In 2021, the focus for businesses has been on recovery and growth. Having more visibility on spending helps businesses to understand their financial situation better, making procure-to-pay (P2P) a key process for them to optimize.
Technologies making up intelligent automation (IA) such as robotic process automation (RPA), artificial intelligence (AI) and machine learning (ML) can help organizations streamline the P2P process and remove error-prone repetitive manual tasks from human workers’ workload. They can allow for more efficient invoice processing, seamless supplier payments, and increased working capital and cash flow.
Interested in optimizing your accounting processes? Register for PEX Live: Finance Process Automation 2022, taking place online on 22 – 23 November 2022
Based on research conducted with P2P experts, this PEX Network report provides an overview of the need for organizations to digitize their process and implement IA technologies to help encourage business growth. Featuring real-life case studies from LafargeHolcim, Tesco Business Services, Deutsche Telekom and Couchbase, this report aims to provide an understanding of how to unlock business value through the optimization of the P2P process.
Optimizing P2P to focus on growth and manage post-pandemic disruptions
More than a year from the start of the Covid-19 pandemic, the focus for businesses has been on recovery and growth. Organizations have realized that they needed to have more control over their spending and greater visibility over cash flows.
This requires optimizing the P2P process: capturing more spend under management, having greater visibility into transactions, increasing control and compliance and reducing risk, such as payment fraud and supplier risk.
Many organizations, however, have manual P2P processes that are time-consuming, inefficient and labor-intensive, which can result in payment delays or errors. These manual processes can impact supplier relationships and prevent organizations from benefitting from early payment discounts.
To overcome these challenges, organizations are increasingly investing in P2P solutions that automate workflows to request, procure and pay for goods and services, allowing for an optimized and more seamless process. The market is expected to grow with a CAGR of 8.5 percent between 2020–25 and is forecasted to reach US$2.4bn by 2025, compared to US$1.8bn in 2019.
NoSQL cloud database company Couchbase has implemented such a solution from Coupa to address its challenges on the P2P process. Accounting teams had to juggle between multiple systems to pay vendors, leading to time consuming and complex processes that exposed the team to potential errors.
The lack of integration between the legacy systems led to data silos, which resulted in a lack of control and visibility over payment status. As each separate system was a potential point of failure, audit and compliance were a challenge.
Ariana Cutler, manager of global AP and travel and expense at Couchbase, says: “It was a sync, to a sync, to a sync, going from Coupa to Netsuite to bill.com and then all the way back. There were so many opportunities for error, for something to not sync over properly, and there was a lot of reconciliation work between every system.”
The team had to select invoices for payment manually, one by one, for them to be approved and payment to be sent out. There was a need to have one single process.
After implementing the Coupa platform, the immediate impact was felt on the efficiencies gained from eliminating manual reconciliations and streamlining processes, which allowed Manisha Goel, director of accounting at Couchbase, to move her team to more strategic projects.
Now, Couchbase is processing approximately 90 percent of its payments, including ACH and digital checks through Coupa Pay, which gives Goel visibility and control over their spending.
“It was also about internal controls,” Manisha recounts. “Before, we had to toggle between various systems to find information or research something, whereas now we just have one source of truth.”
Cutler adds: “The single system integration is what really attracted us. Having it all in one software platform was the biggest thing. It’s now relatively seamless — and painless — to make payments. I can now approve a lot of invoices at the same time, the platform saves me around two hours per week.”
The next section of this report discusses how organizations can leverage IA technologies such as RPA, AI, ML and data analytics on their P2P processes.
Leveraging IA to digitize end-to-end P2P
There are several financial and practical benefits provided by RPA, machine learning and data analytics to optimize the P2P process and remove manual processes, which – as the Couchbase example in the previous section of this report demonstrated – can help teams focus on more strategic, value-adding tasks such as post-order work they could be doing with suppliers to improve the relationship.
Global building materials manufacturer LafargeHolcim processes more than 3 million invoices per year, with more than 500 people supporting P2P across the company’s three service centres in Europe, the US and India.
For Carlo Peretti, its VP – finance and global director of the business shared services program; the aim is to create value for the company by improving the P2P process.
“The end journey is to have an efficient, error-proofed, more robust P2P process driven by IA to let the human talent do more value-adding tasks and understand how we can drive the business with P2P,” he explains.
Initially, Peretti started his P2P initiative by standardizing processes within the cycle to apply RPA for transactional, rule-based processes. Then, he looked to learn from them by using the data collected on them “in a smart way” to adjust behaviour and actions based on experience. This meant injecting AI and ML into it.
ML is applied on data collected on prices and quantity to create algorithms. Peretti explains: “You can learn from it and drive a lot of information to optimize and understand where you have an excess of payment, to increase days payable outstanding and cash flow.”
Related interview: How LafargeHolcim is optimizing its P2P process
Process mining is also used on the data “discovery” to visualize, categorize and represent the data “in a way humans can understand”. He describes it as “an enabler of RPA and IA tools that look at our data to identify which areas to attack and drive first”.
