Look Before You Leap: Best Practices to Avoid RPA Pitfalls
Mia Urman, Oracle Ace Director and CEO of AuraPlayer
Implementing Robotic Process Automation (RPA) can be one of the best business decisions any organization can make. RPA’s ability to ease bottlenecks, increase efficiency and productivity, and boost the bottom line are just some of the reasons businesses across all sectors are implementing automation solutions.
In an earlier post, we tackled the fundamentals of RPA for organizations embarking on their automation journey. I shared initial considerations for getting started, such as whether or not RPA is right for you, selecting the best workflows to automate, and what to look for in an RPA provider.
But make no mistake – those considerations are only part of the process. Implementing RPA without a well-considered plan can lead to a path filled with frustrations, delays, and setbacks. Luckily, with all of the available solutions, it’s not difficult to launch a successful automation journey, and it’s never been easier to automate Oracle Forms and EBS applications.
In this follow-up article, we’ll delve deeper into the RPA journey, examining best practices — from planning to implementation and beyond — while sharing insights on how to circumvent common pitfalls. Deploying automation should be exciting for any organization, and by keeping these considerations in mind, the experience can be both positive and successful.
Read on for a drill down on these best practices:
- Assess the value and costs – Choose a high-value use case and determine the long-term costs to deploy and maintain the automation.
- Plan the journey, draw a map – After selecting a use case, map out the workflow step-by-step.
- Identify the players and decision points – Consider all the actors and triggers, and ensure the integrity of the data.
- Account for outliers – Have a plan for all possibilities.
- Build on stable ground – RPA works best with a stable system.
- Get end-user buy-in – these are not just boardroom buzzwords. Get everyone on board for your unified vision, and delegate a C-suite sponsor for your RPA strategy.
- Plan the long-term endgame – Achieve your goals through routine monitoring, testing, and analysis.
- Select the best tool for the job – With dozens of RPA solutions out there, choose one that can reinforce your organization’s vision and support your system’s needs.
Assess the value and costs
Make sure to home in on high-value use cases that can bring maximum ROI rather than automating too much too fast – or automating the wrong processes. Not all workflows are created equal, and not all business cases should be automated.
Automated workflows should bring you a good bang for your buck; an ideal candidate is a use case that is repetitive, routine, and simple. (See our previous post for tips on picking a good use case.)
Avoid use cases that are one-of, dynamic, or require decision-making. Any attempts to automate them will likely lead you down a dead end of constant fixes and wasted resources, inevitably derailing your overarching RPA strategy and business goals.
However, this step requires more than identifying a high-value use case. When evaluating the value of your RPA strategy, there are four major costs to keep in mind:
- Infrastructure setup and licenses of the RPA solutions
- Building, creating, and testing the RPA automations
- Monitoring and running the bots long-term
- Maintenance, debugging, and error handling – including the cost of any necessary changes
In the early stages, of course the focus is mainly on costs surrounding creating automations and the required licensing. However, as time goes on, it becomes clear that the major value and ROI of RPA will come from implementing stability. The goal is to establish automations that are not prone to instability or rapid changes, therefore avoiding the need to repeatedly re-record and test workflows.
Once you’ve calculated your costs and identified one or two simple use cases to pilot, you’re already on the road to success. But it does not end there. Sometimes there is a big gap between the end goal organizations are trying to automate, and how the business processes the actual “flows” in their operation. That is why mapping out the workflow is a critical step that cannot be overlooked.
Plan the journey, and draw a map
When starting an RPA project, don’t try to boil the ocean. Rather than biting off more than you can chew, break down processes into micro tasks and deploy automation in stages.
In their race to the finish line, organizations often don’t invest enough in planning. They pay the price later on with system errors, backtracking, unexpected results, and popup windows that cause delays. Data and analytical tools can help to track processes through the system and identify use cases that are good candidates for automation.
Successful automation requires mapping a workflow step-by-step, plunging deep into the nitty-gritty. At each stage, the decision point and data set need to be considered, regardless of how inconsequential these steps may seem. This is because you are not just automating the ideal workflow concept; you are running an automation of a human workflow with all the checks and balances.
