Main image of article What Good is Robotic Process Automation (RPA)?

The headlines abound: Business automation software (a.k.a. Robotic Process Automation, or RPA) has helped thousands of companies save money by automation of repetitive, tedious tasks and enabling employees to focus on mission-critical operations.

Many business process tasks can be automated, but not all of them need to be.

An Ernst and Young study found 30% to 50% of RPA engagements fail, citing poor planning as the difference between success and failure. These planning challenges typically occur when business processes are evaluated manually – via a lengthy procedure involving employee interviews and random spot-checks – resulting in inaccurate process discovery, task frequency and time metrics. Manually discovering processes not only takes too long, but it also generates incomplete data for RPA planners and developers. As a result, RPA bots are often implemented where it is “easiest” to apply the tools, not in most critical areas that would maximize ROI.

RPA brings no value if you do not know exactly what to automate. Quite simply, you can’t automate stupid.

Not a Quick Fix

Another problem enterprises encounter is that they attempt rapid implementation of RPA tools based on this faulty information (instructions for bots missing key alternative scenarios). If the enterprise doesn’t take the time or have the tools to discover and optimize business processes, they are simply “throwing spaghetti on the wall to see what sticks.” This is a very costly mistake!

Would a company hire a new employee without knowing what job they are hiring for? No. Similarly, companies must fully develop an accurate “job description” for RPA bots in order to maximize their technology spend. Notably, only three percent of organizations have deployed more than 50 robots, which indicates very low confidence in the value of broad RPA deployment.

The first step toward maximizing return-on-investment for RPA tools is to conduct a detailed and accurate analysis of the company’s complete set of processes using a combination of automated process mining and task mining technologies. This analysis asks, “On what tasks are employees really spending their time?” as well as “What exactly are the processes that can and should be automated?” The ability to obtain end-to-end process metrics early, at the planning stage, would enable planning based on facts, not guesses or assumptions, accomplishing maximum ROI.

Taking Real Savings to the Bank

Let’s have a close look at an actual case study from a major North American bank. The bank, with $50 billion in assets and more than 3,000 employees, was looking to streamline and improve its anti-money-laundering (AML) operations. Despite continuously increasing the number of AML reviewers…

  • The reviews took too long, creating a backlog of transactions. 
  • The process efficiency was very low, with only 3-5 percent of review cases being real suspects.
  • The review process itself was inconsistent, with some reviewers performing excessive validations while others stopping at “just enough.”

Before trying to automate any processes, the bank utilized process and task mining software to reveal the end-to-end tasks performed by all team members. With the technology, executives were able to visualize the work process of each reviewer (including activities, screenshots, and systems used), measure the time each process activity was taking to complete and identify non-value-added activities. They were also able to analyze deviations and variances from the “optimum path” and generate a reliable “Standard Operating Procedures” document for AML reviewers.

Such a high level of accuracy and detail enables analysis even of “unstructured processes,” such as tasks performed by staff in word processing apps, spreadsheets and email. Unstructured processes do not have a descriptive user Interface and are extremely difficult to analyze manually.

After completing their analysis, the bank implemented targeted RPA tools to accelerate and improve reviewers’ workflow and eliminate redundancies, achieving up to 60 percent FTE cost savings in non-post branch validation. Additionally, the bank initiated a new training protocol based on the automatically generated SOP documents.

Beyond its AML application, the technology was used to find FTE cost savings of 40 percent in funding and onboarding of home equity lines of credit and 33 percent cost savings in cash transaction reporting. In short, thanks to proper diagnosis and analysis, the robots knew exactly what they needed to do to help the bank increase productivity and ultimately add to its bottom line.

Robots are great… if you know what you’re hiring them to do!

Sofia Passova is founder and president of StereoLOGIC, a leader in integrated process and task mining technology which helps companies discover inefficiencies in their business processes and suggests solutions to improve workflow and revenue.