Automation Opportunities in Accounts Payable

I have been telling anyone who will listen that the robotic automation of business processes is set to fundamentally change the nature of the BPO market. The stark truth is that, to date, most of the real-live success stories have been in industry verticals such as telecoms, so, if robotic automation is to really live up to its full potential, then it should be able to have a significant impact on those generic business processes that are carried out across all industries. Therefore, we set ourselves the challenge of understanding the scope for software robots in that most process driven function of Accounts Payable.

The first thing we discovered was that real-live examples of full robotic automation in Accounts Payable are very rare. However, there are a number of examples in related F&A areas which demonstrate significant benefits are to be had. We believe that there are enough of these examples to suggest that similar benefits could be available in the AP function.

To date, the biggest improvements in AP processing have been down to three approaches:

·      Getting manual information into the system electronically, e.g. scanning and OCR

·      Automating the processing as much as possible through workflow

·      Standardising the data fields to improve interfaces e.g. EDI

Through our research we believe that robotic software automation presents potential further opportunities in the areas of:

·      Cross-system manual processing

·      Data gathering and reporting

·      Reconciliation of matching errors

·      Monthly account closure

·      Bulk data updates

·      ERP IT processes

We have analysed each of these areas to identify potential opportunities and relevant case studies.

Cross-System Manual Processing

Typically, humans are used to provide a flexible interface between a number of different systems that are used in a process – this is colloquially referred to as ‘swivel chair processing’: data is read by the human on one system or screen and keyed into another system, sometimes with additional steps in-between e.g. adding a zero to the beginning of the account number. 

Middleware (IT systems integration) can provide solutions to these interfaces but they are typically expensive and complex to implement. Software robots provide a much simpler implementation of the interface, carrying out exactly the same steps as the human but at a fraction of the price. This requires no or minimal system intrusion and provides 100% consistency with the process requirements.

In AP, opportunities would exist where multiple systems (SAP and non-SAP) are having to be used within end-to-end processes, for example invoice matching, supplier information checking, letter creation and period end closing.

Cross-System Processing Examples:

·      Barclays Bank has been working with Blue Prism software since 2006. They have seen £175 million p.a. reduction in bad debt provision in their AR function and over 120 FTE saved. Example processes that have been automated include:

o   Automated Fraudulent Account Closure Process – Rapid closure of compromised accounts,

o   Automated Branch Risk Monitoring Process – Collation and monitoring of branch network operational risk indicators,

o   Personal Loan Application Opening – Automation of process for new loan applications.

·      Co-operative Banking Group has automated over 130 processes with robotic automation including complex CHAPs processing, VISA charge back processing and many back office processes to support sales and general administration. 

·      Stadtwerke Düsseldorf AG (a large German utilities company) lacked control of their current billing process due to the disparate systems, so used Automic to integrate and orchestrate all the steps into a cohesive end-to-end process: billing systems are now fully integrated; process data is automatically exchanged between meter-reading systems; process checks are performed to validate outputs and to identify errors as they occur. All of which has resulted in 80% fewer billing errors.

Data Gathering and Reporting

Finance processes require data inputs from many different sources, many of which are in different formats. Often this creates bottlenecks in the process and introduces errors, or creates the requirement for manual intervention before the data can be used effectively.

Robotic automation can manage much of the data gathering burden – effectively managing data readiness, cleanliness and input into the relevant finance process. A form of automation, electronic invoicing (or e-Invoicing), completely removes paper from the invoice process, i.e. no paper, scanning, or OCR, resulting in straight-through machine-to-machine processing from a supplier’s billing system directly into the customer’s AP system.

It can also automate business intelligence process across, for example, SAP BW, SAP ERP and Business Objects to ensure every user has the right report, containing the right data in front of them when they need it.

Data Gathering and Reporting Examples:

·      The manual invoice process of a large steel company did not have any scope for a standard workflow solution, so exceptions generated during the process were manually tracked and resolved over phone, email or fax. Their automation solution allows invoice data to be entered into SAP and integrated with an OCR solution for scanning physical copies of documents. This has resulted in: reducing the number of SAP user licenses, a reduction in the time for resolution, (63% of invoices are automatically posted in SAP), and a reduction in manual interventions by 78%.

·      An international electricity and gas company automated their Accounts Receivable activities (Billing, Disputes, Cash Application, Order Management) through an automated weekly chase report to reduce TAT in Billing, creation of acknowledgement letters through the ARC database, automating the upload and download process in SAP, and automating the copying process of work request numbers. This resulted in savings of 5 FTE (out of 71), an increase in disputes accuracy of 30% and unallocated cash value reduced from £2m to < £600k.

·      Dixons Retail was already processing 90 per cent of its invoices through a mix of high-end Electronic Data Interchange (EDI) software and Contempus Invoice, a customised invoice processing application, but there were thousands of companies who did not have an EDI solution in place. They implemented an e-invoicing solution that can work with any accounting system, which means they did not have to make changes to their accounts payable system, and the barrier for entry for its suppliers was low. Also, neither buying organisations
nor their suppliers were required to implement hardware or software. As a result Dixons Retail Group’s e-Invoicing system has replaced more than 100,000 paper documents.

Reconciliation of Matching Errors

One of the most manually intensive processes in the Finance function, and in AP particularly, is the reconciliation of errors due to incorrect matching of data between documents. Because this process requires inputs from different systems, is inherently non-standard in each case and can require some judgement, it is usually carried out by humans.