Bots need controls telling them how to act when errors or exceptions exist, what information needs to be conveyed, and to whom. For example, bots can be programmed to send an email or notification to the appropriate employee when its workflow is disrupted.
Let’s look at the example of sales order creation, which is an ideal use case for automation. It is both an end goal and a process that can be broken down into smaller “behind-the-scenes” steps that require specific data inputs. To create a sales order, customer data, inventory, and shipping must be retrieved and updated. But the order cannot be completed if that customer doesn’t yet exist in the system. So, if they don’t yet exist, to properly automate this process, you need to set protocols for creating new customers in the system first. This is just one example of the need to plan and map out the flow.
Identify the players and decision points
Once you’ve established the workflow, it becomes critical to ask: What does the workflow require to run smoothly? Each stage of the workflow diagram must include all the relevant decision points, data sets, and actors. Seamless automation depends on a perfect marriage between the required triggers and data. Data integrity is key for the bot to do its job and generate error-free results.
For example, if our goal is to automate Oracle EBS invoice generation triggered by an email with a PDF invoice, we must first verify that the PDF contains all pertinent information (such as customer name and number, date, etc.).
Therefore, it is important to consider how the RPA automation will be triggered and implemented during this step. Generally speaking, there are three types of automation.
First, unattended automation is essentially batch processing, in which RPA bots run from start to finish with no human intervention.
The second type, attended automation, involves cooperation between a human actor and the bot; some parts of the workflow are automated, while others require human intervention. Attended automation is ideal for processes that require more intricate handling, approvals, or validations along the way, before the bot continues to the next stage.
The third type is cognitive automation, which integrates automation with artificial intelligence (AI) and machine learning (ML), enabling bots to gather data, make decisions, and scale automation. Although cognitive automation is not discussed in this article, it is something to keep a watchful eye on as it is the future of automation. A strategic plan should determine which type of automation is suitable because each type requires its own preparation and monitoring.
Other important questions to consider are: is there any approval process or check validation in the proposed process? Who are the actors who need to be updated when the automation is triggered?
Using our example of automating an Oracle EBS purchase order, there may be operations, inventory, and customer support personnel who must be notified, and automations can be set up to trigger an email or notification to do so. In some instances, attended automation can bring greater ROI than complete, unattended automation. Alternatively, other use cases are suited to unattended automation that follows a rules-based logic and requires no employee intervention, such as batch data loading.
Account for outliers
Account for outliers and the full spectrum of possibilities; what happens when things veer off course is as important as when things run smoothly. For instance, with an Oracle EBS purchase order, what happens if the order is not created to completion? Should it be saved as a draft? Should any changes be rolled back? Should an email be generated to notify relevant personnel? Should you exit the whole workflow?
Consider how snags in the system might affect operations and data. You don’t want human actors going in to fix errors all the time. This is why it is so important to map out each workflow stage using response data to ensure a smooth and seamless process.
Build on stable ground
You can’t automate on a rocky foundation. Automation needs a stable infrastructure to run smoothly. If you’re planning on upgrading your system or installing updates or patches, you can go ahead and design your RPA workflows – but wait until the system is stable to deploy bots. Bots are very sensitive to operational or platform changes; even slight changes to processes, data, and user interfaces can cause bots to malfunction.
It is not a good idea to alter methodology and workflows while automating at the same time. Once organizations begin drilling down their operational processes, it’s common for management to start thinking about ways to “tweak” workflows to make them more efficient. But if that is the case, these workflows shouldn’t be automated until the new process is completed; doing so runs the risk of malfunction. An ideal use case should have clear rules and set parameters that lead to clear outputs.
Get end users’ buy-in
The long-term success and ROI from RPA rely on the organization’s people. Your strategy should include ways to get your employees on board as part of your organization’s overall goals.
When employees hear that management is looking to implement RPA, they often become anxious about being replaced by robots. Avoid this anxiety by bringing people together through transparency and a shared vision. Delegate a C-suite figure to be the face of this innovation to clearly convey – to staff at all levels –the automation process, timeline, goals, post-deployment maintenance, and role of human actors in the process.
