Cash application is a part of the accounts receivable process that involves matching incoming payments to their corresponding invoices and client accounts. While seemingly straightforward, the process for a large, global and diverse organization like Accenture is complex and performed at huge scale.
Accenture issues more than half a million client-facing invoices each year from 200 locations globally. Every year that volume increases and is projected to increase exponentially every five years with Accenture’s organic and inorganic growth. Our SAP S/4HANA® system was enabled to successfully match transactions, but the “hit rate” proved to be much lower than expected. A rule-based method requires exact matching of information and time-consuming manual maintenance due to extensive location-specific configuration settings. To complement the technology solution, we leveraged our cost-effective locations, driving results that got to the right outcome but were significantly more time consuming and less efficient. For a rapidly growing enterprise we knew there was a better way.
When tech meets human ingenuity
Accenture Finance’s vision and strategy are to find new ways to deliver value, and one of those ways is in moving to artificial intelligence and machine learning (AI/ML). The introduction of SAP Cash Application offered the potential for change and a leap in capability through a machine learning-enabled matching model.
SAP Cash Application passes new incoming payment and open invoice information from SAP S/4HANA to a matching model on SAP Business Technology Platform (BTP). The machine learning model suggests a proposed match for review by cash application processors, or automatically clears the open item, depending on business needs. The solution offers the potential value of freeing cash processors and clearing cash payments faster, resulting in a far more efficient solution and freeing up our business runners to far fewer administrative interruptions.
Rapid prototype and pilot
Accenture’s Global IT organization and Finance teams collaborated to share our understanding of SAP systems and deep functional knowledge to assess the new offering. Because SAP Cash Application is a Software-as-a-Service product, the technical effort to enable it is low. Given this, we decided to perform a rapid prototype and pilot deployment.
Our Global IT team enabled machine learning proposals for cash application activities for five of Accenture service center country locations, a sampling of environments with varying matching capabilities. For an open bank statement item, SAP Cash Application suggests a corresponding open invoice as a match. The user is given brief information on the reason for the selection. The user then has the choice of accepting or rejecting the proposed match. If the match is accepted, the software automatically applies the cash. This capability changes an entirely manual process to a single automated click. If rejected, the process reverts to manual activities.
The pilot brought forth initial positive results as well as product maturity aspects to address. Given that Accenture is conducting one of the largest global rollouts of SAP’s Cash Application in the world, our Global IT organization participated in SAP’s “angel development” process. Accenture is also among the first companies to test SAP’s next generation of machine learning model, deep learning. We worked with SAP to co-develop the product further to address the needs of a large, complex, global organization. The effort was a demonstration of combining technology with human ingenuity.
"We are setting a new standard for Accenture by moving from a rules-based system to a deep learning model with SAP Cash Application."
Accenture brought to light the specific needs of large enterprises. Our combined Global IT and Finance team identified issues or enhancements to make on regular calls with SAP developers who addressed the new input in agile development sprints. We co-developed localizations and enhanced capabilities with SAP in the following areas:
Removing duplicate internal customers included in transmissions to the cloud. Accenture identified that the solution was considering client accounts that were not relevant to the cash application process in the logic of the model. This action caused the system to needlessly match on those items and send them to the cloud, also skewing the algorithm. SAP recognized this and released the standard ability to exclude non-relevant customer types.
Maturing a lockbox solution. Lockbox is a check processing solution used solely in the United States and Canada. Accenture is actively working with SAP on maturing a lockbox machine learning solution and will be the first customer to pilot it. Accenture completed a successful technical proof of concept in late 2021 and is planning to deploy the solution into production in 2022.
Adjusting intra-country, cross-company code transactions. Accenture was unable to roll out SAP Cash Application to India and the Philippines where Accenture may have multiple country codes for legal and tax purposes that represent a country but receives all payments through a single company code. These codes then need to be applied to the different entities. We raised this need as a requirement and are currently co-developing the model based on Accenture’s system.
Flagging duplicate bank statements sends. Accenture processes bank statements around the clock globally. SAP Cash Application lacked a way to recognize the receipt of a bank statement, resulting in multiple bank statement information re-sends to the cloud service, causing overpayment of line items transacted. SAP recognized the requirement and made the exclusion of duplicates a standard solution.
Enhancing reporting. Accenture participated in workshops to provide extensive input on desired reporting items as well as feedback in the development of SAP Model Manager, a self-service reporting tool. Accenture is the first company to onboard to the reporting solution and continues to provide feedback.
In parallel to the co-development work with SAP, Accenture continues to progressively roll out SAP Cash Application to service center locations. We also continue to co-develop with SAP on requirements needed.
"The collaboration of our Global IT organization and SAP on the development of the technical solution combined with our cash application team’s process knowledge is what led to the right result for the business."
— RONALD STEVENS, Managing Director – Controllership Capability, Accenture
A valuable difference
An efficient accounts receivable process of matching payments to invoices is important to every business, including Accenture’s. SAP Cash Application is projected to help Accenture automate and make the matching process faster and more efficient with fewer errors. Also important, the solution is moving Accenture from a rules-based system to a machine learning model that learns from historical data patterns and user behavior, and grows natively.
Gaining time from the automation of a manual cash application process enables more focus to be placed on receivables management tasks such as contacting more customers. Additionally, a machine learning solution becomes smarter and better over time in identifying and matching new payments with existing open invoices.
Accenture is leading the way in gaining an in-depth understanding of SAP Cash Application for large, global organizations. At the half-way point of our implementation timeline, we are attaining 63% correctness in our machine learning proposals—which is double our original baseline—and are experiencing improved speed and accuracy in matching.
Our involvement with SAP Cash Application enables new customers to benefit from the feedback provided and work performed that helped mature the product. Co-development outcomes are now standard, making SAP Cash Application with features that are more robust and suitable for large enterprises. Overall, these efforts are anticipated to help give users confidence in the solution’s modeling capabilities.
Accenture is moving from extensive manual activities to increasingly a one-click process. Our goal is to roll out the proposals to the remaining service center locations and help build confidence in the solution to facilitate the auto-clearing capability. The machine should automatically match the items and post the clearing document, removing even the one-click step in most cases, allowing the service center cash application teams to focus on higher value-added work.
Benefits to date:
Greater accuracy in applying cash daily
Faster application of payments that in turn reduces open accounts receivable balances
Lower unapplied cash balances
Prevention of capital charges
Reduction in peak time processing—month end, quarter end and year end
Increased speed on collection process
''The shift to machine learning is transformational and will help prepare for Accenture’s projected growth in invoice and payment volume. It has also given us the confidence to consider machine learning and AI for other processes.''
— CARSTEN POULSEN, Managing Director – Finance, Transformation Technology, Accenture
Meet the team
Managing Director – Finance Operations Controllership Lead