Robotic Process Automation: the questions that aren’t being asked - Part 3

Maximizing RPA Impact: Looking for Immediate Benefits

Our examination of RPA from an organizational level has covered the general questions a COO ought to be asking as well as the various ways an organization could adapt its organizational model to anticipate a response to RPA. In today’s article, we will expand to thinking about the proper approach for maximizing RPA impact and measuring performance.

Maximizing Impact

In recent years, process optimization projects have been driven by cost-savings. Success was almost entirely measured by the amount of costs saved. While RPA can also be used as a tactical approach to reduce costs, this vision is short sighted. Reduced cost comes at the end of a long process, but is not considered an immediate benefit.  RPA is meant to improve current process throughput and provide better service for the client by focusing on three main benefits:

1. Efficiency – increasing productivity and reducing time required to complete tasks
2. Quality – reducing errors and operational risk
3. Value added – freeing employees to focus on higher value tasks

Many managers may be tempted to measure the success of RPA implementation by evaluating the efficiency of an individual task.  If they do so, they will obviously see an increase in efficiency and quality of the work. However, the overall goal should not be to only optimize individual tasks, but to optimize an entire process (composed of a series of tasks) in order to increase throughput and decrease errors for the entire process.

Firms should identify major pain points and analyze automatable tasks that can remove the bottlenecks. Examining the entire BAU process can eliminate the occurrence of current bottlenecks while also anticipating future ones. Let’s use a simplified Know-Your-Customer (KYC) example to illustrate.

Let us say that the KYC process at a global bank is split into a number of discrete tasks:

1. Client outreach and evaluation of the business opportunity
2. Information collection about the client
3. Due Diligence
4. Risk Assessment
5. Final Approval

A single FTE can process 10 accounts per day on each of the tasks. For the sake of simplicity, let’s consider that 5 FTE’s are assigned to the KYC process above, with one FTE responsible for each task.

The team, as currently structured, has a throughput of 10 accounts per day and an error rate of 5%.

However, a forward thinking manager decides to implement RPA to help improve this process. Certain tasks are excellent candidates for RPA, and eventually RPA is implemented for tasks 2, 3, and 4. The robots can process these tasks at a rate of 50 per day with no error rate. Assuming a large enough backlog of KYC requests and no additional FTEs, should we expect to process 50 KYC requests per day with no errors?

The answer is no, and it’s because the goal is not to perform as many individual tasks as possible, but to complete the entire process for as many customers as possible. Although the robot can process 50 accounts for the assigned activities with increased quality, tasks 1 and 5 will create bottlenecks in the process as the human FTE can still only process 10 accounts. The table below shows the results. The error rate is reduced to 2%, but the throughput remains stuck at 10 due to the bottlenecks in Task 1 and 5.

In order to build a more efficient process, resources should be re-allocated in anticipation of the new bottlenecks that will arise. As shown below, implementing RPA combined with redistributing resources increases the throughput of accounts per day to 20. It also reduces errors to 1.5% due to additional oversight on value added tasks.

Repetitive non-value tasks are best left for RPA, as robots are not limited to a typical 9-5 work day. One robot can accomplish several tasks at a rapid pace, reducing manual errors on simple tasks. However, a human still needs to oversee RPA to ensure the process is functioning correctly and provide support when there is an unusually complex account. Humans can shift focus on complex value-added activities that may be above the abilities of robots, reducing the bottleneck on the client outreach and final approval.

Additionally, this is why it is imperative to coordinate RPA implementation at a process level. If each of the different tasks above is performed by a different team, RPA implementation for one team will not necessarily improve the overall process.

Measuring Performance

Organizations will need to prepare for RPA implementation by setting guidelines to measure improvements. Measurable key performance indicators need to be assessed for the chosen RPA tasks to ensure ongoing usefulness and efficiency for the overall process. These measures should ideally be tracked before implementation to have a fair baseline by which to assess the effectiveness of the RPA implementation.

• This can be measured by performance divided by some measure of time. RPA can reduce the length of time accounts are processed, so the daily capacity would ultimately increase. Another measure of efficiency would be response time in a customer service setting, as robots can find and process information faster than a human.
• Robots eliminate errors that humans naturally make, increasing the accuracy of the data. Having transparent and accurate information reduces operational risk and ensures compliance in an organization.
• Added value. Without changing the number of FTEs, reassigning employees can result in new services or products within the firm. FTEs can also spend more time on client-facing tasks such as improving relationships and addressing clients’ concerns.

These measures set a clear vision to define the success of the RPA implementation in the BAU process. Without these measures, the organization will not be able to fully understand the gains from RPA adoption.

The takeaway for RPA implementation is to ensure that efficiency and quality of throughput, rather than cost of BAU activities, are being optimized within the firm. A successful RPA implementation focuses on the whole process, rather than just on tasks or teams, improves the speed and accuracy of processes, and frees employees to focus on complex value-added activities and new services.

This article was written in collaboration with Mathieu Ruch and Victoria Rosen.