Many enterprises have been exploring Robotic Process Automation (RPA) as a way to automate functionality. According to Grand View Research, Inc., the global RPA market is anticipated to expand at a CAGR of just over 40% by 2027, thanks largely to increasing demand for Business Process Automation (BPA) through the use of Artificial Intelligence (AI) and software robots.
IT organizations are interested in RPA because they want to quickly automate their most-needed legacy functionality at a low cost. But is RPA a good strategy for modernizing legacy applications to deliver value over the long-term? Or is it a shorter-term, surface-deep solution that automates select functionality, while leaving the spaghetti code and security vulnerabilities of legacy code to cause problems down the road? To answer that question, let’s first define RPA and explore its limitations.
What is Robotic Process Automation?
According to the Association for Intelligent Information Management (AIIM), RPA consists of “software tools that partially or fully automate human activities that are manual, rule-based, and repetitive.”
Essentially, the goal of RPA is to replace labor-intensive, repetitive tasks formerly performed by human knowledge workers and instead complete them through software robots. Deloitte states that RPA solutions can reduce costs, decrease cycle times, improve accuracy, increase throughput, and improve employee morale.
There are some obvious advantages of replacing humans with bots for repetitive tasks:
You can better utilize your human knowledge workers on more complex tasks
Bots work 24/7/365 while humans go home at the end of the day
Humans have an average of 60% productivity if making few errors, while bots can be 100% productive with zero errors (usually).
Where Can RPA Work Well?
Deloitte notes that “Robotic process automation tools are best suited for processes with repeatable, predictable interactions with IT applications. These processes typically lack the scale or value to warrant automation via IT transformation. RPA tools can improve the efficiency of these processes and the effectiveness of services without fundamental process redesign.”
AIIM also defined the types of applications suited to RPA as well as where it could run into significant limitations:
“…handling structured data in specific formats. As the variety of input formats increases (e.g., different invoice formats from suppliers), the effort required to train, deploy and maintain multiple robots can increase.
…basic exception processing but is not the right tool for complex exception processing.
…standardized processes. If the processes change frequently or if there are multiple variants of a process, RPA also needs to be constantly tweaked.”
Based on these parameters, there are a number of business use cases for which RPA might be beneficial, including Data entry and validation, data import/export and upload/download, data consolidation, automated formatting, message creation, and web scraping.
But it is important to consider the complexity of the function you are automating. What system(s) must the RPA access in order to accomplish its task? Are these systems full of technical debt that will limit scalability and flexibility? Is the data the RPA needs structured or unstructured? How complex is the process and is it likely to create a lot of exceptions?
Is RPA a Comprehensive Solution or Short-Term Band-Aid?
In our view, RPA has its useful and proper place, in limited use cases. But it is NOT a near-term or long-term strategy for migrating or modernizing legacy applications.
Gartner predicts by 2021, 50% of RPA implementations will fail to deliver sustainable ROI absent their combination with other solutions.
Although the surface-level automation may scale quickly to meet your demands, the underlying core technology remains constrained by legacy code—especially in monolithic applications with lots of technical debt. With legacy enterprise systems, in most cases, RPA simply kicks the modernizations can down the road. Security vulnerabilities? Not eliminated. Spaghetti code? Still piling up.
To the customer, you can present a surface sheen of digital experience, but internal business processes can still be a real mess. For example, using RPA to scrape your legacy UI and render a screen in an iPhone app does not mean you have created a modern, interactive digital experience. If that mobile interface still pings a legacy system that is neither resilient nor scalable for the level of service your customers expect, the result might be unavailable data and an error-filled, clunky user experience.
EY estimates 30-50% of initial RPA projects fail. Some of the most common reasons they’ve observed for the failure of RPA projects include trying to use RPA on highly complex processes, failing to create an adequate IT production infrastructure to support the automation, and assuming that RPA is all that’s needed to achieve good ROI.
Common limitations of RPA are found around change management, performance/scalability, and legacy technical debt. At first, RPA may seem to represent cost savings over doing the real work of modernizing legacy platforms. However, if those underlying legacy systems create performance bottlenecks as the core applications fail to keep up with the skyrocketing API calls and data requests from the RPA microservices, customer experience will suffer. In the long run, you could face higher costs, less satisfied customers, and lingering technical debt/scalability limitations.
Accelerated, Deep Modernization Using Automatic Microservices Extraction Technology
Enterprise IT organizations need an approach to transformation that offers real depth of value by reducing technical debt while maximizing performance and scalability. Instead of applying RPA over obsolete underlying applications, it would be better to identify the important, discrete functionality inside those monolithic legacy applications and apply automation to transform that vertical slice of legacy functionality into a modern microservice on a digital platform.
That is exactly how the new microservices extraction capabilities in our Synchrony Modernization Lifecycle Platform (MLP) works. Companies can gain the best of both worlds, by fully modernizing the code base for individual microservices without a heavy investment in programmer hours for comprehensive line-by-line recoding. Companies are free to cherry-pick the most critical legacy functionality and migrate those discrete services to cloud-native architectures.
The result is deep, rich, entirely scalable digital functionality at a low total cost of ownership. In this way, it is possible to speed digital transformation in an affordable way while retiring/decommissioning non-essential code along the migration journey. By preserving business-critical functionality while simultaneously reducing technical debt inherent in legacy applications, microservices extraction delivers long-term value in many situations where RPA cannot.
Yes, I’d like to learn more about Synchrony Systems microservices extraction technology: