Machine Learning Capabilities in Finance Processes
With or without realising it, Machine Learning has been simplifying and automating our daily life for several years now with automated time estimations, recommendations for products, movies and songs, face and image recognitions as well as machines talking to us in our natural language. Even though the potential of such Machine Learning capabilities shouldn’t be a big shock for anyone, companies are still facing quite a lot of challenges adapting them to their business processes. There are several reasons for that: no time, no budget, no resources. A lot of companies that we talk to associate Machine Learning with huge, complicated, long projects and efforts enabling and changing their employee’s capabilities. And it can be. But what if we could make this a little bit easier by offering Machine Learning capabilities that are already ready to use for your business processes? After all, Netflix doesn’t expect their customers to write their own algorithm for personal recommendation of movies. Google doesn’t expect their customer to make their own forecast on when they will arrive somewhere by collecting all kinds of traffic data. Apple doesn’t expect their customers to teach their phone how to talk to them. That is all already integrated in our daily routine.
That is why SAP has been investing heavily in seamlessly adapting machine learning capabilities in the business processes of our customers, especially when it comes to finance. In this article we would like to give you a brief overview of the machine learning services that can help you to automate, enrich and improve your finance processes.
SAP offers a variety of solutions, functions and features to leverage our knowledge of business software and enrich it with data driven insights, especially when it comes to finance.
Before we get more into those solutions, let’s first make sure, we all have the same understanding of what Machine Learning is.
Machine Learning always starts with data, historical data. This could be coming from all kinds of sources, such as machines, transactions, humans, social media, market research etc. In this data we’re looking for typical and reoccurring patterns and events. We use these findings to create predictions, forecasts and initiatives.
In order to get more comprehensive insights on Machine Learning at SAP in general, check out this blog.
As mentioned before SAP offers a wide range of services that enables you to use Machine Learning in finance. Let me start with the AI Business Services. SAP AI Business Services are services that offer machine learning capabilities to automate and optimize processes in order to enrich the customer experience across different business processes. They provide re-usable content for several kinds of use cases. SAP offers solutions for different areas of expertise that can be applied to different business process.
One service in the area of finance that I would like to point out here is the SAP Cash Application. This application automatically matches incoming payments with outstanding customer invoices. Usually companies take rule-based approaches to automate the application of cash, however rule set-up increases implementation costs and rule effectiveness declines over time. Let’s try to make this a bit more tangible with an example.
Imagine a family of six people are all going to the same dentist. This would mean the dentist creates different invoices for each of the visits of those family members. The father (or mother) of the family decides to pay for all of these incoming invoices with one payment. Now the dentist is facing the challenge of correctly matching this incoming payment with each individual open invoice of each family member. This exactly the use case the SAP Cash Application covers. For more detailed information please refer to this blog.
Another service that I would like to highlight is the bundle SAP S/4HANA Cloud for Intelligent Accounting Automation which brings Machine Learning functionality to GR/IR Reconcilation, Intelligent Accruals and Inter Company Reconciliation. Let’s look deeper into the use case of GR/IR Reconciliaton and how it works.
Imagine you create a purchase order with a certain number of items. Then, at some point these items get delivered, which will create a goods receipts with the number of items that actually got delivered. And then, finally, you will receive an invoice receipt from the vendor with the items that you have to pay. These three documents always need to be matched (3-way match) because you don’t want to pay more than you got delivered. Or get delivered less than you ordered. Via machine learning functionalities intelligent accounting matches this information in order to find deviations in them. If a deviation is detected it identifies the root cause including the probability of that root cause. This not only increases the accuracy and efficiency of financial closing but saves the employee a lot of time going through different kinds of documents finding a mismatch and is therefore ultimately supporting the end user in his decision making.
Another service that is relevant for finance is the Document Information Extraction. This is a service used to automatically extract information, such as invoice numbers and amounts out of an unstructured document for further processing. This functionality becomes important for any financial process involving an unstructured document like an invoice coming from a customer, vendor or other stakeholder.
Another way to automate finance processes via Machine Learning, are the Smart Predict capabilities of the SAP Analytics Cloud. The SAP Analytics Cloud is SAP’s strategic analytics platform for all line of businesses and industries. The platform comprises the three key elements of analytics: Business Intelligence, Planning and Predict. This combination makes the SAP Analytics Cloud a valuable and powerful tool.
There is one predictive use case in SAP Analytics Cloud for Account Receivable Payment Forecasting that I would like to describe in particular. In an organization one of the largest assets of the financial statement is usually the Accounts Receivables. With B2B transactions increasing in volume and complexity, poor management of Accounts Receivable can lead to unnecessary write-offs and cash flow problems. This complexity and collection of information about the customer can be very well optimized by adapting machine learning models. If we could predict when customers would pay their invoices, we would have better insight into future cash flow. This would mean that the Collections Managers could better prioritize their time in order to focus on high value invoices which are predicted to be paid late. This could ultimately lead to lower Day Sales Outstanding (DSO), reduced delinquencies, and reduced collections related expenses. In order to get more insights on this particular use case, please refer to this article.
Finance processes in general consist of a high number of manual repetitive tasks that need to be accurate and compliant. Executing these tasks can be very frustrating for the employees who would like to focus on more high value tasks and do their daily job. The result of this can be lack of resources, inefficiencies and delays in processes and a deceleration of digitizing and optimizing the enterprise business strategy and working capital. In order to overcome this challenge, I would like to introduce SAP Intelligent Robotic Process Automation. This solution offers the opportunity to automate tasks in several areas such as invoice management and collaboration and reconciliation of manual tasks during financial closing. Furthermore, it can help digitizing parts of the finance helpdesk operations and automate manual tasks in your account receivables and payable process. SAP offers predefined S/4HANA Finance bots that are ready to use in the SAP Intelligent Robotic Process Automation Store.
One example that is often mentioned is the automated payment advice upload, which can be very well combined with the SAP Cash Application and Document Extraction Service which is mentioned above. One requirement of the matching of payments with outstanding customer invoices as it’s done by the Cash Application, is, that the payment advices are uploaded into the system to then be able to automatically extract the information. How can the payment advice get into the system? Imagine you receive 200 payment advices per day via e-mail in a PDF attachment that need to be manually entered into SAP. Assuming this end-to-end process from incoming payment advice to matching takes about 5 minutes. This means this simple repetitive task will consume 1000 minutes or 16 hours per day, which is equivalent of 2 full time employees. Don’t you think these employees could use their time more efficiently for more high value tasks? Via a bot in Intelligent Robotic Process Automation (iRPA) you can automate checking the e-mail inbox for new incoming payment advices, extract the relevant information out of the PDF attachment and uploading this information in the S/4HANA system in order to even further increase your matching automation. For more inspiration about use case examples please check here.
These are just a few of many examples of how SAP is more and more integrating machine learning capabilities into business processes with providing predefined content. For any more detailed and technical information on any of these solutions please don’t hesitate to contact us. We’re looking forward to your interest!