Robotic process automation (RPA) is a form of business process automation that combines traditional software user interfaces with optical character recognition, machine learning, and artificial intelligence (AI) to help automate the mundane and routine tasks in your business processes.
Most of the technology used in RPA is not new. Rather, RPA technology is the combination of existing technologies into easy-to-use automation platforms. Bringing together the suite of tools has created a new class of automation that can be adopted across industries.
At the most basic level, an RPA task can be likened to a next-generation macro. Like macros, RPA allows you to assemble a collection of actions into a sequence that automates a business process task. Often the tasks transform a set of data into a standard output through calculations and other data manipulations.
Applications of RPA can also be much more complex. One example might be automating an accounts payable solution, where RPA handles a number of associated tasks:
- Setting up new supplier accounts
- Receiving and recording invoices
- Tracking payment due dates
- Generating recurring reports and data analytics
RPA differentiates itself from software macros, however, through its broad scope of interactivity. Interactions can occur across disparate applications with nearly any computer software system. RPA software is capable of interfacing with applications just as a person would. Traditionally, interacting with applications required an application programming interface (API), but RPA can use multiple methods to interface with applications, including a graphical user interface. Websites, legacy applications, mainframes, scanned documents, emails, and more can be used in the business process making up an RPA task.
Leading RPA platforms are advancing the core functionality of RPA software by integrating AI and machine learning. Use of these technologies can create solutions for understanding and manipulating your business data. Some estimates put upward of 80% of business data as either unstructured or semi-unstructured. Without structure, data becomes practically unusable. A transformation to structured data is necessary to extract useful insights, which you then can apply to business decisions.
The ability to transform unstructured or semistructured data is the key differentiator among RPA software alternatives. The AI algorithms built into innovative RPA engines allow the software to understand the language used on different document types, such as invoices, purchase orders, bank statements, and other common business documents. RPA with built-in AI capability uses this understanding to transform data from source documents into business actionable data. AI-augmented RPA bots can classify documents into groups, and then extract meaning from the semistructured data present in documents. The data is then processed into standardized data sets for use in your business process.
Should an AI-enabled RPA bot fail to process a document, the document is sent to a human operator for further processing. During the employee’s review, the bot watches the human interaction with the user interface. Observing how a person corrects an item trains the bot to handle similar exceptions in the future. Over time, the bot increases its “confidence” in how to process documents. The increased confidence decreases the number of help requests that are sent for human review. Machine learning allows bots to become better workers as experience accumulates.
RPA technology creates efficiencies in rules-driven, repeating processes. The automations are designed to handle the work using defined action sequences. But, as the saying goes, all rules have exceptions. Business process rules are not different. Because predicting and handling 100% of the possible scenarios is hard and an inefficient use of development resources, RPA bots should be designed to allow for exceptions. Bots handle the routine and expected cases, and deliver exceptions to human operators for resolution. In this way, bots transform the work that people are performing. Employees are removed from handling boring, repetitious tasks. Instead, they focus on resolving issues of unusual circumstance that require thinking and judgment.
Advantages to Automation
As with other forms of automation, the inherent speed and efficiency of machine processing gets work done faster. Automation can also scale to accommodate increases in volume. For a growing company, this can mean not having to add additional head count to handle the extra paperwork that comes with expansions to meet customer demand. Alternatively, scalability could mean being able to audit more documents to find exceptions to defined business rules.
Greater traceability is another advantage. Every action a bot takes can be logged and controlled through strict codified rules. This feature makes all bot activity easy to monitor, trace, and audit. When good software governance practices are applied to developing and deploying the RPA bots, business leaders can expect consistent and reliable results from their automated tasks.
RPA helps manage risk, too. Limiting workers’ access to sensitive information is often a key control in business processes.
Incorporating RPA into the workflow can reduce the possibility of data exposure by limiting access to sensitive data even further. RPA can also manage risk by being a control that provides continuous monitoring and audit of high-risk transactions and processes.
In the accounting field, there are many processes where the required tasks are repetitious and rules-based. The nature of the work provides many opportunities to employ RPA technology to reduce the amount of time people spend on basic data processing, allowing CPAs to work on more strategic and impactful decision-making tasks.
