The agilification of automation
March 4, 2020
Many businesses try to implement RPA to realize its full savings potential. But only a few organizations manage to effectively scale RPA. To do so, you need an effective methodology and structure, be agile. But where do RPA and agile connect? How can RPA and agile help you execute processes in a more intelligent way?
To answer these questions, we will go back to the properties of RPA and agile and show you how our operating model can guide your RPA implementation.
<<< Start >>>
<<< End >>>
Businesses are being shaped by all kinds of forces in this fast-paced and ever-changing world. While some changes come from the outside, others come from within organizations. Two major catalysts of internal change are the agilification of things and automation.
We found that these two concepts bring their own sets of opportunities and challenges, and what's more, that they have a lot of common ground. But how exactly are they similar and how can they reinforce each other? To explore these questions, we tap into our knowledge and explain our thoughts.
Please be aware that this is the second article in our automation series on RPA. Read our first article below.
<<< Start >>>
Related articleChange is coming: what intelligent automation means for people
<<< End >>>
RPA (Robotic Process Automation) market revenues worldwide more than doubled from $271 million in 2016 to $626 million in only one year (by 2017) and are estimated to rise to $3 billion by 2022. The potential savings that companies are likely to experience through RPA by 2025 are between $5 trillion and $7 trillion. In terms of automated work, RPA technology may be able to perform tasks equal to the output of 140 million full-time employees (FTEs) by 2025.
Thanks to the rapid growth and compelling benefits of RPA, many organizations are interested in implementing RPA effectively and realizing its full potential. In fact, only 13 percent of RPA initiatives are scaled throughout the whole organization. In order to achieve a scaled RPA implementation, an effective and consistent methodology and structure throughout the whole organization is essential.
<<< Start >>>
of RPA projects are scaled
<<< End >>>
Our agile center of excellence (CoE) operating model offers you a structure and best practices that address the characteristics of RPA implementations and helps you deploy RPA in an effective and seamless manner. In this article, we’ll explore the connection between agile and RPA. We'll also tell you how the CoE operating model and best practices can be a solid guide for your RPA implementation.
At its core, RPA uses software that mimics and automatically performs tasks normally performed by human workers. Usually, the tasks automated by RPA are partial or end-to-end business processes. RPA adds the most value in high volume, repetitive, structured, and rule-based processes.
Although the development time can vary with scope and complexity of the process, functioning Proof of Concepts (PoCs) can be delivered within just a few weeks or even a couple of workdays. The benefits of RPA have triggered a fresh wave of transformations in all sorts of industries.
The main benefits of RPA, therefore, are:
<<< Start >>>
Related articleRobotic process automation: 4 huge benefits for businesses
<<< End >>>
Despite the potential of RPA to deliver fast business benefits, some organizations struggle with making the right structural changes to support the deployment of robots. Waterfall delivery techniques are not sufficiently tailored to the implementation of large-scale RPA projects.
Why is that? To answer this question, we need to understand the two leading paradigms on project management: agile and waterfall.
Typically, traditional delivery approaches consist of non-iterative, sequential stages, usually involving:
These waterfall approaches are beneficial when the requirements are clear, the technology is well understood, and you can strictly set clear deadlines. However, these characteristics can hinder the implementation of software projects—such as RPA—for a couple of reasons.
First, waterfall methods generally focus on the large-scale deployment or ‘launch’ at the end of a stage. Hence, there is no room for flexibility or iteration. Robots, especially if not well understood, don't always produce the desired results. Meaning that focusing on one large deployment of robots, brings great risks. Once the robots are already in the new stage, it is very difficult to correct for changes, which were not well thought out in previous stages.
Second, traditional waterfall methods generally only launch robotics deliverables in the implementation phase. This means that throughout the planning, analysis, and design phases, no robots are being deployed at all. When waiting for the completion of these phases before releasing the automation, organizations will miss out on quick wins and valuable feedback loops.
Agile is often seen as a solution to some of these shortcomings of the waterfall approach...
Agile can be defined as "a range of values and principles of iterative and incremental approaches to software implementation". In contrast to the waterfall method, agile embraces adaptive planning, continuous interaction, close collaboration, and early delivery.
The benefits of agile include:
Agile overcomes the rigidity of waterfalls’ requirements for scheduled communication with specific content. Agile focuses on meaningful interactions and exchange. Close collaboration between the Business and IT enables alignment between different project stages and teams.
In agile delivery, documentation deliverables can be flexibly adjusted and written to contain only essential information. In contrast, the traditional waterfall method requires a detailed project plan, requirements specifications and design. This produces a lot of documentation that does not necessarily add value.
Within the agile approach, the client can be involved and collaborate throughout the whole development process, rather than just at the starting and end phase. Client needs and requirements are thoroughly studied and sometimes challenged. The setup ranges from including the customer at intervals for periodic demos to having an end-user as a daily part of the team and attending all the meetings.
In Waterfall, change is seen as an expense to be avoided by developing detailed, elaborate plans. In contrast, the agile approach gives flexibility for incorporating new features, improving existing ones and generally making use of feedback loops. Agile project delivery can easily be adapted and put together flexibly, just like building blocks.
So far, we identified two areas of limitations: the limitations of the traditional project management approach to application development and the limitations of delivering and upscaling RPA implementations. Possibly, the solution lies in looking into the fit of RPA and agile, as both have flexibility and iteration as guiding values. So how do RPA and agile fit together? Let us look at the agile values from the perspective of RPA delivery.
RPA delivery is in unison with this motion, as unlike some heavyweight IT projects where most of the work relies on a development team, RPA delivery needs close cooperation of business and IT. It requires IT expertise such as developing, testing, system security, as well as business knowledge in terms of identifying business needs, modeling processes, and assessing new IT solutions.
