Automated invoice processing or invoice automation is an automation implementation that frees up finance and procurement team from the repetitive and time-consuming tasks involved in manual invoice processing.
Most of the time, companies describe this automation in 3 simple steps:
- Invoice Capture via OCR
- AP Automation
- ERP Integration
In a nutshell, the invoice is received by the company and captured by an OCR (Optical Character Recognition), the data is then sent to the AP automation solution’s core capability, namely the workflow automation, and finally the deciphered invoice information is sent to the ERP for payment.
The reality is that it is a bit more complicated than this. OCR capabilities vary depending on which tool the AP automation provider is using, workflow automation is also different from one solution to another and not all providers can offer seamless integration, although many argue they can.
1. OCR Capabilities
OCR invoice processing is a powerful tool allowing us to extract data from digital invoices, reducing the need for manual data entry. In other words, it transforms data into metadata to be sent to the next step.
Most commonly, the OCR process follows the following pattern:
Step 1 – Image Import: The OCR receive input from different sources and different format (TIFFs, JPEGS, PDFs, and other image formats as well as digitally created Office documents.
Step 2 – Image pre-processing: To increase recognition accuracy, image quality is enhanced during the pre-processing step.
Step 3 – Document analysis: During this part, the software analyses the layout of each page as well as the document structure. This process defines the areas for text recognition and delivers information about layout and formatting elements for the final document reconstruction.
Step 4 – Recognition: This part is unique to each OCR and its process varies. Most commonly, the software extracts the various information from a set of pre-defined fields and makes a comparison against similar field entries. More advanced OCR’s utilize a combination machine learning capability and AI-driven parameters to creates an unique dictionary of words and patterns attributed to each project. From there, OCRs can increase their recognition accuracy depending on their respective capacities.
Step 5 – Text export & document reconstruction: The data exportation and document layout differs from one software to another. Some OCRs provide many formats such as XML, PDF, Words while others only provide restricted exportation options.
Some more performant OCRs have an extra layer of capacities classifying and matching information against master data in addition to PO matching.
Now, most OCRs follow this process and commonly advertise a rate of extraction accuracy of 90%. However, the reality is that 68% of invoices processed with OCR still require manual input. That is more than two-thirds of the solution currently on the market.
So why is that?
For multiple reasons really:
Firstly, it is apparent that there is a clear difference between solutions out there:
Some are just better than others when it comes to assessing new invoices, discerning between multiple types of layouts, and extracting the data accordingly.
A stronger machine learning capability:
Machine Learning, a subset of AI, analyses the structure of an invoice for patterns to structure the data. The solution might make a mistake with an invoice because of its singular form but will learn from it and match the correct information the next time. While for every mistake, manual intervention is required, strong machine learning capabilities will reduce the need for any intervention as the OCR processes more invoices with less human interaction when weaker capabilities might require more time to adjust.
One-off transaction and infrequent purchases:
For the same reason stated above, infrequent purchases aren’t suited for OCR’s with limited capabilities and require manual data entry.
What to look at when choosing an OCR solution?
Obviously, the first factor to look at is an OCR’s strong AI capabilities and its efficient capture capacity. We partnered with Abbyy, one of the leading OCRs on the market, as they provide interesting classification functions and a powerful machine learning feature.
The second aspect is the variety of languages and currencies the software can process. Depending on your situation, this might not be essential but when investing in a solution, you should always look at its capacity to provide solutions alongside your expansion.
The third component is whether the solution is cloud-based or on-premises. If you want to be able to run your operations from everywhere and have real-time updates, then you should opt for a cloud-based solution.
What are the OCR’s limitations?
OCRs are powerful tools and drastically reduce manual input, but it can’t be used as a standalone technology at is sometimes the case. Whether it could be considered as a limitation or not, OCR capabilities are leveraged once they are coupled with an AP automation workflow that can handle discrepancies and notify users of a mistake.
Without it, you could extract the wrong information without any sign of warnings as the OCR would not react unless you consistently configure the system, which requires manual input.
In other words, OCRs reduce the time-consuming data extraction but require a more sophisticated solution on top of it to monitor its actions.
2. AP Automation
This part is the core of any invoice processing automation and is the most complex. The step is at the center of the whole automation and connects the OCR’s extraction to a company’s ERP while classifying, notifying and handling exceptions.
It enables to process invoices through a touchless experience, hence removing the need for human interaction while drastically reducing costs and time.
There is a broad range of companies offering automated invoice processing software packages but these solutions can be split into two camps:
- Tools that provide automation as an extension of their core solution. Many ERPs offer some interesting features to automate invoices although limited.
- Specific software for accounts payable automation that provides a broader spectrum of features, most notably an invoice library, more complex business rules, different permissions levels, exception handling capabilities, etc.
To give you a better understanding, here are 3 stats comparing manual, OCR and OCR coupled with AP automation:
The average number of invoices processed per month for a company that uses manual processing is 906, processing 5 invoices per hour.
The average number of invoices processed per month for a company that uses OCR is 3,821, processing 22 invoices an hour.