Adopting bots doesn’t make a team of humans less valuable. Show employees how RPA can make their jobs more interesting by freeing them up to focus on high-level work and giving them opportunities for advancement by learning new skills. Additionally, using citizen development tools can be another way to involve employees throughout your RPA journey, enabling them to capture processes or design simple data entry screens, all with little technological skill needed.
While the decision to implement RPA is made in the boardroom, a partnership between management and the end users is the key to success. Super users know the ins and outs of the system – and all of its quirks too. They can be a significant asset during the planning phase, providing insights that help limit errors and malfunctions later.
Including end users in the planning phase allows employees to take on responsibility and ownership of the automation process. This will catapult RPA beyond being just a buzzword and encourage widespread acceptance and engagement.
Plan long-term endgame
Your grand plan should include a vision for success and how it will be measured. Once you’ve identified a high-value use case that can impact your business operations, focus on monitoring and refining it over time.
Think of bots as part of the team, just like any other employee. Similar to human staff, bots must be set up with varying levels of security, permissions, and access to complete necessary tasks. And along with the huge value that RPA brings, bots still require monitoring and re-calibration. As humans run into obstacles during their work and need to adjust accordingly, so will bots.
While some setbacks are normal, the true success of your RPA strategy will be determined by how often these setbacks and disruptions occur and how long it will take to get them back on track. Post-implementation issues might stem from changes in the user interface, updates in controls or policies, or overlooked steps during the bot development process. For this reason, adopting a long-term perspective when selecting your tools and crafting your strategy is crucial to minimize friction and ensure minimal downtime.
Establishing a robust strategy to monitor the bot’s performance from the outset, and verifying that it aligns with your goals, can help you mitigate future frustrations and disappointments.
There also should be a clear and consistent methodology for updating bots when a change is made to the back-end system, including how to test them in each environment – from development, to QA testing, to production. Testing must be conducted continuously throughout the bot’s life cycle to ensure that it is capable of effectively responding to any errors or exceptions that may occur and maintaining optimal performance.
Select the best tool for the job
In addition to the considerations and preparations outlined above, ensuring your RPA solution fully supports your Oracle system’s complexities and dynamic nature is equally critical. You can do everything right and have everything mapped out, but without a compatible RPA tool, even the best-laid plans can go awry. Each tool has its niche and specializations, whether automation for specific back-end systems or packaged applications for Oracle or SAP.
Ask the following questions to ensure that you are selecting the best automation tool for achieving your goals:
- How does the tool capture Oracle workflows?
- Can the automations handle dynamic pop-up windows and hidden fields based on personalizations?
- Is the solution able to deliver Oracle error messages in real time?
- Can the automations withstand UI changes, patches/installs, and other environmental changes?
- Can the tool move quickly between environments – from development to staging to production?
If you feel overwhelmed by this article and the amount of effort RPA entails – don’t worry! Keep in mind that there is a wealth of information and expertise out there. The right RPA vendor for the job will have experience working with systems similar to yours, and they will guide you through this entire process from start to finish.
Another option is to bring in an RPA consultant to oversee this journey, rather than expend resources better suited for operational activities. Also, it’s worth considering pre-built RPA accelerators or templates that support your Oracle or SAP applications which can easily be customized for your more complicated workflows.
RPA automation is a great way to streamline complex back-ends, such as Oracle EBS applications. Automating simple, repetitive, and manual tasks can bring substantial results, specifically in use cases that rely on batch data entry, such as invoices, sales orders, and purchase orders.
There is no question that RPA automation is an incredible adventure and worth getting excited about! But the key is to move ahead with a measured, calculated plan with proper direction and preparation.
With a dedicated cross-silo RPA team that remains focused on the grander RPA vision, organizations can achieve optimal results and ROI by deploying automation piece by piece within a framework for continuous monitoring and measurement. By taking the right steps and considerations, the possibilities are limitless. The sky is truly the limit.