Top 5 RPA Pitfalls and How to Avoid Them
Even with the pandemic-induced current economic climate, the RPA Industry has been growing at exponential rates throughout 2020 and is predicted to continue that trend over the next several years. Currently, the market is flooded with a plethora of “Success Stories”, best practices, and use cases that guide companies towards successful automation programs and specific delivery methodologies. However, as an RPA Services provider, we have noticed that the industry is severely lacking stories of failed implementations. Without these stories of “What went wrong?” companies new to the technology often fall victim to the same mistakes that other organizations have made early on. Recently, AI Multiple published a research piece on the Top 20 Pitfalls of RPA Programs and How to Avoid Them, after reading, we wanted to highlight a few of these pitfalls that we see quite often in the industry and offer our advice for how best to avoid these scenarios and solve for these issues when building out an RPA Program.
Choosing a Process with Insignificant Business Impact
A common pitfall that organizations encounter when starting out with RPA is choosing processes with insignificant business impact. Often an organization will ask people within the selected department what types of tasks they do daily, weekly, etc. that they would like an automation bot to do instead. The mistake here is that employees will suggest tasks they dislike or want to get rid of as opposed to ones that offer real value to the business when automated. The other way companies fall victim to this pitfall is when organizations look to automate a department’s large predominate process as it takes up the most time. The mistake here is that a process as such could be difficult and cost too much to automate, or worse involve difficult steps and numerous human validation points causing it to not save as much time as they expected and making the investment not worthwhile. When deciding what to automate, organizations should prepare a list of goals and required metrics for what their automations should accomplish. They should also be sure to fully understand the list of attributes that make a process ideal for automation. In other words, companies should avoid automating for convenience and prioritize automating for ROI.
Selecting the first round of processes in an RPA journey can be daunting. The high-overhead costs add pressure on the organization to select processes that give a large amount of ROI in order to justify the investment. This leads to rushed projects that are high-risk and fall short of business expectations at the end of development. Taking the time and making the decision to invest in a strong Process Discovery practice is the best solution for running a successful automation program while guaranteeing ROI.
Process Discovery is viewing business processes under a lens for automation. A strong Process Discovery practice should aid businesses in selecting processes that are fruitful and remove mundane tasks from their employees. It works by observing how standardized the process is, acknowledging the inputs and outputs expected from the process, and deciding whether technologies outside the realm of RPA are needed for implementation. The most important resource required for smooth process discovery operation is an SME who understands the process and knows what is necessary to complete it. When an employee or manager proposes a process for automation, the next step in the discovery phase is for a business analyst to interview the SME to determine if the process is fit for automation. What makes or breaks this decision is the process SME providing all relevant information, from there, the process discovery team can easily identify if the process is worth pursuing. Overall, based on the business goals and standards set in the early stages of building an automation program, if the ROI, time savings, feasibility, and ease of implementation do not meet the standards the company has set then the process will fail to make it through the discovery phase. Evaluating business needs against the set metrics is what process discovery is made to do. Building out a strong process pipeline and proposing automations that deliver real results back to the business are what prevent companies from making the mistake of selecting processes whose output does not justify the investment made to create and deploy it.
Striving for End-to-End Automation When It’s Not Cost-Effective
As RPA becomes more universal, there is a justifiable desire to automate cross-functional business processes from end-to-end. The benefits of doing so are clear: cost savings, productivity and efficiency gains, capacity increase, general process improvement, etc. However, in most cases, this is a tricky situation that organizations are lured into based on perceived benefits. We often witness across various implementations that an organization’s “boil the ocean” approach to automating an entire end-to-end process often proves to be impossible, or, at best, extremely difficult. RPA is not a one-size-fits-all solution. For a true end-to-end automation, other technologies such as OCR or machine learning are likely required and, in some cases, extensive process re-design or optimization efforts are needed just to enable a process for automation. These extra steps are all likely needed if the approach is to focus on an entire process, which in the end proves to be a costly endeavor that requires a substantial amount of effort. However, this should not be a barrier to realizing value from your RPA program.
