RPA- Five biggest hurdles and how to overcome them

Despite all the hype generated by RPA and automation, the ground reality does not match the heightened expectations. Many implementations are excruciatingly painful and there is a common thread that run through many of these projects. Here I list the 5 biggest hurdles that organizations already face or may face as they rush to get into the Automation bandwagon.

Hurdle #1- Limited scope of automation : The bulk of all the processes have a small headcount(10-20). A small number of processes have hundreds or thousands of advisors. However this does not mean that a single implementation will cover all the advisors. Most processes have many use cases or variations and even in very large processes, a handful of agents (10-15) work on a single use case. What this means is that a RPA implementation has to focus on just 10 -15 agents that work on the particular process scenario. The implications are twofold- 1)It becomes difficult to justify an implementation for just 10-15 agents and 2) The very assumption of rapid scalability across all processes is dramatically weakened.

Hurdle #2-Lack of tools for performance measurement: Most organizations have very limited data on performance, much of the data being restricted to overall process performance (10 transactions per hour and so on). The exact benchmark for specific use cases, tenured Vs non tenured advisors etc are not available. Most organizations do not possess the talent , tools or skill sets necessary to measure performance. Performance measurement is a science that draws knowledge from  multiple disciplines- modeling, Statistics and simulations. Most organizations lack the skillset, tools and experienced members who can handle such diverse areas of study and implement them to their specific environment.

 Hurdle #3- Lack of benchmark and data for existing process variations: Some processes are not stable enough and  this is one of the major causative factor for wide variations in performance. Even for stable processes, individuals perform differently at different times. Lack of benchmarking affects implementations in two ways -1) It becomes difficult to figure out which use case need to be automated first 2) Lack of data on current performance makes calculation of  ROI for a use case a huge challenge. Availability of accurate data can help an organization to leverage and squeeze out more savings from automation. Maybe instead of blanket implementation across the entire user group, it might be preferable to roll it out to say 1-2 months experienced agents. Intelligent rollout decisions can be made if an organization has the benchmark data available and is well equipped ,understands the science of performance and is able to apply them to their organization.

Hurdle #4- Long cycles of implementation: Only the trivial process variation can be implemented rapidly- most process variations take weeks or months to implement. In many situations, knowledge on existing variations is not available as documented knowledge in the organization. As a result discovery of use cases themselves become a trial and error approach. In other cases, Documentation and training is static and does not change with the changes in processes. Entire infrastructure of maintaining knowledge on AS-IS and TO-BE process variations are nonexistent. Lack of process infrastructure, costly and timely implementation schedules make it difficult to scale up various RPA initiatives.

Hurdle #5- Rapid Change management: Processes change rapidly. Some changes are small and insignificant and some can have widespread ramifications. Small changes are typically a result of changes in business rules, change in policy, organizational changes, mergers etc. Some changes happen when the entire platform or application is replaced. In case of a platform or a large scale change the entire RPA implementation needs to started all over again. The frequency of change can act as a dampener to many RPA implementations.

How can organizations overcome these hurdles. Some recommendations(specifics may vary depending on the organization):

1. Create a Performatics group within the overall framework of the Process excellence team. This group should be adept in advanced mathematics/statistics and should be able to develop models and simulations for organizations (In the next article , I will cover the basics of the Science of Performance).

2. Document and maintain Capture data and Simulations for AS-IS process variations. These assets can be helpful in many ways. Not only can they help to build a business case for an automation project , but the same information can be leveraged by the Training group to onboard agents. This group should preferably be a part of the Process excellence group and the content and data should be made available to the Training team/Transition teams and other QA functions in the organization. Since most processes undergo frequent changes, the  team should focus on creation of processes that can update the content in real time.

3. Benchmark all the process variations across multiple range of users. The benchmarked data can be used not only to create a business case, but can also be used to determine the root cause for variations, to understand the extent of process stability and help organizations to improve performance.

4. Create a process datasheet that captures all the variations, performance of user groups across these variations. key metrics (TAT, errors or some other metric that is a business challenge), benchmarks, frequency of change, criticality of process to the organization etc. Decisions on automation/ performance improvement or root cause analysis will be easier if the organization consistently and regularly maintains data across all their processes.

5. Ultimately the time taken to implement bots for specific process variations will be the single most impediment for rapid proliferation of Automation. Organizations need to invest on tools and technologies that have a clear roadmap to reduce the time taken to create Bots.