Here’s Why RPA Fails to Meet IT Expectations

Robotic process automation (RPA) tools are great at automating rule-based tasks. Yet despite the hype, RPA projects often fail. Here’s why RPA fails at times.

Written by . Last Updated:
RPA projects have a success rate of about 50%, due in large part to the inflated expectations of business leaders.

When cloud computing first became popular in the early 2010s, early adopters expressed disappointment –and a fair bit of regret– that the new technology was not living up to its hype. Here’s why RPA is caught in a similar hype cycle.

The market for robotic process automation is surging. In 2019, the market grew over 60%. 2018 saw similar figures. 2020 is expected to be another boom year for RPA providers as organizations double-down on digital transformation initiatives in order to adapt to pandemic-related issues and economic uncertainty.

Much of this should be expected –RPA removes tedious, manual tasks from people’s day-to-day working lives, giving them much more time to spend on higher-value cognitive tasks that require creativity and forward thinking. At least, that’s how the marketing goes.

Over the last two years, analysts have advised prudence around RPA, hinting at something that often gets drowned out in the hype: a lot of RPA implementations fail. According to Ernst & Young, up to 50% of RPA projects fail.

Common Problems with RPA Projects

Problem #1: Brittle scripts

Robotic process automation tools are great at automating routine, rules-based tasks. But the RPA software isn’t actually a robot, it’s a software script that is programmed to execute instructions that are narrow in scope. 

Additionally, and just as crucial, the RPA tool is operating at the UI level –instructions might include inputting data into another piece of software, or pulling names from an Excel sheet. If something happens to the UI, for example if the software is updated or someone adds a new column to Excel, then the RPA script will stop running or return improper data. 

Problem #2: Inflated expectations

RPA use cases require planning, research, and a solid understanding of the tasks RPA will be automating. Quite often, after a task is automated, it’s discovered just how dynamic the process actually is (especially when multiple applications are involved). This can include variability in how employees fill out forms that RPA solutions must extract data from, or common mistakes such as using the wrong data format, or changing the name of a file.

Organizations have high expectations that are often inflated by marketing hype. This isn’t to say that organizations don’t find success with RPA –they certainly do use the tools– but that the ultimate goal of pervasive automation (“a bot for every desktop”) is frequently unachievable. The organization overestimates how many of its processes are suitable for RPA, and underestimates how much work is required to fine-tune process rules. As a result, key milestones such as cost savings never materialize and political buy-in dissipates.

Problem #3: Uncertainty at scale

RPA tools require no coding skills. They’re designed for non-technical employees in business functions who need to automate basic tasks. They then automate as many tasks as they can, without realizing how brittle the software robots are, and how prone to variation their tasks are. 

RPA tools offer few tracking capabilities, making it difficult to determine who created what bot, or what programs and datasets those bots depend on. So when RPA tools are hastily implemented at scale across the organization, it becomes impossible to predict what processes are going to stop functioning next.

What are the Disadvantages of RPA?

Disadvantage #1: Codified inefficiencies

Robotic automation projects intended to reduce process times or to increase the reliability of tasks often fall short of these goals. This isn’t the RPA tool’s fault, to be clear –the problem is that the process itself has always been complex and poorly designed, and task automation alone isn’t going to fix that.

RPA is most common outside of IT, where it’s deployed by non-technical employees who do not always have a deep understanding of the full process or task that is being automated. Once the process is automated, it still takes too long to complete because it is overly complex and includes additional, unnecessary steps.

The disadvantage here is how those processes are approached. Nobody in HR is going to consider redesigning a process when handed an automation tool –they’re going to automate and forget it. RPA tools are task automation tools that are not designed to optimize and reorganize tasks into processes.

Disadvantage #2: Long-term technical debt

It sounds counterintuitive that a cost-effective automation tool would, on the other hand, give IT teams more work to do. Again, this isn’t so much the tool’s fault as it is a misunderstanding of what RPA tools can and should automate.

RPA tools interact with software at the UI level. When an update is made to a user-interface, it results in an RPA failure. When RPA is used to string tasks together into processes, the whole workflow can be thrown off by a small change within an application. It doesn’t matter who implemented the RPA bot, IT will be asked to fix it. If the organization decides to migrate away from Oracle products, those RPA bots will need to be scrapped and redesigned.

