Invoice automation (also called automated invoice processing) is a maturing area of automation with limited implementation risks and significant benefits. Invoice automation would free up back office finance/procurement teams to focus on higher value added tasks.
What is invoice automation?
To understand invoice automation, it makes sense to understand what an invoice is and the different types of invoices from an automation perspective. Invoice automation is a critical part of a company’s procure-to-pay (p2p) process (also called source-to-pay s2p) and is part of the company accounts payable management.
Invoice automation allows straight through processing (no human interaction) most of the time for the entire invoice process.
How does it work?
Invoice automation involves
- monitoring for invoices: Invoices arrive in companies as structured XML documents from Electronic Data Interchange (EDI), PDFs and image files via email and increasingly rarely as hard copy documents.
- EDI software can automatically process structured XML documents, complete the necessary payments and create the necessary accounting entries.
- For digital invoices, an RPA bot or a simple email automation tool can flag emails with invoices and forward them for data extraction. Some companies use a dedicated email address for invoices to further simplify invoice monitoring.
- For hard-copy invoices, companies are switching to using a single address to centralize invoice scanning
- invoice capture: Extracting relevant details (e.g. bank account, ordered item) from the invoice. If software does not have confidence in the results, it is sent to employees for a manual check.
- evaluating invoice against order records and criteria to ensure that the payment is indeed a valid one. Evaluations include
- cross-checking invoice against purchase orders
- cross-checking invoice for duplicity
- using working capital optimization policies to decide payment time
- using limits to to decide whether to manually process invoice. Invoices that are abnormally large compared to a suppliers’ usual invoices may need to be manually verified to ensure that wrong payments are not done
- checking the invoice against VAT rules
- recording invoice-related information in systems. For this, invoices without purchase orders need to be added to a general ledger account and machine learning solutions can be used to match invoices to accounts.
- making the necessary payment to settle the invoice
All steps except invoice capture are rule-based processes. However, invoice capture relies on machine learning to extract the data in the invoice. For more, please read our article on invoice capture.
Before automation, back-office teams would
- look at invoices, understand the relevant data in the invoice
- feed it to the relevant systems so payments and system records would be complete
- In rare cases, team would notice irregularities in the invoice and contact the supplier or the ordering team to resolve the issues.
What are the benefits of invoice automation?
Benefits include
- reduced invoice processing expenses – Hypatos indicates that businesses can achieve up to 90% cost saving.
- increased focus on activities with higher value add
- reduced invoice processing errors as personnel only focus on hard to interpret invoices
What is the current level of invoice automation?
Almost all invoices submitted through a company’s Electronic Data Interchange (EDI) are automatically processed.
However, for large companies a significant share of invoices (up to 50%) are sent by smaller companies that are not part of the company’s EDI. While public info on automation rates on these documents are scarce, we had a chance to talk to multiple vendors. Most Fortune 500 companies using technology that was not developed in the last few years have 10-15% automation rates while leading edge, deep learning based invoice extraction solutions can increase that to >80%.
Why is invoice automation important?
Since the whole invoice management process is already digital, we have seen companies deprioritize invoice processing automation. This would be a mistake because even though the whole process is almost completely digital even without automation, it relies on a lot of manual labor. In the pre-automation process, invoices, the extracted data and payments are digital but the process is no less time consuming than a completely analogue process.
This issue of manual labor in digital processes is a widespread issue as companies focused on digitizing processes but not automating them. But this is being fixed as companies focus on automating repetitive processes. They achieve this mostly with Robotic Process Automation. Our collection of RPA use cases revealed that invoice automation was the most common area of RPA automation. If you are curious about RPA, we have the most comprehensive set of answers on RPA.
What are the leading invoice automation companies?
Invoice automation can be split into 2 parts:
- Invoice capture: This is the most critical part of invoice automation and is about extracting data from invoices. We have looked into invoice capture companies in our article on the topic.
- Rest of invoice automation includes rules based processes and there are 2 types of solutions
- RPA based solutions can help automate the entire process. Please read our article on RPA tools to see all 50+ RPA vendors.
- Accounts payable management solutions such as those provided by Kofax are providing specialized invoice automation solutions.
If you have questions, feel free to contact us:
Featured image source
2 comments
Good points on why automation is important. Are you familiar with human in the loop automation? I’ve been looking a lot of places and have found other software that include this feature (bisok is one). Do you know if Kofax does something similar? My department is looking at several solutions. Thank you for your help
Hi Matthias, thank you for contributing.
Most ML based invoice automation vendors (e.g. Hypatos and others) offer human-in-the-loop (HITL) functionality to enable continuous learning. Kofax allows users to modify templates which I wouldn’t call human-in-the-loop functionality since template configuration is not as simple as HITL. In HITL, you improve future extraction results just by clicking on fields on the document, which is what ML based vendors offer. However, in Kofax, you need to improve an existing template which requires a lot of thinking and ultimately results in a solution that is not robust, suppliers often change templates or churn.
And heads up, had to remove the link in your comment as per our policy. If we allow links, we get too many low value comments.