notion of “remaining workload”. Let us explain.
One of the main goals in digitalising the accounts payable process is to reduce the amount of manual workload remaining for accounting teams. However, the remaining workload is directly linked to the performance of automated reading rate of invoice data:
Therefore, when going from an automation rate(1) of 80% to an automation rate of 90%, the remaining manual workload for the accounting teams is divided by 2. And for an additional 2% of automation (92%) the remaining workload is divided by 2.5.
It is easy to see that even small differences in the automation rate have a strong impact on the remaining manual workload.
So many additional invoices that the teams no longer have to deal with! This frees up more precious time to focus on higher added value tasks, such as supplier relations, analyses, risk prevention, and more.
The Holy Grail of automation? Straight-Through Processing!
There is another way to measure the impact of automated data reading on the performance of supplier invoice processing. This method is based on the notion of “Straight Through Processing” (STP), also known as “touchless” invoice processing. The term refers to the percentage of invoices received that do not require any human intervention up through the payment order. This is the Holy Grail for all finance departments.
With ITESOFT, 96% of our fields are read automatically.
The result is that 70% of our invoices go to “no touch”.
Rachel Finkelsztejn, CFO at LIDL
Of course, calculating STP in some cases may be a purely theoretical exercise. For example, business rules in some organisations might require systematic human intervention for certain types of invoices, or for specific fields – and this must be done even when the data is recognised correctly.
Nonetheless, it is worth spending some time with the exercise as it demonstrates the importance of every additional percentage of automation.
Let us start with the example of an organisation that needs to obtain 8 data fields from an invoice (net amount, VAT amount, invoice date, etc.). What would be the proportion of invoices not requiring any human intervention, based on a given rate of automated field reading?
A simple mathematical formula tells us that the theoretical rate in “no touch” is the automatic reading rate from one field to the exponent of the number of fields.
Based on that hypothesis, the proportion of invoices without intervention is, in this case:
% STP = (Automated field rate)8
As such, the stronger the automated reading, the higher the theoretical STP percentage. Above 90%, every percentage point counts… and significantly!
>With 90% automation, maximum theoretical STP is 43%
>With 95% automation, maximum theoretical STP is 66%
>With 96% automation, maximum theoretical STP is 72%
The automation rate is clearly information that must be considered. It is certainly one of the most central and important points for implementing a successful digitalisation project.
Capture as a Service for invoices: ITESOFT’s unique technology
Given the critical role of high automation, ITESOFT developed a disruptive technology: a combination of unique OCR, Artificial Intelligence, and exclusive expert systems.
While OCR is used to convert an image into text, and it is indeed an essential piece of technology, it is still not enough, though many solutions on the market seem to think so.
AI algorithms contribute by grouping identified characters and forming words. Just like a child learning to read, who starts by learning the letters of the alphabet and groups them together to discover words.
Expert systems take the whole proposal to a higher level. Again, like a child learning to read, this involves knowing grammatical rules in order to make sense of what is read. Is the date referring to the date of delivery, payment, or order? Is this amount a total, excluding tax, including VAT? What currency is being used? And so on…
By leveraging these unique technologies based on many years of experience and over 1 billion documents processed annually, ITESOFT is the only provider that guarantees all its customers a minimum automation rate of 85%. In reality, most clients observe automation rates higher than 92% in their own production platforms.
(1): percentage of fields read automatically
Find out more: