Top 13 RPA challenges and ways to overcome them

Our RPA expert, Dzmitry Kliuchnik, highlights 13 RPA key challenges:

These challenges can be further segmented into organizational, process selection-related, implementation, technical, and post deployment issues. Below you will find a detailed rundown of RPA challenges together with possible solutions.

Deloitte considers this one of the major challenges of RPA implementation. When starting with automation, organizations tend to separate business and IT responsibilities and delay involving the IT department till after the proof of concept phase, which causes frustration and skepticism.

David Wright, Consulting Partner at Deloitte, emphasizes the importance of IT involvement, “IT are absolutely critical to the successful deployment of RPA. This was a lesson we learned early on in our own RPA deployment in Deloitte. I’ve found there is a significant difference in both speed and cost to deliver between clients that have an engaged and supportive IT function and those where IT is less supportive.”

The business department can highlight the fact that RPA efforts aim to unburden IT staff and not completely replace the system. Also, be prepared to address any security concerns the IT department might have. To ensure the business and IT can understand each other and communicate freely, you can establish an RPA competence center containing members from both sides.

Another challenge of RPA stems from the fact that organizations don’t appoint a person responsible for different parts of an RPA project.

This can be confusing for everyone involved and will hinder the decision-making process.

Also, after the RPA solution is implemented, it is crucial that employees understand their roles in the new environment.

It is best to explicitly specify who is in charge of various project aspects, such as approving the project’s design, monitoring the execution, and measuring the success rate.

If a company selects an unappropriated business case for their automation efforts, they will get rather limited profit.

When choosing a case for automation, it is vital to understand when exactly you need to automate and what you aim to achieve. There are two directions that you can take with RPA:

Also, when projecting the overall RPA costs, consider all the contributing factors. This would include the price of the automation tool itself, infrastructure costs (if any change is needed), and costs of monitoring and maintenance of the deployed processes.

Defining automation strategy is an essential factor in avoiding RPA pitfalls, according to Zeeshan Rajan, Senior Manager at PwC. Here is what he said, “RPA is a technology that is easy to use, but things take time, and so does an RPA implementation. A sound foundation must be created for the purpose of further scaling, and this requires a solid basis with a defined strategy, communication plans and change management, trained RPA experts and a stable IT infrastructure – just to mention some examples.”

Before implementing RPA, every company needs to define a strategy to guide the project. While doing so, you can ask yourself the following questions:

Not all processes are suitable for automation. Sometimes, automating a small process can result in significant savings, while automating a large but wasteful process will not bring improvements. As Gina Schaefer, Intelligent Automation Practice lead at Deloitte, puts it, “Just because you can automate something doesn’t mean you should.” For best results, you need to choose RPA candidates carefully.

Here are some characteristics to consider when evaluating a process’s potency for automation:

After selecting a process, the automation team needs to make sure they understand every task involved and how it fits within the business context.

Attempting to automate unoptimized processes only results in wasted efforts. When people are used to performing tasks in a certain way, they will stick to it even if it is not the most optimal solution.

Consider the following real-life process example – an employee receives an email containing text and some numbers. These numbers are the important part. Instead of simply obtaining the numbers from the email, the employee had to print this email, bind the printed file into a folder, then scan the file again, and add it to the system as a digital image. Only then others would look up the numbers from the scanned image instead of simply referring to the original email. If this process is to be automated, the steps with printing and scanning need to be dropped.

To address this RPA challenge, you can optimize and simplify a process by finding and eliminating unnecessary steps, such as redundant approvals, and minimizing the number of exceptions, if possible. This job can be done by subject matter experts, RPA developers, business analysts, and other participants.

It is not always feasible to automate the whole process. 70-80% of a process can be automated without great efforts, while the last 20% can increase automation costs five times. So, do you need to go the extra mile? Is it worth the effort?

To overcome this RPA issue, determine whether you need to automate the whole process(es). If not, identify which tasks suffice. Where RPA is not good enough but you still want to relieve your human employees from a particular duty, it is possible to combine RPA tools with AI-based technology.

There are a plethora of off-the-shelf enterprise automation solutions on the market. There is no need to reinvent the wheel if a reliable tool with all the desired features is already available for licensing. Suitable ready-made solutions can minimize the difficulties associated with RPA automation.

