Robotic process automation: Why IT Ops needs to lead

Robotic process automation (RPA), a type of software that mimics the steps that human beings take when carrying out a task within a process, has been hailed as a crucial element for sparking digital transformation, as an affordable alternative to DevOps and agile investments, and as a key technology for cutting down on expensive, highly manual tasks across the business.

But while most analysts agree that these robots have legs—predicting tremendous growth for the technology in the next five years—there have also been many failures. Here's why IT and business leaders should temper their expectations for the technology, and IT operations should play a key role, engaging with the business to help get deployment right.

Ovum Decision Matrix for Multicloud and Hybrid Cloud Management 2018-19

The promise of RPA: Cost savings

Analysts at Market Research Reports expect the RPA market to grow at a rate of over 25% annually through 2021, while another recent study conducted by Horses for Sources (HfS) and KPMG shows that at least half of organizations in seven of nine major verticals plan to make significant investments in RPA within the next year.

What's driving the predicted growth? Most enterprises are looking at RPA as a way to cut costs, eliminate manual errors, and free up the humans to do more brain-intensive work.

"RPA has revolutionized how organizations run their back-office works to automate business tasks that are often mundane," explains Sachin Garg, associate director of electronics and semiconductors for market research firm MarketsandMarkets.

"The biggest benefits of deploying RPA are seen in process centers that have large volumes of repetitive processes, such as data entry and migration, invoice processing, loan processing, and claims processing."
Sachin Garg

Unlike something such as IT service automation, RPA isn't aimed specifically at automating IT service delivery tasks or software delivery tasks. Instead, most RPA products target the business, providing an automated platform to help workers to easily set repetitive, manually intensive work on rails. Targeted tasks are often trapped within legacy systems that aren't going anywhere and are completely enmeshed within the organization's workflow.

This could be anything from tasks necessary for analyzing financial data to processing human resources documents to routing a constant influx of inbound customer emails, says Jakob Freund, co-founder and CEO of workflow automation tool vendor Camunda.

"RPA has a well-defined sweet spot: automating specific tasks for organizations that are stuck with legacy software that doesn't offer an API."
Jakob Freund

In the last five years, RPA has been primarily used for "old-school" screen scraping, which mimics human behavior on the front end, with the notion that you don’t have the underlying application to enhance, says Vishal Awasthi, CTO for Dolphin Enterprise Solutions Corp. "The installation tries to mimic the sequence of actions that the user takes."

"RPA was primarily designed to not remove the need for the humans to work, but to make it faster."
Vishal Awasthi

RPA's role in software and services delivery

While RPA isn't an IT operations play per se, it plays a part in the overall delivery of IT software and services. It's at the heart of the low-code movement, which some savvy CIOs are using to ease some business pressure for long-term DevOps and agile transformations to deliver immediate and sweeping results. RPA can help tech leaders deliver automation to monolithic legacy systems while incrementally moving toward a more widespread microservices software delivery model.

"Organizations should realize that the RPA approach is a temporary remedy, not a sustainable solution. Eventually, it is inevitable to replace the respective legacy applications with a modernized infrastructure."
—Freund

For example, RPI is not needed when replacing monolithic applications with microservices.

There is a reason why you don't usually see splashy RPA case studies coming from digital-first companies such as Google and LinkedIn, he says. They've already got processes in place to quickly deliver automation directly through their continuously delivered software stack. But for the rest of world, RPA can be an effective tool for inserting automation into parts of the software portfolio that won't be seeing any DevOps love for a long time yet.

Bringing this understanding of the transitional role of RPA will help organizations make better decisions about where and when to use it, rather than trying to overdeploy it because it's hot right now.

"The word robotic has a connotation that sparks the imagination and perhaps obscures what RPA really does best."
—Freund

What this means: You might be disappointed or have missed expectations when trying to use RPA to solve a problem it’s not suited for, Freund says.

How to avoid failed RPA projects

Like any newly hyped automation technology, robotic process automation has certainly had its fair share of disappointments in recent years. According to estimates by EY, RPA deployments have a 30% to 50% failure rate.

There are myriad reasons for this, but one of the most common is that organizations expect to wave RPA like Harry Potter's wand on fundamentally broken processes.

RPA isn’t magic, says Awasthi. "A broken process—whether done by a person or RPA—is still broken. Organizations need to develop a process roadmap to figure out areas that need to be streamlined, and then leverage RPA wherever it makes the most sense."

Organizations should be able to define the processes up for automation to the keystroke level, and they need to put in place standards for architecture and logic creation in order to ensure success, says Mark Davison, global research partner at Information Services Group (ISG).

"If a process requires complex or fuzzy logic, it is best to wait to introduce RPA until after the company gains confidence with the technology with other less complex processes," he says.

Otherwise they risk running into what Awasthi calls the "dark side of RPA."

"For the classic, rules-based RPAs, if the rules get too complex, or get set up by business analysts not familiar with all the nuances of the processes, RPA can produce incorrect results and put companies at risk," Awasthi says. "This is a classic case where you gain efficiency at the cost of effectiveness."

Why IT Ops should deploy RPA

This is where IT operations can provide value in the process of deploying RPA. Yes, it is a business-centric tool that's ideally meant to help democratize the creation of automated processes out to the nontechnical workforce. But the platform itself is best optimized by IT, and business users will do best if someone trains them on how to use the tool.

Unfortunately, the divide between IT and the business often presents a bit of an RPA deployment paradox for organizations, Davison explains.

“On the one hand, an IT organization that has achieved a level of maturity in these processes is well-positioned to succeed with robotic process automation," he says. However, the RPA tools are often used by the business and not IT. And the business is most likely not trained in the software development lifecycle. "There is a need to train the business on the same process, which is called the ‘automation lifecycle,’" Davison says.

Organizations that have maturity around process discipline are the ones most likely to succeed with RPA, Davison explains. This is a key refrain taken up by analysts and consultants who are five years into the nascent RPA movement, writes Phil Fersht, CEO and Chief Analyst at HfS Research:

"This isn't simply a case of buying software and looping broken processes together to remove manual efforts. This requires real buy-in from IT and operations leaders to invest in the technical, organizational change management, and process transformation skills."
Phil Fersht

Doing RPA? Success requires a disciplined approach

Even though RPA is a transitional technology, it takes discipline to use it well. As McKinsey consultants Alex Edlich and Vik Sohoni explained last year in an article on recent RPA failures, this solution is just another hammer in the IT toolbox, and you need to plan for its use alongside everything else the IT team uses.

"A bot is a tool in a toolkit, just like self-serve tools, work-flow tools, lean-process maps or six-sigma methodologies. Companies need to apply these tools as part of an orchestrated action, not in isolation."

Hybrid Cloud: New Challenge For Monitoring Solutions