Where It Starts to Fall Apart – Losing the ROI on Your RPA Initiative


Expect the best, plan for the worst.

Robotic process automation (RPA), which uses software bots to automate rules-based business processes, presents enterprises with a low-risk opportunity to build new capacity and reduce costs.

Do you need to automate how you move data from one environment to another? Check! Do you want to automate complex workflows in an enterprise resource planning (ERP) system? Check! Do you wish you could automatically pull reports from various platforms and perform financial validations into one report? Check!

When teams use RPA, they are not only freed up to contribute higher-value tasks, but they see quantifiable benefits that come from a built-in audit trail and a dramatic improvement in accuracy.

RPA is appealing because it automates existing proven processes, and the automation performs the job accurately day in and day out. It is quicker to implement RPA than traditional software development because the existing process doesn’t need to be redesigned prior to implementation – and because RPA uses the same programs and software interfaces a human does. Setting up an RPA tool allows existing business rules and controls to be followed, and the “virtual worker” can be assigned the same security and access as a human worker.

RPA centers of excellence (CoE) can examine a business’ needs by considering several sets of factors, including the complexity of automating a process and the viability of the return on investment (ROI). Based on the business’ priorities, a CoE can create an RPA deployment strategy that factors in the start-up costs, including software licensing and talent.

Without governance controls in place, it’s easy for that ROI to erode. Here are five challenges businesses need to watch for:

  1. Solution designs going rogue. Prioritizing automation and realizing ROI begins with an initial assessment of the viability of a process, and the RPA team makes certain assumptions about scope. If the design analyst and process owner try to solve all the problems of the world beyond those assumptions made in the assessment, the level of effort to build an automation grows exponentially beyond the plan. Often, a Pareto (80/20) analysis is used to determine how many exceptions to the core functionality should be included in the design. The idea is that 80 percent of the process requires 20 percent of the effort to resolve. Teams should remember that solving for the remaining 20 percent of exceptions can take five times the development effort to build!

  2. Communication doesn’t happen. Stakeholder participation is crucial to the success of any RPA implementation. If you don’t have a communication plan ready before your start date, stakeholders don’t know what is expected of them, which deliverables each resource or unit owns, and what the timeline is for overall changes and deployment. Create a project plan and communicate it. Conduct sprint reviews during each stage of the build process and communicate progress and roadblocks. Have a checklist and review for each working session: What is the agenda? What can I expect to happen today? What are the inputs, and what are the outputs from today? Do I have any follow-up tasks I need to do or assign to someone else to escalate?

  3. Test data and test scripts not prepared in advance of build. Before developers write a single line of code, they should be able to review the test plan written by the process owners — the experts of the process — to compare their understanding of the process, the solution design and the test scripts. This reduces the risk of a user discovering process requirements during the testing phase that were not discovered during process definition and design, eliminating unplanned work that would increase the cost of an automation and reduce capacity available for other development work. No one wants to get to the finish line and find out they were disqualified from the race.

  4. Forgotten organizational change management. Organizational change management (OCM) for RPA happens much faster than OCM for the traditional software development lifecycle. To achieve the full capacity and value of automation, the business unit may need help understanding what factors and data points the CoE will measure, so it can plan for the expected support required of the business once the automation deploys. Many CoEs are set up to be a shared service that uses a collaborative approach – the business still owns the process and is responsible for handling exceptions that are routed out of the automation – and the business shouldn’t be tossing the process over the fence for the CoE to manage.

  5. Poor planning and implementation. Manage each automation like any other project – with a start and finish, timetable, resource commitments, milestones and deliverables. Failing to manage a project means it will manage itself, and the overall timeline will extend far beyond the original schedule.

ISG helps organizations optimize their RPA journey and accelerate results to dramatically improve business performance. Contact us to see how we can help you optimize the RPA experience.


About the author

Heather Lapinski

Heather Lapinski

Heather Lapinski is a consulting manager in ISG’s Robotic Process and Cognitive Automation practice, with 20 years of experience in leadership roles in process automation, IT program/project management, operations, process re-engineering/design and outsourcing. As a project lead in the practice, she has led complex RPA implementation projects for global clients in the insurance, consumer packaged goods, and manufacturing sectors. Each of these RPA projects included managing large teams to implement RPA from process assessment and software selection/installation, through training and automation configuration/deployment, to center of excellence training and implementation.