By using inputs from other data sources, processing different matching options far faster, and applying semantic reasoning, robotic software automation can replace much of the reconciliation task, thus significantly reducing the number of people required.

Reconciliation Example:

·      The excess queue procedure at the Co-operative Bank is carried out daily to accept, reject and return direct debits, cheques and standing orders as a result of the customer having insufficient funds to meet payments. Overnight BACS processing results in a daily ‘queue’ of customers with payments due to leave their accounts and with insufficient funds to meet these payments. A nine-person team in the bank would have the daily responsibility of manually reviewing the 2,500 or so higher risk accounts. They would then make a decision to either return or process the payments depending on the account profile of each customer. The automation of the entire procedure means that the bank now has a ‘virtual’ team of 20 people completing the workloads by 11am each day instead of a team of employees working to meet a 3pm daily processing deadline.

Monthly Account Closure

The monthly account closure is typically a complex process involving many data inputs, plenty of reconciliation and some elements of judgement. The number of people involved in the process is typically very high, and the time taken to close the accounts has a direct impact on the financial position of the company, but must be 100% accurate.

Previously, much of the reconciliation work has required human input across many data sources – robotic software automation combines a number of the approaches already mentioned into one critical process. By being able to access multiple data sources, make fast but relatively complex decisions (and to do that 24×7) software robots can significantly reduce the labour required, and the time taken, to close monthly accounts.

Monthly Account Closure Example:

·      A group of 250 National Health Service trusts have automated their month end-close process. The process initially took 15 people 12 days; but it is now down to 2 people and half a day through automation. In this example the automation software works on top of Oracle ERP.

Bulk Data Updates

It can often be the case that multiple data records require updating due to changes in compliance regulations or because of the introduction of new systems. Additionally, monthly data processes, such as Billing runs, can cause bottle-necks and require extensive human intervention to be able to run smoothly. Whilst scripting tools can provide a robust solution, they can sometimes take a significant time to develop and test, and will require IT involvement.

Robotic software automation can simply and accurately provide a means to update records using the existing systems and screens but without human (including IT) involvement. It can also allow monthly processing runs (using, for example, SAP IS-X) to be run in parallel, and is able to cut out significant amounts of manual intervention. 

Bulk Data Update Example:

·      Many NHS Trusts suffer from the problem that no notification is provided by the Patient demographic Service (PdS) in the event of a significant change of circumstance, such as the death of a patient. This means that, even though a GP may have informed the PdS of a patient’s death, the hospital may well remain uninformed. As a consequence, unnecessary appointments may be held open and potentially upsetting appointment reminders may be sent out. University Hospital Birmingham had circa 50,000 deceased patients’ records to be updated. The software carried out the bulk-update and also helps to keep the Patient Master Index (PMI) in the PAS up to date which enables crucial audit and reporting activities to be delivered.

ERP IT Processes

Because of the large number of data records involved and the complexity of the processes, some of the IT processes involved in ERP systems can be very time consuming and labour intensive. In particular, where there are a large number of jobs to be run, the ERP scheduler can struggle to manage the exchange of large data volumes amongst different systems. Additionally, complex jobs that take a long time can cause issues with handovers between shifts of IT personnel.

Robotic software automation is able to integrate the data gathering from the disparate systems and automate manual tasks in order to reduce error rates. The processes can be significantly speeded up, reducing system down time and the potential need for shift hand-overs.

ERP IT Processes Examples:

·      Linde Group, a gas and engineering company, sends around 100,000 invoices monthly. They consolidated their SAP systems at 41 production sites into a large single system for Central Europe to increase the effectiveness and efficiency of customer and enterprise data processes but, as the number of jobs increased the internal SAP scheduler could not keep pace with demands of exchanging vast volumes of data with legacy systems, handling route-planning, gas cylinder administration systems. Linde automated their business processes end-to-end by integrating their SAP and non-SAP systems so that manual tasks are now automated and errors are reduced. Benefits included $30m in cash flow due to faster invoicing of European customers.

·      Air Products is a supplier of hydrogen and helium. They need to make a weekly copy of the volume SAP R/3 system that runs all transactions worldwide in order to create a Production Query system for end-user reporting. The original home-grown solution was semi-automated , with operators following 53 pages of steps taking 36 hours to complete and requiring 11 hours manual effort. Human error, and lack of coordination were the biggest problems with handoffs between shifts leading to confusion and missed steps. Air Products DBAs and the BASIS team automated the entire process so that it now takes 19 hours to perform a system copy.

The AP Process Opportunity

Of the key AP processes, we have identified those that have the greatest potential to deliver automation benefits

AP Process

Potential Benefits

Manage document receipt


Verify and Process invoices, including Matching


Process Payments (Disbursements)


Respond To Enquiries (Query Handling)          

Period-End Closing

Signatory File / Authority Matrix Management  
Reconciliation of GR/IR Clearing  Account  
Debit Balances Management  
Statement Auditing                  
VAT compliance  


Records Retention                 

Performance Measurement and Effectiveness



With a software agent costing around one third of a typical offshore BPO FTE, and one ninth of an onshore FTE, there are clearly some significant benefits to be gained from exploiting this new technology. The examples given above show there are ways to extract savings from Finance and Accounting, and specifically Accounts Payable, but we haven’t yet seen a compelling case that would have the impact that we suspect is available. It may be that the automation solution is a combination of various tools and approaches, therefore an ‘automation strategy’ is likely the best way forward for a consumer of BPO services. However, at the same time, we urge the software vendors and BPO providers themselves to focus attention on this potentially huge opportunity.

Share on:

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Follow on Feedly