The automation team can compare the suitability of ready-made and custom RPA solutions to their case. For example, if you operate in a heavily-regulated industry and the candidate process includes a standard set of features, then it is worth considering trusted off-the-shelf RPA solutions. However, if your process has unique requirements, it can’t be adapted, and legacy systems are involved, then prepare to employ a custom RPA solution provider.

Without a proper infrastructure to support RPA deployment, companies will not get the desired results.

To address this RPA challenge, think of the capabilities your organization will need after automation and whether the current infrastructure can support all this. You are looking for two main aspects in your infrastructure:

Additionally, make sure your system is centralized, not easily influenced by external factors, and that any updates installed will not cause damage.

We have an example from our portfolio to support this point. A healthcare tech company based in Massachusetts turned to ITRex to automate a process on appointment scheduling and storing all the relevant information in an electronic medical record (EMR) system. Appointment scheduling was executed manually by one employee using a desktop computer. Our team was responsible for implementing a solution that will copy the clinical data gathered by this one employee from a temporary database to the EMR.

The system was old, inflexible, and slow, making it challenging for a modern RPA tool to fit in. One of the main pitfalls was Windows updates. Our script was scheduled to run at night to copy the data acquired by the employee during their day job. And every time the Windows OS updated, the script would terminate halfway into copying. To overcome this RPA issue, we included a feature to flag all data fields that were successfully migrated. After every interruption from Windows updates, the script would discard any unflagged (i.e., partially copied) data and resume the migration from the last flagged item.

Scaling seems to be one of the most prominent robotic process automation challenges, as Deloitte reports only 3% of the organizations surveyed by the consultancy were able to scale their RPA solutions.

Russ Felker, CTO at GlobalTranz, a logistics service company, experienced an RPA scaling challenge firsthand. GlobalTranz has been using automation successfully for a while, and when they attempted to scale, it wasn’t going smoothly. According to Felker, “Bots are a great technology but, like any technology, has a point of diminishing returns from a scaling perspective.” As data flow increased, the bots were not able to catch up. Felker’s team tried to assign the bots more computing resources and reduce data frequency. In the end, they considered “moving from a bot to a more integrated process.”

To reduce the hustle around scaling, implement your processes to work in parallel and independently of each other. Then you can scale by increasing the number of bots, which is easier than increasing the capacity of one particular bot. Also, make sure your RPA solutions’ architecture is simplified, and they are audited after deployment.

One of the more prominent examples of implementing RPA at a large scale comes from Japan. The country’s leading financial institutions, Sumitomo Mitsui Financial Group and Sumitomo Mitsui Banking Corporation, teamed up with UiPath, EY, PwC, Deloitte, IBM, and Accenture to roll out a comprehensive RPA solution. They successfully automated more than 400,000 hours per year across 200 different operations.

If the processes you want to automate involve input from a third party, it is best to understand how they fit in and whether their interaction with the process is uniform and consistent.

Coming back to the GlobalTranz example above, the logistics company struggled with RPA issues when automating a process involving third-party truck drivers who needed to submit different types of documents, such as delivery proof, to the RPA tool. Drivers used different layouts and didn’t bother to submit documents in a predefined order, not to mention their attempts of sending over paper-based documents instead of using the system. As a result, bots failed to process tasks with such variability, so GlobalTranz had to monitor their RPA software and improve its responsiveness.

Study the input your external stakeholders supplied over time. Are they consistent in their formatting and order of submission? Are their requests and contributions uniform? If not, maybe you can consider informing them about the changes at your company and ask them to be more consistent. If this is not possible, perhaps you can enhance your RPA with other digital technologies to handle this type of variation.

As regulations and business needs change, organizations might need to adapt their RPA tools to reflect this. Even if there are no tangible changes, monitoring automated processes will help you uncover hidden issues which the RPA team missed during implementation. Additionally, any slight change in the process itself can confuse the bots and result in errors.

Even when the system operates properly, without any modifications, it will degrade over time – the RPA tool will accumulate bugs, a database can reach its data capacity resulting in memory overflow.

To overcome this RPA pitfall, companies need to appoint someone, for example, the process owner, who will take ownership of RPA solutions and perform the following tasks:

Deploying RPA forms another potential loophole that can be exploited. Bots can access CRM, ERP, and other critical business systems. They can move data freely along different processes.

Gartner proposes the following four steps to overcome this RPA challenge:

Successfully automating the right processes will bring about many benefits, but there are RPA challenges to address on the way. Here is a checklist that will help you prepare for automation: