Five most critical points of failure in RPA implementation
A successful robotic process automation (RPA) implementation starts with understanding the RPA opportunity and developing a good business case for investment. This requires thorough analysis at the beginning of the RPA journey. Everyone can identify an issue or a problem in your business, but what people often don’t realise is that they are identifying the symptom not the cause.
Root Cause Analysis is a great way of tracing the pain back to its source. Once that’s been achieved, the next step is to identify the issues it causes, how it manifests itself and to quantify the level of pain.
How much time is lost, what level of wasted manual effort, what kind of errors occur, what customer impact it has, how does it affect your own staff and finally how much it really costs.
Once that is done and things are confirmed, the analysis moves on to mapping out the actual process. Organisations that have failed or run into difficulty have overlooked the detailed mapping that’s required. Without this, you run the risk of selecting processes that are not the most suitable for automation. This leads on to the second failure point:
Failing to understand complexity
When organisations first look into the benefits of implementing RPA, there’s a lot of material available to provide guidance on how to go about it. Generally, this recommends organisations select simple, non-complex processes.
This sounds great, but determining what a simple and non-complex process is is not as easy as it sounds. A process may be short and appear to be completed by a single person or a specific role, but what isn’t so obvious are the dependencies between that function and any that come before it or after it’s been carried out.
It certainly helps to apply logic – such as ensuring the process is consistent, has rules that can be automated, is non-subjective and so on – but there’s more to it than that.
Mapping out the user roles, the application access and how data or information is sourced is also vital. Likewise, ensuring everyone has a common understanding of what the process actually does and knowing when that process or function is considered to be complete, is also important.
A structured analysis approach to break processes down to basic functions, roles, data flows and decision points will help to avoid making the mistake that, simply because someone uses or executes the process in their job, that they assume they know everything about it and what the dependencies are to make it work correctly.
Running a PoC or selecting an RPA product before proper assessment
It’s highly tempting to assume that running a Proof of Concept (PoC) will help identify whether RPA is appropriate and will solve the business challenges that you are facing. It’s certainly a key part of a selection process, but the PoC requires preparation.
Measuring your current effort and costs to operate your business is a significant step in quantifying what and where the inefficiencies are. If you go straight to trialling various products or solutions without a comprehensive understanding of the problems and a clear definition of what success looks like, this will likely result in spending money on a solution without a clear plan of what areas are prioritised and how to measure the value and benefits.
The outcome of this is that you get caught in the trap of attempting to automate as much as possible to justify the spend on a solution. This can lead to the next failure point.
There is a belief that in order to see a good return on your investment you need to automate as much as possible and, for a process that is deemed to be a suitable candidate, that the whole process must be automated. This is likely to lead to failure and disappointment.
While a process may have been identified as a good candidate, you must avoid the temptation to attempt to automate it completely. Start small and take a modular approach. Find the simplest, basic functions that meet the automation criteria of:
- Being consistent
- High volume
This is where you’ll also find the quick wins and gain some early traction with the technology. Focus on these as the core candidates and avoid getting stuck attempting to automate what will quickly become a long, complex process that will consume a lot of effort and quickly see costs spiral.
Over-estimating the capability of RPA
We are seeing and hearing much about machine learning and artificial intelligence. It’s easy to be seduced by the marketing and sales messages that imply these emerging technologies will solve all of our issues, including process automation, right now.
There’s no question that Intelligent Process Automation through machine learning can be incredibly powerful and enable organisations to evolve at a much faster rate and improve their business intelligence. However, to be truly effective, machine learning algorithms require big data to learn.
Given that the secret to success for RPA is to start small and take a modular approach, there is little or no opportunity for machine learning to provide any benefit early in a journey towards process automation.
The outcome and benefit of machine learning is to predict things with far greater accuracy through statistical analysis. The larger the volume of data, the more accurate the predictions and the clearer the information on hidden trends and insights.
This does not always directly correlate to process automation which is solely focused on driving out inefficiencies, speeding up delivery, reducing operational costs, eradicating manual errors and freeing people up to do more interesting and valuable work. The ultimate aim of a successful RPA implementation is to be able to achieve rapid results and demonstrate value through automation enablement.
Optimising the implementation process of RPA
Developing an optimal process for implementing RPA will help you avoid each of the common points of failure discussed above. It also keeps you true to an ‘optimising ethos’ because, after all, if RPA represents an opportunity to optimise business efficiency, it’s almost unthinkable to implement RPA with an inefficient process! This is why we developed ‘RAE’.
This is why we've developed our Assurity Process Automation Framework. Find out more at Intelligent Process Automation.