The use of RPA can also have downsides. RPA creates risks, like basically every automation technology does (Kirchmer, 2017b). RPA helps to do routine work faster and at a higher quality, but it also can make mistakes faster and with certainty. There is no human check before executing an action. Humans apply intuition and experience even to routine tasks. Poor data quality or the insufficient definition of business rules can lead, for example, to the ordering of the wrong parts – fast and in big quantities. Or missed claim types can lead to significant rework in the claims handling, overcompensating the automation benefits. RPA requires detailed knowledge about the business process it is used in – otherwise expected performance improvements will not be realised.
The use of RPA may also just cover symptoms without correcting the real reasons for issues. RPA was, for example, used for the automated reconciliation of account differences in an investment bank. However, in the mid and long-term it would be much more beneficial to correct the up-stream issues leading to those differences. In this situation, RPA has become an obstacle to real progress. It is a transformation that brings change but not the full possible improvement.
While one of the benefits of RPA is lauded as being efficiency, the reduction of time for specific working steps does not automatically lead to a workforce headcount reduction. Saving a few hours for different roles may lead to more time for the related people, however, they are still required to do the remaining work. In an insurance company this led to a situation that RPA was perceived as not creating any benefits at all. Real cost reduction, should this be the objective, requires a systematic re-structuring of roles and appropriate workforce management, including human and digital workforce.
RPA vendors stress that their tools are easy to implement and use – also for a businessperson. This may be right for simple straight-forward applications. However, to achieve full potential of sophisticated larger RPA environments some expert know how is required for implementation and ongoing adjustment (Harmon, 2017). For a robot to be reliable in an enterprise context, it can be necessary to provide for anticipated exception conditions. This exception processing is often developed using coding and is difficult to do by the average business user, this should be part of the process management capabilities of the organisation. Failing that, RPA expectations may not be met at all or at least not be met fully. A basic business process management discipline should be in place or established to support RPA operations.
An RPA implementation, just driven through various business demands, without a proper automation strategy, also leads to significant issues since up and down stream process effects are not, or not sufficiently considered. The lack of appropriate prioritisation and roll-out planning limits the business results significantly as some of the above examples show.
It is also worthy of note that automation initiatives do not replace systems that are already in place. They create agile ways to overlay new capabilities while improving the existing core systems. If automating processes creates interactions with underlying systems and that system needs to change then it defeats the purpose of RPA as a non-invasive technology.
Combining RPA with artificial intelligence capabilities can also lead to challenges. For example, if Machine Learning (ML) is used to handle more complex process instances, such as the handling of specific claims, the results depend heavily on the available source data AI learns from. If the historical data, input into the ML algorithms, is of a low quality, resulting bad decisions and actions will be only executed faster. Artificial intelligence turns into artificial stupidity.
As shown diagrammatically in figure 4, new technologies, especially disruptive gamechangers such as IPA, may be difficult to understand and integrate into organisational thought and action at scale. As demand for automation increases, multiple tools may be adopted across the organisation. Ungoverned adoption may result in costly maintenance or violation of compliance policies (Workfusion 2017, Intelligent Automation).
These potential challenges can make RPA a dangerous illusion. All these effects are mainly due to a functional technology focused approach to RPA. Value-driven RPA minimises those risks.
Figure 4: Technology Adoption (Brinker, 2013)
The initial focus of most RPA implementations is on the technology. Having reviewed a number of RPA projects, it has been clear that success is much more dependent on a thorough understanding of the end-to-end business process, enhanced through the appropriate application of RPA technologies.
Your email address will not be published.
Company Name *
GeneralStrategy Execution in a Digital World: The Discipline of BPMProcess Modelling IntroductionValue-driven Repository ManagementRapid Process ImprovementAchieving Compliance with DMNBuilding an Organisation for Value-driven BPMPerformance & KPI ManagementTarget Value, Create Your Process AgendaImplementing a Successful Automation (RPA) PractiseValue-driven System ImplementationIntegrated Customer Journey PlanningSystem Design & Requirements GatheringProcess SimulationTargeted Innovation to Maximise PerformanceEnabling Process Standardisation & HarmonisationThe Process of Process ManagementValue-switch to DigitalisationControlling your Systems with Application ArchitectureLean Six Sigma in a BPM ContextProcess Mining: Delivering ImprovementProcess Mining Advanced Technical Training
What would you like to discuss?
Area of Interest