Robotic Process Automation (RPA) promises to reduce tedious manual tasks, freeing employee time for higher value work. But to see the full potential of RPA, companies must understand what it is and what it takes to implement it.
In the tech world, we often find ourselves talking about the latest buzzword, but it’s important to look past the tech speak to understand the concept, what benefits it provides, and what it will take to bring it from bright idea to reality. One of the latest buzzword topics is Robotic Process Automation (RPA).
What is RPA?
Software technology has been available for decades to automate the GUI (Graphical User Interface) interaction associated with business workflow processes in the form of “record and playback” technologies. For many years these techniques have formed the basis of widely used test automation tools. RPA takes business process automation further by using software “robots” to capture the on-screen actions of employees via graphical user interface (GUI), which a Machine Learning Artificial Intelligence (ML/AI) algorithm then analyze, recognize patterns, and turn into software-driven Virtual User scripts. The scripts are then used to automate human processes, such as sending an automatic response to an email, entering and processing forms, or automating work in a company’s customer relationship management (CRM) system.
But human processes are not completed exactly the same way each time; if this approach relied on simple record and playback capture alone, it would simply create an enormous amount of scripts, which eventually would become impossible to use and maintain. That’s why RPA takes advantage of artificial intelligence (AI) and machine learning (ML) to automatically analyze those scripts and recognize common and variable elements, making the suite of scripts lean and re-usable so they can accomplish work efficiently.
Benefits of RPA
RPA holds the promise of faster processes, standardized customer service, and reduced manual tasks and human error. These benefits can trim staffing costs and/or free employee time for higher value work that humans can do better than machines, such as handling one-off processes and recognizing information or circumstances that are non-standard and thus do not fit the use case for RPA.
Companies across multiple industries, including finance, manufacturing, healthcare, and retail are using RPA. They are also seeing positive results. According to a Computer Economics Technology Trends 2019 study, about half of early adopters of RPA saw positive return on investment within 18 months of deployment, and half reported breaking even. Many companies are also seeing improvement in customer satisfaction due to more accurate and faster response times.
Challenges of RPA
While companies are seeing noteworthy initial results, RPA comes with potential challenges. Note that in the otherwise positive report above, half of early adopters could be interpreted as seeing no value. Consider these possible stumbling blocks:
- RPA tools are relatively new. It’s possible to overestimate the capabilities of today’s tools and underestimate what it takes to set them up properly and maintain them long-term. Setup and maintenance of RPA requires application, subject matter, and data science expertise as well as a clear understanding of how a company’s processes and business work. Without this human knowledge and expertise, RPA algorithms may fall prey to a phenomenon known as “overfitting’, with the scripted results veering off track from the original intent.
- RPA requires precise application of AI and ML to create concise scripts that avoid overfitting, which results when robots start excluding less statistically significant data too early. At the other extreme, inclusion of too many workflow variations can eventually lead to the system making incorrect assumptions and result in script “bloat,” which makes the system unmaintainable over time. Having a human in the loop who understands how machines learn is the best defense against both overfitting and bloat.
- Because RPA offers companies the ability to reduce unnecessary human work, some employees may be afraid that it will put them or their team out of a job, and thus may resist or even actively sabotage the effort. While it’s true that some jobs may be replaced by RPA, a successful RPA project requires a shift in both leadership and workforce mindset to embrace consciously the increased productivity realized when people are freed from tedious manual tasks to focus on higher value work. It can also lead to interesting new jobs involving management of RPA suites and programs and in-depth analysis of processes and business functions of their departments.
Moving forward with RPA for your organization
Huge enterprises including Walmart, American Express Global Business Travel, Deutsche Bank, and Vanguard have begun adopting RPA. They are paving the way – showing other companies the benefits of the technology.
It’s clear that RPA offers a glimpse into the way work will be done in the future. At the same time, implementing RPA requires a clear understanding of the capabilities of today’s emerging RPA tools and how to apply them properly to a company’s processes to ensure they remain accurate and manageable.
If your company is in one of the industries benefiting from RPA, we at SQA are well-equipped to determine if and how RPA can work for you. Our method includes:
- Identifying tasks and processes suited for automation
- Determining the tools and resources needed for implementation
- Developing and implementing a “training plan” for the “robot under instruction”
- Calculating an expected return on the RPA investment and the elapsed time to allow before evaluating the results
These can be daunting tasks for a process that too often is advertised as “launch and forget”, but SQA can help.
The SQA team has worked for more than 20 years with the technologies that have come together to enable RPA — including test automation, AI, ML, robotics, and the Internet of Things. We understand how RPA works, how it can fail, how to implement it properly, and how to test it to make sure it works the way it was intended to work — and that it will continue to do so in the future.