The average number of invoices processed per month for a company that uses OCR coupled with automated processing is 10,387, processing 60 invoices per hour.
You can check our AP automation Guide if you wish to learn more.
Best-in-class software will process an invoice using the following steps:
- Step 1: The OCR capture the data and send it to the AP automation software
- Step 2: The software analyses the data searching for any anomalies and routes the invoice to the person authorized to approve this vendor, previously configured in the business rules.
- Step 3: At the same time, the invoice is also recorded and listed in the invoice library alongside a complete audit trail of who handled it and when it was handled.
- Step 4: Should there be an exception, the system would match the invoice to its vendor and carry on the workflow while notifying managers of that event.
- Step 5: The information is sent to the ERP to fulfill the payment.
In most cases, these AP automation solutions can provide ROIs in less than 8 months.
Which features are important for choosing an AP automation system:
- A document library to monitor your invoices – from inception through to the payment within your ERP, every touchpoint, action, state-change should be monitored and visible at the click of a button.
- A good exception handling function capable to manage your exceptions automatically without the need for human intervention or with smart escalations features
- A flexible business rules capability – configuration is the key to a good system, not customisation. Look for a solution that has a broad set of in-depth features “out-of-the-box” paired with configurable business rules to allow for speedy deployment, lower-cost and quicker ROI to be achieved.
- A user-friendly interface – nowadays people are expecting systems to work fast, be intuitive, have a nice design and add value to the staff members working with it. A system with all of these key points means staff adopt the solution quicker and provide greater positive sentiment to the wider business, which also helps with determining what a successful AP automation project looks like.
What are the AP automation limitations:
This depends on the software you are relying on. Many have limited business rules or cannot handle exception handling very well. Others lack visibility and have no document library. Some offer everything cited above but are not cloud-based or can only integrate with the top 5 ERPs on the market. All in all, limitations will fall under your choice.
We, and a few other competitors, provide this set of features while also having seamless integration and a cloud-based solution. However, many companies miss one or more features that fully enable a touchless experience.
This is why choosing the right solution is quite difficult. From our experience, many fail to assess the market correctly which causes significant long term costs when the process cannot entirely be automated.
We wrote an entire AP Automation Business Case Guide which will give you more information about how to assess your current situation and the trap to avoid when choosing your next solution.
3. ERP Integration
ERPs are typically a suite of integrated applications using a common database. The system provides a continuously updated view of core business processes and tracks business resources as well as the status of business commitments.
Simply put, ERP is the software for integrating people, processes, and technologies across a modern enterprise.
These systems were firstly exclusive to the manufacturing sector and used for production schedules. In the last century however, ERPs evolved drastically as they started to offer more function and provided value to other industries.
Formerly installed on-premises with an important implementation cost, cloud-based ERPs have since replaced these costly installations while improving function applied to finances, human resources, manufacturing, and supply chain.
There are two major types of ERPs currently on the market:
- Smaller ERPs targeting specific industries such as the life of TradeGeko in the retail industry, Tyler Technologies within the public sector or IQMS in the manufacturing industry.
Depending on a company’s needs and situation, a generic yet strong ERP such as Oracle would be suitable when another company might require an industry-tailored solution and will thus turn to an industry-specific ERP.
Why can’t an ERP automate the invoice process alone?:
ERPs systems have constantly evolved to remain competitive in a world of perpetual changes. Hence, these past few years have seen a new movement within finance teams to transform AP from a cost centre to a profit centre by embracing new technologies and innovative strategies.
Alas, and while there has been a major push to further automate the AP lifecycle and enhance processing times and reduce costs, ERP continues to lack the comprehensive and innovative capabilities essential to fast invoice processing such as exception handling for instance.
As a result, most companies must rely on manual human intervention even when using the ERP system-specific function.
That is not to say that ERP providers are incapable of providing such service, but it does not come out of the box and requires a custom-made solution which can be extremely onerous.
The reason why is simple: ERPs are suited to create a collaborative framework around a business’ departments. However, they have not been created to automate a specific process such as AP automation, even less to outperform the specialists within this industry.
That’s why the key is to leverage best-of-breed automation solutions that are specifically designed to automate the invoice process and which provide real-time integration with ERP systems. This, coupled with strong OCR capabilities, will make the automation worth a company’s investment as results have shown that top performers applying best practices have a cost per invoice processed of $2.
Hence, when the ERP is part of this trinary structure, once the invoice has been processed, it is then sent to the company’s ERP, which will add the data to the common database to trigger the relevant action. When automated properly, the process is straightforward as the two previous steps have captured and organised the data according to the company’s business rules.
Automating invoice processing, or AP automation, requires 3 distinctive yet integrated components to achieve great profits and reduce costs:
- An OCR
- An AP Automation Solution
- An ERP
Most commonly, companies will have an ERP with either an OCR or AP automation capabilities, sometimes both but more often than not they aren’t able to reap the benefits of invoice process automation as the structure lack essentials AP automation features such a strong exception handling engine, seamless integration or simply lack an invoice library.