ML is also used to help the company capture the information contained in purchase orders (POs) and invoices and convert it into the ERP accounting system. Human workers, or a large volume of data loaded directly, depending on the technology used, must teach the machine on the first sample of invoices (from 500 to 1,000 approximately) about the suppliers structure of their invoice, such as where the name, tax codes, amounts and line items are located, so the machine can learn how to do it on its own afterward. Then ongoing learning will continue on the machine with the next processed data.
Peretti says: “Machine learning has improved the success rate and the quality and reliability of the information we capture a lot.”
At Tesco Business Services, a branch of supermarket chain Tesco, a similar approach was taken to implement IA solutions on its P2P process.
The business has been on an automation journey and has worked with RPA over the past four years to increase efficiencies in processing the 8.5 to 9 million of invoices the company generates every year in the UK and Ireland alone.
Prashant Shah, head of procure-to-pay at Tesco Business Services, explains: “One of the strategies the business focuses on is the supplier side of the P2P journey. In the last few years, we have used data analytics to uncover opportunities in cash flow and working capital.”
Processing invoices is key at Tesco Business Services. The company uses optical character recognition (OCR) to translate paper invoices into digital records, which allows it to operate at scale with a large number of suppliers across markets and geographies. By applying RPA into the core matching and procurement systems, the business can answer its suppliers as quickly as possible.
The organization also has an electronic data interchange (EDI) framework to communicate with suppliers across the market, with 98 to 99 percent of all invoices being EDI invoices, meaning there is no paper involved.
Related interview: Optimizing the P2P process at Tesco Business Services
Tesco Business Services and LafargeHolcim are not alone in applying evolving technologies to invoice processing. Deutsche Telekom Services Europe, the shared service center that oversees the telecommunications company’s internal transactional tasks in several departments including P2P, struggled with a high workload in accounting.
This was the case for invoices related to non-POs, where no tax or general ledger account information was available, which meant they took a lot of time to process. In Germany, of the 1 million invoices processed per year, 100,000 are non-POs related.
To overcome the challenge of processing these non-POs invoices, a multidisciplinary team of lead accountants, process specialists and data scientists built in-house software SmartFI. Enriched by OCR and a ML algorithm to identify patterns in the data provided by those non-POs invoices, the software has an accuracy level of 99.2 per cent for tax codes and general ledger account predictions. It is also able to check if the invoices comply with German VAT tax law.
Fabian Stadler, data science manager at Deutsche Telekom who led the initiative says: “We are now confident enough to have the machine post fully automated bookings for us […] in numbers, this accounts for 20 percent of all non-PO related invoices. In near-term future that number will constantly grow with every new AI model version.”
To ensure data quality and compliance, a threshold value was implemented in the software to ensure that only invoices answering a certain level of confidence could be booked automatically by the AI software. For the other invoices, the software recommends at least the most probable values to the accountants.
The next section of this report provides recommendations on how to create a long-term and sustainable P2P initiative to deliver strategic value.
How to implement a long-term and sustainable P2P initiative
The strategic value of an optimized P2P process is increasingly recognized. At LafargeHolcim, the P2P initiative was initially spot-based, with P2P work being increasingly implemented in the service center Peretti manages. Gradually, leadership looked to harmonize it with the company’s strategy via a holistic design to improve, automate and standardize it.
Related PEX Network report: Leveraging intelligent automation to link strategy with process improvement
Due to the size of the company and the complexity of Tesco Business Services’ systems, any change conversation on P2P needs to involve all the relevant teams and stakeholders, and the focus is on managing it with suppliers too.
There are change governance counsels and committees to involve all stakeholders in the process and to present change to senior leaders in the business with participation from tech teams to allow the business to look at it holistically.
Forums have been created where conversations with relevant teams such as tech, digital transformation operations, business and global process owners happen frequently to keep everybody notified of any change, as Shah explains.
“We want to ensure we are not waiting for a moment of change and we talk on a day-to-day basis in those forums,” he says. “Any new idea can come from any team and it gets discussed and debated – this helps land the change more seamlessly, otherwise it can become challenging. The more we collaborate on the roadmap of our mission and objective, the more we feel shared accountability as a group. It helps getting an idea spoken about and implemented.”
When it comes to the future of P2P, LafargeHolcim’s Peretti believes the process itself will remain the same, but the way organizations conduct it will become more tech-driven. Fewer humans will do manual work and there will be a great opportunity in vendor portals.
QR codes on invoices are also set to make a difference. Currently, two to three percent of LafargeHolcim’s suppliers use them, making the company adapt its work and IA document recognition tools. Peretti says: “Through the use of QR codes, the exchange of information on the invoice is much stronger but you need to adjust your way of reading the documents with technology, and thus adapt your way of working. When you have made these adjustments, the process is more accurate and straightforward and does not need much human intervention.”
In an ever-evolving world, where the pace of technological pace increases, Shah believes the key will be how organizations are going to be agile to adapt and implement new technologies. This is how they will truly make a difference on P2P innovation and spending control.
This report was originally published on June 8, 2021.
How are you optimizing your P2P process? Let us know in the comments below.