The key is to identify your quick wins. For example, in an order-to-cash process instead of automating from end-to-end, identifying the sub-processes within it that are best suited for RPA such as Invoice Processing and Customer Account Updates could prove more valuable to the organization in automating them separately. To identify ROI-heavy solutions and plan out an automation road map accordingly, an organization needs to assess which processes would deliver the most value in conjunction with the effort required to implement the proposed solution. For this to happen effectively, investing in Business Analysis prior to Automation is a crucial step in an organization’s journey.
Having a trained automation business analyst involved from the beginning stages of automation mitigates the risk of tackling a project too large, or worse, attempting to automate one without enough ROI to justify the time and effort it would require. BAs are trained to identify the best processes based on a variety of benefits to the business. Outside of time and money they are trained to calculate throughput increase, error reduction opportunity, ease of implementation, and feasibility scores. Relating to this pitfall specifically, feasibility scoring solves for this as it determines how much of a process can be automated and what outside technologies and efforts are required to do so, thus preventing a company from trying to tackle an end-to-end automation and failing. A trained Business Analyst would also be able to break apart a proposed end-to-end automation and create a structure for automating the sub-processes within it. In most cases, automating sub processes within a major business process offers a higher ROI with similar time reduction savings while requiring less time and monetary investment to develop and deploy the solution.
Company Lacking a Clear RPA Strategy
A common mistake companies make early in the automation journey is not setting up a clear RPA strategy. Without a clear plan from management, departments and sub-groups within a company will set up different governance models and each development team will organize and store their automations in different formats and locations. What happens then is the RPA program becomes disorganized, causing bottlenecks in the automation request pipeline and delays in implementation. Eventually, the problem escalates, leading to an overall halt of program growth or the decision by management to defund the program all together.
In the early stages of automation, companies should plan to evangelize their RPA goals and set up an automation Center of Excellence (COE). From opportunity assessment to execution and escalation procedures, a COE handles the overall goal and management of automation programs. Without a well-defined governance model to manage the automation infrastructure, promote completed projects to production, and set up a shared location for automation materials, companies find themselves in an up-hill battle they will inevitably lose.
There are various models for RPA deployment and governance; options include federated, centralized, and hybrid models. A centralized model manages the automation program from one main governing body, a federated model also has one governing body however delivery is delegated to each individual business unit; a hybrid model is the combination of the two. Each has its pros and cons, but every organization should know which version will best suit their business needs.
Setting up an automation program for the enterprise requires both a thorough understanding of challenges unique to RPA and expertise in infrastructure/Dev Ops. A COE provides this guidance along with various services such as demand generation and management, process analysis, automation development/configuration, and ongoing maintenance and support. In order to provide these services efficiently, a successful RPA-COE will need to establish three verticals: DevOps, Infrastructure, and an Operations Center. Governance and business continuity should also run horizontally across these verticals to standardize and manage overall operations throughout. A company’s approach to ensure that teams effectively run their automations truly determines the overall success of any program. With proper structure and aligned goals, organizations can avoid risks to their RPA program caused by a lack of a clear strategy.
Choosing a Process That Changes Frequently
Often times we see organizations gravitate towards automating processes that change frequently, or as we onboard a new client, we see that one of their major pain points in their automation program is dealing with a process that is always down or needing support to correct changes. Choosing a process that changes often is not typically a strategic decision as the automation will crash frequently and need modifications quite often, making it not worth having automated it in the first place. However, if a frequently changing process is a major function of daily operations or will turn a large ROI, with an organized plan of action, it’s not something companies need to avoid entirely.
The business or RPA team might not know exactly why frequent changes are happening or how to predict them, in turn, this deters businesses from automating these desirable processes. To both avoid the headache of constant breakdowns and corrections, process mining can be used to standardize, optimize, and predict changes, allowing organizations to turn this downfall into an automation win.
Process mining can use underlying data created by the automations running systems to provide a proper diagnosis as to why it’s constantly changing. The diagnosis will return insights and provide an understanding that can help the organization see exactly what is causing the changes by examining all the exceptions and reasons that they occur.
