About the Author
Ask any garage workshop hobbyist—there are many tools for many different purposes; one may well work for a certain task, but not for another.
So it is with robotic process automation (RPA), the rapidly-growing technology that, broadly defined, allows a business to deploy intelligent software systems—or software “robots”—that mimic the actions of human users.
RPA can be applied to a wide range of industries in which humans perform repetitive, high-volume, highly transactional functions, freeing people up for higher-value work.
The technology is still in an early growth stage, but it holds immense potential: dramatic improvements in accuracy and speed for some functions and potential cost savings of 20 percent or greater compared with your more labor-intensive processes.
Companies are understandably excited about RPA. Many organizations are kicking the tires of RPA, often driven top-down by COOs, CFOs and CIOs. They want to use RPA to help them reduce the cost of manually intensive efforts and free up capital and resources for higher value activities. RPA also enables improvements in quality, compliance, customer satisfaction and time to market.
Let’s tap the brakes for a minute.
Before latching onto RPA as the answer for a specific use case or business problem, ask these seven questions first:
1. Could the problem be solved by changing the business process?
Automation is not the solution to a broken process. Better to solve the process issue first and then automate if the business case still supports it.
2. Is it a manual process with enough volume or compliance risk to justify automation?
A high-volume manual process is usually a “no-brainer” for RPA consideration. A lower-volume manual process requires you dig a little deeper. It might still be a candidate, if there is a potential for the same team to increase volume (for example, help the sales team increase closure rates). Or, if automation can help reduce or eliminate costly compliance mistakes resulting from human error.
3. Does the process involve structured or unstructured data?
RPA requires structured data (for example, a file or a set of fixed fields on a screen). If the data is coming in an unstructured format, like email, text or paper, then additional technology will be required for pre-processing the data into a structured, digital format. For example, an Optical Character Reader (OCR) might be required to convert to a structured digital format (some RPA solutions also have rudimentary OCR). Cognitive technology could also help to convert, classify and categorize emails and text into a structured format for RPA to process.
4. Is it a rules-based process, or does it require human judgement?
RPA works for relatively simple, if-then type rules. For example, “if a customer number in file X matches an SAP customer number, then update SAP.” If human judgement is required, you may need to consider a workflow or case management solution to help route, track and support the business operation. Artificial intelligence (AI) deep learning solutions are another option to help automate tasks requiring human judgement. For example, AI can help determine whether a warranty claim is covered by a policy.
5. Are there a high number of “exceptions” in this process?
RPA is most effective when exceptions are low. Exceptions require workflow and/or case management and most RPA tools to date do not provide this capability. For exception processing, you either need to integrate with your existing workflow or case management toolset or implement one as part of a broader transformation initiative.
6. Is the process fairly static?
For RPA to be most effective, you need the process and systems with which you are interfacing to be relatively static. On the simplest level, RPA is recording and then replaying the work performed by a user. If the steps or systems change, then RPA needs to change as well.
7. Could a change to the underlying technology, system or root cause solve the issue?
You’ve gone through steps 1 through 6 and determined that RPA is the right fit; the final step is to make sure there is not a better way to solve the problem at the source. For example, you might be planning to use RPA to help your spreadsheet jockeys automate the conversion of reports into an easy-to-use pivot table. However, rather than spend money on RPA software, perhaps the best course is to just change the report.
Key takeaway: RPA solutions have delivered significant value over the last couple of years, and I expect that to continue. RPA is a specialized tool for certain jobs, but it isn’t the right solution to automate every process. Consider the questions above to find the right path forward.