Here’s how Gartner explains the technical debt: 

“Organizations must manually track the systems, screens and fields that each automation touches in each third-party application, if they want to predict the impact of a third-party system change. Most products support this critical need very poorly.”

Disadvantage #3: Difficulty scaling

RPA is designed to automate discrete tasks at the individual level, and that’s where the bulk of RPA automation takes place –by teams and individuals creating attended or unattended bots that run on desktops or local servers.

These automations for individual tasks are difficult to scale into long running or end-to-end processes, in part because of how rigid and rules-based the underlying scripts are, and also because RPA tools do not provide API-based integrations necessary for reliable, cross-platform processes.

Without additional tools that provide extensibility and orchestration, RPA initiatives tend to turn out like a work of abstract art –lots of colors and dots, but nobody can tell you what the bigger picture is.

Read the latest trends that are driving digital transformation in our new ebook, "Stop Fighting Digital Transformation."

Don’t Let IT Complexity Hamper Your Digital Goals

Get the latest trends and data on what’s driving transformation and how IT teams are addressing these challenges.

Is RPA Worth It?

It depends.

RPA tools can be highly effective at moving data, extracting information, and inputting that information into Excel sheets, third-party applications, databases, and more, so long as the organization can dedicate sufficient time to research those tasks and then monitor the applications involved in order to prevent broken processes.

RPA success depends on where and how RPA is being used. For tasks and processes that include variables or are more dynamic than a rules-based script can be expected to manage, broader automation tools that provide programmatic integrations, API accessibility, and machine learning (or artificial intelligence) are highly recommended. These automation tools can include digital process automation, workload automation, or iBPMS.

Does RPA Have a Future?

Yes, but with some caveats.

The RPA market over the last few years has enjoyed a tidal surge. This has in large part been driven by business leaders anxious to fill in the gaps of enterprise automation. In other words, RPA has been filling a significant need –in this case to reduce human intervention and human error in day-to-day business tasks.

RPA tools are an important piece of the organization’s broader automation strategy. For years, organizations have been automating IT, development, and, guided by CIOs, long-running business processes. But these strategies have yet to reach desktops and daily tasks for many line-of-business roles. This is similar to the last mile problem in transportation –it’s one thing to provide buses and routes that serve broad, general needs, and another thing to deliver that service to thousands of people independently.

Successful RPA implementations cover that last mile of automation. It takes automation down to the individual, and for that reason is unlikely to disappear. It’s a part of an automation ecosystem that includes task automation, process automation, and process orchestration solutions (going from small to big). 

The ultimate goal for organizations is to automate as many tasks as possible, in order to fully automate end-to-end processes that seamlessly provide data and services down to the individual employee to directly impact the customer experience. 

To bring RPA into the fold of enterprise automation, RPA tools should be used in coordination with intelligent automation solutions (including OCR), IT infrastructure and workload automation tools, and iBPMS or service orchestration and automation platforms. Mature workload automation solutions (SOAPs) can automate data transfers at the programmatic level, enforce enterprise-wide governance policies, and provide the necessary monitoring and compliance to keep track of all instances. Additionally, these tools provide low-code REST API adapters that make it possible to coordinate task automation and orchestrate long running processes.

Basic task automation will be a part of this evolution towards orchestration, but it might not require a separate technology forever.

“By 2021, task-centric RPA offerings in their current form will be obsolete. The simplistic task-focused RPA deployments that focus on routine, repetitive, rule-based workflow will give way to zeal and demand for automating more complex workflow. This does not mean the RPA market is going away. We foresee that remnants of the current RPA deployments will be around for the next decade or more (similar to how we still have green screen applications and mainframes). However, we predict a renaissance of the existing market offerings — a shift from task-centric to more process-level automation and eventually to process orchestration.”

-Gartner, Hype Cycle for Business Process Services

Ready To See How We Make Workload Automation Easy?

Schedule a demo to watch our experts run jobs that match your use cases in ActiveBatch. Get your questions answered and learn how easy it is to build and maintain your jobs in ActiveBatch.

Brian is a staff writer for the IT Automation Without Boundaries blog, where he covers IT news, events, and thought leadership. He has written for several publications around the New York City-metro area, both in print and online, and received his B.A. in journalism from Rowan University. When he’s not writing about IT orchestration and modernization, he’s nose-deep in a good book or building Lego spaceships with his kids.