The SAP Cash Application enables an intelligent incoming payment processing powered by machine learning.
The SAP Cash Application learns matching criteria from payments history and automatically creates matching proposals for incoming payments. It takes advantage of machine learning to enable an effortless setup and to increase automatic matching rates.
The SAP Cash Application provides the following key points for your business:
- SAP Cash Application is a cloud service that is tightly integrated with the SAP S/4HANA Cloud for Contract Accounting and Invoicing.
- SAP Cash Application contributes essentially to the Total Cost of Ownership reduction because it provides higher automation rates than the classical approaches without any configuration effort. It boosts efficiency of the traditionally very labor-intensive clearing process, being executed with nearly no user interaction.
- This higher automatization level allows you to focus more and more on strategic tasks for your business.
The historical clearing information is sent to the SAP Cash Application to train the model and derive matching criteria. Training happens regularly to ensure changing behavior is captured so the model can adapt.
When new bank statements are received (most times on a daily basis), those not processed by the standard rules will be sent to the service along with the open receivables so the machine learning model can infer matching proposals.
The Machine Learning approach can capture much richer detail of customer- and country/region-specific behavior, without the costs of manually defining detailed rules.
Proposals are returned to SAP S/4HANA, and those that meet the configurable confidence threshold are automatically cleared for full automation. When there are multiple proposals for a payment, they're presented to the accountant in the appropriate SAP Fiori app.
The machine learning process is able to reduce the number of items that fall to clarification cases.
The machine learning process carries out the following steps:
- Payment lot items are created from bank account statements.
- The items that do not correspond to rule-based matching fall to clarification cases.
- Clarification cases appear in clarification worklist.
- Machine learning creates new proposals after analysis of memo lines.
- New proposals can be detected with confidence level (based on criteria). They are cross-referenced with open items to further increase proposal accuracy.
Let's assume you've defined the following values in the SAP Cash Application configuration:
- X1 = 90%
- X2 = 70%
- X3 = 50%
The system applies these values to the following examples:
Example 1: Proposal with a confidence rating of less than 50.
The system ignores the proposal. The clarification case has to be processed manually.
Example 2: Proposal with a confidence rating between 50% and 70%.
If you select Adjust Confidence Rating for the Process Returned and Reserved Clarification Cases job, the system checks whether the proposal can clear a receivable.
If the receivable can be cleared, the system increases the confidence rating to 70% and proceeds as illustrated by example 3.
If you didn't choose this, the system ignores the proposal as illustrated by example 1.
Example 3: Proposal with a confidence rating between 70% and 90%.
A proposal with a confidence rating between 70% and 90% is displayed in the Clarify Incoming Payments app.
Example 4: Proposal with a confidence rating of 90% or more.
If you select Adjust Confidence Rating for the Process Returned and Reserved Clarification Cases job, the system checks whether the proposal can clear a receivable.
If the receivable can't be cleared, the system reduces the confidence rating to 70% and proceeds as illustrated by example 3.
If you didn't choose this, the proposal can lead to an automatic clearing if you choose Clear Automatically for the Process Returned and Reserved Clarification Cases job.
The machine learning training needs the following input data:
- Master data such as business partner and contract account number, name and bank data.
- Historical payment data such as electronic bank statements, note to payees, payer bank account data and finally all available payment lot items.
- Customizing settings such as the selection category and the selection field name.