Robotic Process Automation: Plan Carefully Before You Automate

By James Hejka, Practice Director, Business Optimization Services Center of Expertise, Jefferson Wells

Robotic process automation (RPA) is rapidly gaining favor across many industries as a labor-saving, error-proofing solution for repetitive tasks. However, companies that automate with unrealistic expectations or inadequate preparation have struggled to achieve desired results. Longer than anticipated timelines, low returns on investments, projects stalling without completion and an inability to scale to more processes are all results that can be attributed to crucial mistakes made prior to beginning the project.

Choosing the correct processes to automate, optimizing the process prior to automating, and selectively automating the portions of the process that make sense are the keys to success. This article will discuss the importance of making the right choices prior to automation.

RPA and its advantages

RPA is a category of software tools that enable complex digital processes to be automated by performing them in the same way a human user might perform them, using the user interface and following a set of predefined rules. The newest versions of RPA are enterprise-wide tools that can harvest data from disparate applications. Proper use of RPA can achieve double-digit reductions in error rates, cut data entry costs and reduce the costs of offshore staff — but only if implemented correctly.

Choosing which processes to automate

A task which employs the use of any knowledge or skill repetitively in a routine manner could have potential for automation. However, many companies fall into the trap of applying RPA to any task that appears to meet these requirements.

Here are factors to consider:

  • Is the process itself optimized? Have workers been performing this task in exactly this manner for a long period of time? Are there any unnecessary steps? Look at the efficiency of the process before setting it in stone with RPA. You could reap double rewards.
  • Will automation impact the customer experience? If you fully automate your telephone system, for example, making it impossible to connect with a live operator, this could cause resentment among current or potential customers.
  • Are there decision-making processes that can also be automated if they follow a predictable logic? Examples include insurance and financial services transactions which automate financial analysis, risk assessment, and credit checking.
  • Are there exceptions within the process? If there are too many, it may be too complicated or simply not cost-effective to automate the process.

If you’ve determined that a process is a good candidate for RPA, it’s time to move on to the planning stage.


  • Set clearly defined start and end points. Understand the beginning and end of the process with clear definitions and stick with it. This allows for finite requirements gathering and a clean build process.
  • Identify human decision points. If human intervention is needed, program the automation to stop at a certain point and signal that more input is needed. The automaton can always begin again after the human intervention is complete but knowing those human decision points is critical.
  • Don’t try to design for all exceptions. If process exceptions occur rarely (a couple times a month or so), it’s not worth the trouble to program them in. 
  • Separate long end-to-end processes into logical segments. Creating one long, complex bot takes longer to design, longer to build, and raises the potential for error due to its complexity.

Optimizing the process. Here are more tips to make the RPA process go smoothly.

  • Consolidate multiple formats into common templates. If multiple users are storing the same information in different Excel templates to execute the same process, an example could be to migrate the information into one template style to simplify the RPA development process.
  • Eliminate variants of the process if possible. Some users may have slightly different methods for performing the same task. Ascertain which process is the most efficient and eliminate the variants.
  • Drill down even further into the “why.” Which steps of the process are truly necessary? Many times we find process steps are unnecessary and only executed because “we’ve always done it this way.” 
  • Document the “unwritten.” Some steps of a process may not be documented in writing. It’s crucial to make sure all steps are accounted for before designing an RPA bot.
  • Maximize system functionality. Workers may be using unnecessary or outdated steps to complete a process. An example would be printing out and scanning a document instead of sending a PDF via email, simply because the user wasn’t aware of existing system functionality at their disposal. 
  • Consolidate process data. Instead of forcing a Bot to search across various drives and servers, place all associated documents in one central repository. This will vastly speed up the process and reduce the complexity.

The bottom line? The more you simplify a process before you begin, the easier it will be to automate, the more effective it will be to scale, and the better your chances of achieving desired return on investment.

For more information on how to optimize before you automate, contact Jim Hejka, Practice Director, Business Optimization Services, Center of Expertise at Jefferson Wells.

Did you miss the chance to join your peers across the country at our Jefferson Wells Annual Virtual CPE Conference November 16-18? Our industry experts – including Jim Hejka – highlighted emerging topics in Data Analytics, Technical Accounting, Cyber Risk, Internal Audit Priorities, and more. The event kicked off with a Keynote from Becky Frankiewicz, ManpowerGroup North America President, who revealed the latest insights on workforce transformation in the U.S. Watch the on-demand conference or individual sessions here.