By Leslie Chacko and Dermot Carroll
This article first appeared on Brink on July 6, 2018.
A recent study found that people in business typically make errors at work at a rate of 10 to 30 errors per 100 opportunities. The types of errors noted in the study range from simple tasks such as reading a digital display correctly, to more complicated, non-routine tasks such as responding correctly within one minute of an emergency situation presenting itself. Even the best performance possible in a well-managed workplace has an error rate of five to 10 in every 100 opportunities.
The study concluded with two points: Even with highly experienced and competent people doing the work, there are excessive rates of failure, and letting people work from experience and knowledge creates unwanted random variation that too often produces wrong outcomes.
Let the Bots Do the Work
How can this issue be tackled? Enter robotic process automation (RPA)—a promising stepping-stone for companies looking to free up valuable resource time and minimize human error by automating manual, repeatable tasks. RPA software (or “bots”) mimics human activities, while making processes more automatic, repeatable, faster, and less prone to error.
There are a number of activities across scale and complexity that can be automated through RPA. On the simpler end of the spectrum, screen-scraping technologies can take data from websites or legacy applications, manipulate the data, and key it into another system for use in other processes. On the other end of the spectrum, robots can be combined with more unstructured data and algorithms to manage more complex tasks and become more intelligent and independent over time.
Bots are tireless and can process requests on a 24/7 basis, conservatively delivering a throughput advantage of at least 8-to-1 over a human performing the same task. Such automation frees up people for higher-value tasks. Simple processes can often be automated in less than a month, with returns on implementation typically seen in less than three months.
Figure 1: RPA Advantages
As more organizations add more RPA tools to their service repertoire, they need to consider the following practical steps to maximize the advertised benefits—and to minimize the risk of failure.
As more organizations add more RPA tools to their service repertoire, they need to consider the following practical steps to maximize the advertised benefits—and to minimize the risk of failure
1) Pick the right processes to automate.
Not all processes are good candidates for automation. Good candidates for RPA are processes that are performed often, require users to repeatedly perform similar tasks, and have structured data inputs with high-quality data. As with any process automation, there is the danger of automating bad processes. The ultimate goal should be to improve processes and maximize their efficiency, using RPA as a tool.
Once processes have been identified as candidates for RPA, develop a consistent pathway to sequence the development of RPA automation. Factors to consider include the nature of the application, hardware/software requirements, such as service-level agreements for provision and licensing, and underlying data quality.
2) Pick the right RPA tools.
Different RPA tools have different attributes, strengths, and weaknesses. Not all tools will work in every environment, so it is best to experiment. Test the application with a variety of RPA tools of choice, such as login bots, to give the RPA developer an understanding of the application and allow for an accurate estimate for the development effort.
Many successful RPA teams use multiple RPA tools to automate a wider range of processes. For example, an RPA process that auto-populates customer complaint forms, which minimizes the need for manual data entry, may be supplemented by a chatbot that automatically engages customers on easy-to-solve issues. Best-practice design will ensure that these fit-for-purpose bots “talk” to each other.
3) Engage the right team.
Although RPA providers frequently state that anyone can be trained to use RPA tools, people with technical skill sets will certainly be required for databases, code objects, and report creation. Uniform, technical proficiency is especially important to ensure that developers are able to employ a unified approach, which will streamline ongoing maintenance.
During development, it’s important to engage experienced project managers, business analysts, and day-to-day managers of RPA processes. Ideally, a subset of the team will have been in the business unit for some time, with hands-on knowledge of the processes to be automated. Finally, senior stakeholders need to understand and support the RPA program to provide the organizational momentum required for success.
4) Identify and tackle organizational constraints.
Many RPA implementations face logistical, procedural, and structural challenges. Frequent constraints include lack of a secure access-restricted environment, lack of standards and procedures for creating user credentials, and insufficient existing hardware/software or funds for provisioning those additional resources. Once RPA starts to scale, its appetite for infrastructure can be voracious. It will be important to identify these organizational constraints and to formulate a plan that addresses each challenge.
5) Identify, assess and test the data.
The quality of data supporting the RPA program will have a massive bearing on the success of the deployed bot.
The first step is to identify all data types involved in the process—from internal or external sources, scraped or delivered—and assess their quality. If the data is not sufficiently clean, clean extensively before any automation. If the process rules have been documented in sufficient detail, it’s possible in some cases to automate the data cleansing.
However, even more important is the second step: testing the RPA process on a large data sample to test for unforeseen issues that stem from data anomalies. Typically, you would want to test the system with six months of data to understand the type of errors that might occur and under what circumstances.
6) Anticipate and address resistance as early as possible.
It is not uncommon to meet significant resistance to RPA from within the organization. Prospective users will have heard about RPA in the media, where it is often portrayed as a precursor to job loss. It is important to communicate that this is far from the reality; RPA most often alleviates overwork and allows users to focus on higher-order tasks. Developing a simple product demo can enable users to understand the benefits of the RPA program, since it’s often easier to observe these benefits than to intuitively understand RPA mechanics.
7) Document the process.
It is important to document the RPA program development with design documents and solution design guides to establish proper-use protocols during implementation. Thorough documentation may allow other parts of the business to determine if and where RPA is applicable and if it can deliver additional process improvements. And since an initial RPA process may be a stepping stone to a more complex process, it will be hard to go back and fix errors without proper documentation.
The Difference Between Success and Failure
RPA should be perceived as a stepping-stone to a long-term strategic solution. New, high-speed, digital systems can integrate company data in multiple ways. These solutions require more end-to-end design and longer implementation, but also provide bigger benefits.
However, to progress from concept to reality and to realize the full benefits of RPA programs, companies need to keep these seven leading practices in mind. Aligning with these practices could represent the difference between RPA success and failure.