[INTERVIEW] Prof. Dr. Mike Schulze, CBS International Business School | Robotic Process Automation in Controlling - Results of an empirical study

The interview was held with Prof. Dr. Mike Schulze, Vice Dean and Professor of Controlling, Accounting and Financial Management at CBS International Business School, Mainz, Germany. You can learn more about his helpful and up-to-date insights into the use of robotic process automation at the 2nd Finance Conference, which will be held online on Tuesday, October 27, 2020.

Can you briefly describe what is behind Robotic Process Automation (RPA)?

RPA is a term that has recently begun to spread widely and receive broad attention. It is referred to as a technology which is mimicking human behavior on computer systems. It is software-based and can act like virtual employees that perform tasks which have three main characteristics: The associated processes are manual, repetitive and rule-based. The use of RPA aims at the automation of such processes. Automation robots use structured data as input and they interact with different systems and software.

Can we categorize RPA software tools?

Two specific types of RPA software tools can be distinguished: click agents and dialogue agents. Click-agents are using the same interfaces that any human user would use. They are therefore typically operating on Microsoft Windows, opening, closing, maximizing applications in their respective windows. They log on to applications, search for or filter information, they cut, copy, paste, input, type and switch between contexts of business applications and internet browsers. They also take decisions based upon rules that their companies devised before, just as humans would. Dialogue agents are agents that the user can engage with in the same ways in which they would engage with co-workers. A currently very typical approach is to set up chatbots. These bots are most commonly either text-based or natural language-based. Text-based chatbots can be found on client facing websites very commonly already these days. Natural, spoken language-based agents are found in the services of Amazon’s Alexa, or Google’s Google Home. Dialogue agents are therefore an alternative way of querying databases. In extension, robotic processes can be connected to other applications and in this course also trigger processes, save or update information that is stored in IT systems.

What are the main elements of RPA tool architecture?

Although the structure of RPA software differs from provider to provider in detail, a comparable basic architecture consisting of four components can be identified: a developer component, a process recorder, the actual software robot, and a monitoring and control component. Depending on the specific RPA provider, the naming of those components might be different.

The developer component is typically a graphically assisted workplace where workflows can be designed, whose elements represent not only actions the bot must take, but also elements of the graphical user interface of the operating system that would commonly be operated by a human user. The second type of component is the process recorder. The process recorder allows humans to “show” to a program how they operate their computer while executing a process. After the click of a “record button” they start opening the applications they need, logging in to them, searching for files, reading or writing attributes, saving data, etc. These recordings may be used not only to document processes in a level of detail that few companies are able to present. In addition, in some RPA applications they serve as blueprints for editing the automation process in the developer component. This way, the modeler must only focus on variables and executions and the process of modelling is starting from a much higher level of detail already defined. The result of the modeling in the developer component is a software robot. The robot can either be installed on a single desktop (desktop client) or on be held on a central server (application server). Whether the robot is running on a central server or a single desktop, it usually accesses the application level and not the data management or data storage level, although this is technically possible. The monitoring and control component is found in RPA solutions that are aimed at enterprise-scale deployment. Such modules manage multiple bots and commission work packages to them. One may think of them as a sort of team leader for digital worker robots. It is often also the hub for operational reporting and the first place to go to when a human automation administrator wants to gain an overview over the state of affairs.

What are the typical RPA uses in finance?

Simple use cases comprise the collection of data from various sources, the grooming of base data as well as the compilation of lists and low-complexity reports. Typical use cases in that regard are for example:

• attribution of credit card positions to travels in travel reimbursement processes,
• booking of incoming invoices,
• review of open orders,
• mail-to-ticket-processes in service centers,
• mail attachment extraction,
• master data updating/integration, and
• duplication checks.

For controlling especially the report preparation offers high potential for use cases, all the preparatory work that a controller normally has to do, such as data collection, merging data from different sub-companies, group consolidation of the data and to some extent standardized commenting, all this preparatory work can certainly be done well with RPA.

Complicated cases consist of the same identified building blocks, but display increasing lengths and may include communication with humans. For example, such cases implement the use of email or software notifications, resulting in robots that have to wait for human input. What makes use cases complicated is not the number of sources or output formats, but rather the number of rules and exceptions to rules to obey. Another driver of complications is the integration of human feedback into the automation process.

What are the potentials of using RPA in controlling?

Experts especially see a high degree of usability of RPA in the controlling main processes management reporting, planning, budgeting and forecasting, cost accounting and data management. In management reporting the report preparation offers high potential for automation, the same applies for data enrichment and data quality assurance in data management and the summary and consolidation of individual plans in planning, budgeting and forecasting.