When a process changes frequently it is usually attributed to major exceptions thrown by the process or an important changing variable within the bot actions. An example of this would be the loan application process. In a bank’s loan application process the majority of applications are processed based on income, credit score, outstanding loans, etc. however, exceptions can be incurred from applications for veterans, teachers, customers with disabilities, or any other group that receives federal or lender discounts and unique interest rates. In this case, the process can become bottlenecked due to availability of processers working against the number of applications the bot requires human actions. Other errors that may require human correction would be changing variables such as interest rate fluctuations, preferred lending company policy changes, or even internal operating systems with frequent updates. Any lending company would want to prioritize the automation of their loan processing task, however, to avoid the constant breakdowns and large amounts of exceptions, the organization would want to use process mining to eliminate and predict the changing variables, increasing automation ROI and throughput.
Process mining has been gaining momentum within the automation industry as it is able to break down processes to a micro level and provide the business with a list of action items to resolve the issues and optimize the process overall. Back to our loan application example, the frequent process exceptions can be extracted, mapped out, and transformed into fully automated subprocesses based on the information returned by the accurately mined process maps. This eliminates the need for human involvement with the exceptions without having to reconstruct the original application processing automation. By dissecting the process and automating the error handling for a specific step, the overall automation can run end-to-end without stopping and requiring human input to solve the error. Process mining transforms faulty automations based on a deep understanding of the process. The finalized solutions are then capable of anticipating encountered changes, allowing for organizations to automate the desired process regardless of frequent changes.
Forgetting About Automation Maintenance and Support
One pitfall that we see clients stumble across often is not taking maintenance needs into account. When building automations, whether its attended, unattended, machine learning, or an AI-integrated process, a critical error is made by the organization when they forget that their automations need updates, bug fixes, and general support after they are built and deployed. Whether it’s a bot failure, a server failure, or even a change to a system’s login credentials, sooner or later all automations are going to need maintenance and support.
Not adding a support plan to your automation program is like buying a car without factoring in the service visits and oil changes; you can’t expect to operate at full capacity without it. Much like a car, bots and automations need to be serviced, upgraded, monitored, and tested regularly. Typically, when a check engine light comes on, the owner would bring their vehicle in to a mechanic. When a bot’s “check engine” light comes on, its best to have a dedicated support team to look into the issue rather than pulling an in-house and unprepared developer away from a project, or worse, having someone less experienced look into the issue.
Without a trained automation support team or IT group that understands RPA bots and their respective platform infrastructure, who can an organization turn to for reliable support? Managed Robotic Operation Centers or ROCs are becoming a necessary and invaluable pillar to successful automation programs across all industries. A ROC is implemented to ensure that deployed automations continue to be productive throughout their daily operations and are continuously improved over time. Continuous monitoring not only reduces bot down time, but it can work to optimize license usage and amplify bot productivity, thus increasing the ROI from automation program costs already accrued and strengthening the deployed digital workforce.
When RPA is implemented within an organization, support should be included as part of the company’s overall IT operations from the beginning. As dependency on automations increase, the need to support those automations along with the necessity of mitigating risk of automation failure within business functions become critical. Without the proper monitoring, maintenance, and reporting, companies are not maximizing the value of their digital workforce. Acquiring a ROC for automation support covers all of these needs and increases the returns of automation programs overall.
Pitfalls and Their Automation Solutions
The pitfalls discussed in this article are some of the most common mistakes we see in the industry. In some cases, it’s good to know what to avoid, however, when the technology and expertise are leveraged correctly, there is a way to turn each misstep into a strategic solution that benefits the organization. RPA is becoming increasingly popular and reports see this trend continuing for the next several years. This is true even more so after the coronavirus pandemic as business are looking to put solutions in place now that will solve for if a similar scenario happens again in the future. A digital workforce can scale processes up and down by adding bot capacity instantaneously; solving for staffing and demand volume fluxes as they occur. When companies go all in with Process Automation and the surrounding solutions, they can return major hour and dollar savings to the business. Knowing what to avoid and how to solve for mistakes is what separates organizations that are making strides within automation from those that struggle to break even and eventually fall short of necessary digital transformation.