The 5 P'S of Robotic Process Automation (RPA)
The 5 P'S of Robotic Process Automation (RPA)
RPA is a hot topic: the technology is rapidly gaining popularity; every company wants to adopt the technology and see fast return on investment. But in their quest for quick savings, I have seen organizations too often rush in and use RPA in wrong ways. These '5 P's of RPA' (Purpose, People, Processes, Platforms and Procedures) will help you to ask the right questions before diving into implementation.
Purpose: if you don't know why… then don't.
My first advise is to avoid starting with RPA without any purpose. Of course, RPA is exciting, causing some companies to rush into automating as many processes as they can, as quickly as they can. But before starting a project, make sure all stakeholders are on board to decide which benefits to realize and how to measure these benefits: for example, ROI, cost savings, manual effort savings, success rate or productivity, accuracy or compliance. Take your time to start with RPA and do not only focus on innovation as such.
People: RPA is amazing… unless no one knows anything about it.
I once set-up and coached an RPA Center of Excellence (CoE) at the customer to centralize RPA governance and capabilities and to communicate success stories throughout the entire organization. We appointed the RPA sponsor and RPA governance board with head of RPA, IT and business representatives (e.g. from HR, Finance, etc.) and a robotic operating team to analyze, develop and monitor the deployed solution.
In this way we were able to embed RPA in the organization, transfer knowledge to all team members but also monitor and allocate the results of robots per business unit. Together, we managed new processes, defined the delivery methodology (define, design, build, test and deploy the solution) and procedures on support, security and compliance.
Processes: don't try to solve problems that can't be fixed.
RPA is not a silver bullet for all challenges, neither will one tool automate all processes. RPA is based on rules so in certain cases it might be better to opt for another solution, add AI, execute a process manually or even delete a process. A good candidate for RPA is for example the processing of high volumes of structured invoices in SAP. The more unstructured the invoices and the higher the number of suppliers, the higher the need for e.g. OCR engines and machine learning.
And if you ask my advice: never automate a broken process. Plan a process discovery phase before process selection and make sure documentation and subject matter experts are available to avoid automating a process in a limited timeframe with too many steps, applications or business segments. Make sure to define beforehand process logic and exceptions, the trigger for the robot (new email, new file in a folder, etc.) and if approval from a business user during execution is required or not (attended execution vs unattended). And finally, start simple, add more complexity gradually, prepare a detailed UAT plan and sign-off with all stakeholders on the hyper care plan and RACI before go-live of each process.
Platforms: try to drive an engine without a car.
Sorry to disappoint you, but RPA will never work without the necessary infrastructure. Define for each process the number of robots, machines, users, environments, drives and target applications you need. Diving too fast into implementation will result in re-work, connectivity issues to retrieve data and execute processes, re-installation of robots and a lot of time lost. Make sure to get business, IT, infrastructure and security teams on board to discuss changes, solutions design, requirements, back-up scenario's, audit trails and security and compliance policies.
Procedures: formalize each P of the 5 P's of RPA
Your project will probably fail, unless you state all the previously mentioned in an RPA bible. Define procedures for the RPA strategy, business benefits, roles and responsibilities, coding practices (e.g. readability, naming conventions, reusability, release management and version control), process management, delivery methodology, change management, support procedures, infrastructure, security and compliance, just to name a few RPA topics. Without proper communication, business and IT will never realize they have to solve exceptions in business data or deviant behavior in the applications. This can be avoided when all stakeholders sign-off on the hyper care plan and SLA's. Not to mention the confusion about sensitive data in queues, archiving policies or access for robots and developers in production.
Finally, remember that quick wins can be enticing but consider these lessons learned before building the proper RPA solution. Companies should expect RPA deployment to take from six to 12 weeks, followed by short- and long-term hyper care scenarios to tweak the robots and add efficiencies. After that, companies can expect to see the full ROI realization.
Contact us
and transform your business to tackle the digital age more efficiently.
