AI & automation · Article · 4 min read
30–50% of RPA projects fail. Here's how to get RPA-ready.

Published · Updated
Key takeaways
- • RPA can reduce the operating expense of a highly manual process by 20–40% within a year, without changing systems.
- • The business, not the technology team, must drive RPA and select the processes.
- • Complex processes must be re-engineered before they are automated.
Conceptually it sounds simple: take a part of a process that is manual and apply algorithms to automate it. It is about creating a virtual workforce. Its goal is to reduce the operating expense of a highly manual process by 20–40% within a year of implementation, without changing systems. By definition it is disruptive, and at the same time transformational.
ProcessArc's checklist for RPA readiness
- The business is driving the RPA initiative, not the technology team. There is strong partnership and reliance between the two, but it is the business side that needs to identify and prioritise processes for RPA.
- There is a larger transformation strategy that uses RPA as one of its levers. RPA is not the only tool needed to transform a company and its client experience. It can automate large sections of a process, but not the process end to end.
- The goals and targets for RPA have been defined. Applying RPA to random processes across the business carries two risks: automating sections of processes that a larger transformation will eliminate or change, and automating low-value steps while missing the synergies of focusing on a department or a series of interconnected processes.
- The "right" processes have been selected. The right process is simple, repeatable, logical, mostly free of exceptions, high volume, entirely manual and therefore costly. Organizations tend to pick the most painful process, which is usually complex. Complex processes need to be re-engineered and cleaned up before automation.
- The process is optimized before RPA is applied. RPA is a virtual helper: it does the low-value, low-risk, high-volume activities so people can focus on what creates value for the company and its clients. It should reduce the white noise in the process, so the process sometimes needs re-engineering first.
- RPA is delivered with an agile method. Traditional software delivery methods are over-engineered for RPA, which rarely changes systems and documents the process in the tool.
- There is a trained, skilled team to design and implement RPA processes.
RPA initiatives can support digital transformation. Treating RPA only as a cost-reduction strategy sells it short: it can also improve client experience, enable growth and improve regulatory responses.
Go further
AI & Automation training
Process-first automation and safe generative AI, online and live.
FAQ
Why do RPA projects fail?
Most fail because the wrong process was selected, the process was not optimized before automation, or the technology team rather than the business drove the initiative.
What makes a process right for RPA?
A process that is simple, repeatable, logical, mostly free of exceptions, high volume and entirely manual, and therefore costly, is a strong candidate for RPA.
Sheila ShaffieCo-founderBusiness transformation leader who honed her skills at three GE businesses: Plastics, Healthcare and Capital. GE Master Black Belt, University of Chicago MBA and co-author of The McGraw-Hill 36-Hour Course: Lean Six Sigma.