Discover the attraction and benefits of using RPA in Accounts Payable and read about a successful invoice automation project that resulted in a high-tech firm saving time and improving quality.
There is a lot of content about using RPA (Robotic Process Automation) for Accounts Payable. For many organizations, Accounts Payable is where they start their RPA journey. The legacy process of automating invoicing can be complex, with many non-standard formats and approvals.
However, most organizations and RPA firms assume that this makes it easy to achieve actual results with RPA. But the fact is, many invoice automation projects have failed to deliver the expected results. There are also specific cases where clients have been extremely successful.
The attraction of using RPA in Accounts Payable is that AP (Accounts Payable) is a low-risk function, with rules-driven processes, often with a high volume of transactions currently being processed manually.
What makes the difference between success and failure with RPA in Accounts Payable?
Let us look at an example of a successful RPA project in Accounts Payable.
The client is a global high-tech firm headquartered in California that specializes in Internet connection and data centers. This firm has managed to save tens of thousands of hours in processing using invoice automation.
It is not just about the hours saved – it is about the quality of the working environment for their teams, the freedom this creates to do higher-value analysis, and the fact that the bots work 24 hours a day, error-free, with a clear and accessible audit trail of actions.
The four key components of successful invoice processing automation.
- The invoices that you are trying to process.
- The RPA platform that you will use to automate your invoice processing.
- The OCR (Optical Character Recognition) solution that you are using to extract the data from these invoices.
- And the ERP solution that you are using as your AP system.
1. The invoices that you are trying to process.
In this case study, the firm receives most of its invoices via email. These invoices have the same common, recurring information, regardless of how they are laid out and structured:
- Header, with some header data
- Footer with some footer information, like totals and sub-totals, tax, and shipping data.
- Line items were line items with quantity, unit price, the total amount for that line.
- They can have a single page or hundreds of pages.
- A single PDF document that comprises multiple invoices.
- Content could be in English, Chinese, Spanish, or German.
- It could be in the U.S. Dollar, British pound, Singapore dollar, or one of 100+ other currencies.
- It may be a PO invoice or a non-PO invoice.
- It could be a credit memo or a statement of accounts.
That is an incredible number of variations and an even bigger number of permutations.
2. The RPA platform that you will use to automate your invoice processing.
The firm in this case study is using one of the leading RPA robotic platforms. You can choose many platforms from; UiPath, Blue Prism, Automation Anywhere, Microsoft, or other RPA platforms out there. The decision is yours. We can help you with that decision, or you may already have chosen a strategic platform to use.
3. The OCR solution that you are using to extract the data from these invoices.
You might be using ABBYY, one of the leading OCR platforms. You might be using IBM Data Cap or one of the built-in features of the main RPA platforms, like Blue Prism Decipher or UIPath Document Understanding Feature.
4. The ERP solution that you are using as your AP system.
You can be using SAP or Oracle or any of the other Accounts Payable/ERP packages available. How did this major firm deliver one of the most successful automation programs for invoice processing when many others have failed to achieve any value from RPA? The answer is a pre-built solution that is the foundation of their results in automating invoice processing. It is one of our fast-to-implement solutions, targeted at finance and accounting teams, to enable successful invoice processing using RPA.
How Pre-Built RPA Solutions Work in Accounts Payable
InvoiceBotz is a smart-engineered RPA and cognitive document processing solution that uses your AP system to handle PO and non-PO invoices from receipt through registration.
Speed-to-value since; they are deployable in less than a month, for your top 20 supplier invoices.
This can be easily extended to other suppliers to further improve the throughput of invoice processing to reach your target. And, with an additional extension, you can even initiate payment after the approver sign-off of the invoice.
InvoiceBotz has business rules ready to be applied on extracted data from the invoice.
These are externally configurable during the deployment of the solution. This set of business rules is the crux to increasing the quality of invoice processing. It answers questions such as how do I handle:
- Invoices when a supplier is set in different locations?
- A situation where the PO line description is different from the line on the invoices?
- Invoices that come in multiple currencies in each line item?
- Installment payments?
- The logic for duplicate invoices?
There are 30+ out-of-the-box business rules included in the Invoice solution, with more being added all the time.
OCR does a powerful function in extracting data from invoices using OCR.
But it is the application of the business rules that makes the solution more successful, and faster.
In addition, InvoiceBotz can be configured with parameters configured externally for the process automation, for example:
- How invoices are going to be received.
- Where they are going to be stored.
- What OCR platform is used for data extraction.
- How to invoice PDFs should be loaded into the AP system.
These are the built-in components that make this solution powerful, effective, and fast to implement, whatever size your business is.
How Do the WonderBotz’s Accounts Payable and Invoice Automation Solutions Actually Work?
InvoiceBotz makes this happen in five steps:
1. Receiving The Invoices.
So digital worker or the software robot or the RPA. software robots continuously monitor an email account or a shared drive for new invoices, which is very similar to what a human would do
Once it receives invoices, it then saves this new invoice to the robot work queue.
Then comes the second step, which is extracting data.
2. Extracting the data from these invoices
The robot now works with the underlying OCR platform to extract the data from these invoices.
Some of these OCR platforms use machine learning to train their engine to improve data extraction quality over time.
3. The quality loop to improve the data quality.
The quality loop is unique and proprietary to our solution. Now that the data from the previous step, which has been extracted using the OCR platform it is still raw.
The robot then uses a set of business rules to cleanse further, validate, transform, and then confirm that the data is ready for consumption by the AP system.
4. The audit loop to ensure that the right quality of invoices are going into the AP system.
During the quality loop, the rules are applied to catch errors early and tag them.
For those situations where an additional audit is required or when the confidence level of OCR data extraction was lower than a certain threshold, these invoices can now be redirected to the AP team for a human to do further checks and corrections.
Once corrected again by the human, the AP team can click on a button, and now the invoice flows into the process, and it goes to the next step.
5. Registering the invoices.
And the last step is now receiving the invoices or registering them into the system. Now that this data quality is safe, the robot prepares this invoice data to be loaded into the AP system.
Once these invoices have been created in the AP system, the robot uploads these invoices’ individual PDF documents to the AP system.
So, to summarize this, a successful RPA service like InvoiceBotz provides business-managed, low touch invoice processing, from receipt to registration, using OCR along with RPA. This reduces manual keying, focusing your AP team on exceptions and higher-value analysis that only people can perform.