Accounts payable and accounts receivable (AP/AR) automation is an area where accounting departments are currently adopting RPA and seeing a high return on investment. RPA software is well-suited to processing the variety of forms used in AP/AR workflows. Transforming the semistructured data on nonstandard forms used when working with many different vendors is a task in which RPA excels. Data captured from forms by RPA become standardized input to your enterprise accounting system.
Other areas of early adoption often focus on the high-volume and manually intensive tasks that have measurable cost savings or provide timelier task execution. It is in these areas that management will see the value in RPA and commit to the investment. Accounting areas where leadership will want to invest include the following:
- Compiling financial reports
- Managing complex billing processes
- Reconciling bank accounts
- Detecting fraud
- Tax compliance
As adoption of RPA grows, use of the technology will spread to areas where risk mitigation plays a larger role than just a cost-savings return on investment (ROI). RPA use in financial controls and audits can improve financial oversight, but impact on company financials is hard to measure. This may be why implementation here tends to fall to the bottom of management’s priority list.
Initial ROI calculations play an important role in the initial decision to begin adopting an RPA strategy. The first-year cost to establish an RPA program can be a deterrent due to licensing, development, and infrastructure expenses. Looking beyond year one, though, the rate of return improves greatly as maintenance costs will be significantly lower than year one’s development costs. Enterprise software has an average expected lifetime of about six to eight years, so any developed RPAs can be expected to be in service for the same length of time.
The real ROI for RPA comes when you implement automation across a number of processes. Increasing the number of automated processes allows for spreading out the fixed costs across multiple projects. In doing so, more automation projects become viable because of improved ROI calculated values. Once employees see and understand the power of RPA, they begin to see the use cases where automation can improve company performance.
As RPA replaces mundane data processing and entry tasks, accounting activity will shift to higher value analysis and design work. Automation will handle much of the transactional accounting work that people perform today. With this shift in job responsibilities, work typically assigned to entry-level employees will largely be eliminated.
As time passes, a deep understanding of processes will tend to fade due to the lack of personnel performing the work on a daily basis. Companies will need to ensure that their process workflows are well-documented and maintained. Otherwise, the nuanced knowledge of the automated accounting processes could be lost. Companies will need to ensure that new accountants receive training on the underlying process to gain the kind of insights needed to develop and enhance accounting and audit programs.
Future CPAs will be focused on the business impact of the accounting data. RPA will provide an increasing amount of structured data about businesses. As a result, a CPA will have access to more up-to-date financial data about his or her company’s business efforts than ever before. Understanding and using the data to support management decisions will become a key responsibility for CPAs. The accounting profession itself will become much more focused on data analytics and reporting. CPAs will be asked to develop predictive data analytics to forecast future needs and trends within the business. In turn, this will aid management in running performance programs based on financial statistics and trends.
Use of RPA will lead to changes in how financial records are audited, too. Current practices use statistical risk-based sampling for auditing transaction documents. For high-volume transactions, audit test procedures may cover as few as 5% of the population. By incorporating RPA into the testing process, the enhanced speed and efficiency of the automation will allow for a greater amount of audit testing. Initially, larger audit firms will promote their increased testing assurance as a way to win engagements from audit firms that are not using RPA. Eventually, auditors should expect standard testing allowances to rise as automation spreads through the industry and becomes a regular tool in conducting audit test procedures.
CPAs will continue to be valued for the business insights they can provide. The interpretation of complex business laws and regulations has been and will continue to be a core skill for CPAs. Subtleties in language and the meanings of tax and business laws will defy much of the codification necessary for machines to understand them. Business leaders will continue to rely on CPAs as trusted advisers when making financial decisions.
RPA will not replace CPAs, but it will transform the accounting role. The digital world is overflowing with data. As businesses adopt RPA to transform their terabytes of data into meaningful, business-focused, structured financial information, CPAs should expect to be asked to put greater meaning to the numbers.
Don't miss the podcast on this subject.
Adam M. Costa, CISA, CCSP, is robotic process automation solution architect for Schneider Downs & Co. Inc. in Pittsburgh. He can be reached at email@example.com.