• Arianna Campbell, Shareholder

Process Automation Starts with Consistent Processes


The intersection of process and technology has been a key focus area during our Boomer Lean Circle meetings this year. As part of this discussion, robotic process automation (RPA) has been a hot topic. Of course, RPA is a compelling issue throughout the profession as firms look for ways to leverage technology, become more efficient and free up capacity to become more consultative. However, among Lean champions from some of the profession’s top firms, the focus inevitably turns to the process aspect of RPA.

Let’s ignore the R in RPA for the time being. Essentially, RPA is process automation. Process is a series of actions or operations conducing to an end. Automation is the technique of making an apparatus, a process, or a system operate automatically. Simply put, process automation involves taking away routine, repetitive tasks so the people doing them can focus on higher-level work.

The Deloitte Insights article, Robotic Process Automation: A Path to the Cognitive Enterprise, includes an excellent illustration of how a typical back-office process can be automated.

As you can see from the figure above, four out of five steps in the process can be automated, with a human responsible only for validating the actions performed by the computer. The key is, the manual process that was optimized started with consistency. You’ll run into issues if you try to automate inconsistent processes.

Automating inconsistent processes

In 2016, EY noted that 30% to 50% of initial RPA projects fail. That high rate of failure is not due to the technology itself. Often, it’s because RPA needs clearly defined processes to work. When companies don’t have clearly defined processes, they either automate the wrong things or get lost trying to reverse engineer the process.

Imagine trying to automate the process of invoice approval in a company where there are more than a dozen different approval processes. That lack of consistency might stem from acquisitions where each office has its own system for approvals. Or perhaps you have variations in the process even within the same office due to personal preferences of partners. Either way, trying to target RPA at highly complex and inconsistent processes is time consuming, costly and will likely result in failure.

Identifying good candidates for RPA

The processes that are the best candidates for automation are stable, repetitive and structured. Some examples include:

  • Opening emails and attachments

  • Logging into web/enterprise applications

  • Moving files and folders

  • Copying and pasting

  • Filling in forms

  • Reading and writing to databases

  • Scraping data from the web

  • Connecting to APIs

  • Performing calculations

  • Extracting structured data from documents

  • Collecting social media statistics

  • Following if/then algorithms

So before diving into RPA headfirst, remember that RPA works best when processes have been thoroughly optimized to be efficient and effective. Take the time (and cost) upfront to review your processes, reduce decision points and loopholes and make them more efficient.

Using the Lean Six Sigma process improvement methodology, you may find that you’ve created a process efficient enough that automation is no longer needed. Alternatively, you may discover technologies already in use in your organization that can automate parts of your process without the need for RPA. Even if you do decide that RPA is a viable solution, you’ll be much further ahead in meeting your process requirements than you would be if you’d skipped the process improvement step.

Automating without first investing in process improvement ultimately results in establishing an inefficient process that returns poor results.