RPA projects require minimal documentation, which is reflected in the fact that–according to our delivery method–two major documents are needed for the actual development: The Technical Design Document (TDD) and Process Description/Design Document (PDD). These provide details on the design and working of the robot and are refined continuously throughout development and signed off only when complete.
Similarly, the Subject Matter Expert (SME) is involved throughout the whole RPA lifecycle, not only during the initial discovery. This is essential as often throughout development the process (according to the developers) is not fully described yet or some logic is missing. Thus, the SME is a point of contact for clarification right until User Acceptance Testing (UAT), which serves as a final moment of validation before delivery.
While current applications of RPA can be replaced by automated software solutions using APIs, RPA is designed for continuous delivery. Once a minimum viable product is ready, it will be deployed in the production environment. Additional features can be added at any later point.
The early and frequent delivery approach can contribute to getting feedback faster, which allows the team to make quick adjustments. So, breaking down the delivery into multiple waves that can be tested individually instead of one ‘monstrous’ solution that fails to function, reduces the risk of having a lot of defects when the final deliverables are done.
During the discovery of which processes to automate in a business, analysts explore opportunities for process optimization before automating them. As a result, the automated processes take often more efficient paths.
So, RPA challenges businesses’ (operational) processes in terms of how they used to be. Resulting process changes and redesign can bring challenges when adopting or integrating those new automated processes.
<<< Start >>>
<<< End >>>
That is why especially large-scale implementation of RPA should follow a tested and proven model. Our agile CoE operating model ensures that the deployment of robotic operations is executed in a seamless manner.
To scale RPA initiatives, the integrated agile component has been shown to substantially improve the implementation trajectory. The iterative, adaptive, and collaborative nature has provided particularly valuable contributions to the automation journeys of our clients resulting in three best practices.
The decomposition of operational processes in iterative sprints provides quicker returns of investment throughout the agile RPA delivery process. Clients often demand RPA implementation projects for highly transactional and high-volume tasks (e.g. managing invoices). Automating even chunks of a process already allows organizations to take over a high volume of tasks. This is what we like to call the Pareto principle of RPA.
This principle states that automating the first 80 percent of a process delivers the most value. The remaining 20 percent of that process only consists of low-volume, dirty cases. Therefore, the target should be to automate 80 percent of a process, instead of the full process, as those 80 percent tend to cover the high-volume cases. Our agile CoE operating model is a suitable framework for guidance.
For example, for a large Financial Services client, we applied this model to manage their large-scale RPA implementation. After approval from all stakeholders, the client already saw productivity and efficiency gains by implementing 80 percent of the to-be automated process instead of waiting for the final product.
The iterative feedback sprints during the agile RPA delivery process nurtures resilience to regulatory pressure and enhances the adaptive capacity. Clients often operate in an uncertain and dynamic environment. Moreover, large-scale RPA implementation projects are relatively complex and making alterations is a time-intensive and complicated task.
<<< Start >>>
<<< End >>>
Therefore, fast feedback cycles let organizations to quickly adapt their processes to the changing demands. Our model offers a framework to break down process automation into smaller bits, implement some bits, validate their performance, adjust where necessary, and apply the lessons learned before deploying subsequent robots.
As an illustrative example, we managed a large-scale RPA implementation project for a leading financial institution. Here, by utilizing adaptable Lego puzzles, the complex technical RPA solution could be explained in simple terms to client stakeholders.
Having simplified and broken down the RPA initiative, stakeholders could provide valuable feedback for improvements. These activities offered better feedback and decision-making processes, which enhanced the client’s understanding of the underlying consequences of change without taking any risks.
The endeavor of optimizing, remodeling, and automating business processes can bring various unexpected opportunities and challenges, which vary depending on the complexity of the process and the automation technology. Some tasks, for instance can be automated through very simple automation and quickly delivered with rather low risk.
Automation of more complex tasks, for instance when dealing with image recognition or unstructured data, may involve additional technology and tools, such as Optical Character Recognition (OCR) or Natural Language Processing (NLP).
The agile CoE operating model helps identify and generate value from automating the low-hanging fruit first, while drawing on that experience and tackling complex challenges later.
We have seen that agile is a promising methodology for implementing RPA projects. However, agile can be difficult to properly apply and can bring other challenges, too. These include high time effort, for instance for ceremonies, as well as limited readiness of the client to properly implement a project in an agile way.
<<< Start >>>
Related articleWhy many agile transformations fail, and how yours will succeed
<<< End >>>
Some experts say that agile is best applied in cases where solutions evolve through exchange and collaboration between self-organizing and cross-functional teams as it promotes flexibility, adaptiveness and evolutionary development.
On the contrary, it is difficult to apply RPA to cases where the business requirement is not fully defined yet. This is because RPA is often used as a temporary, intermediate or short-term solution to existing, wider business problems.
Both agile and RPA have brought along new opportunities of how to execute processes in a smarter and better way. Their potentials can be leveraged even more so, as both the agile methodology and the RPA delivery approach have been around for long enough for best practices for their implementations to have been developed.
Our advice to both our clients and colleagues is: Learn from what agile has thought us and automate intelligently by focusing on high-volume parts of the process instead of striving for completeness; by developing smaller process bits that can be directly validated and adjusted based on feedback; and by starting with lower complexity and building experience to scale up towards higher complexity use cases.
We have witnessed the potential of those best practices for our clients, there are, however, scenarios where neither agile delivery, nor RPA can provide value or be a remedy for deeper-rooted inefficiencies. Being aware of both benefits and challenges is the first step for our clients to make informed decisions.
One thing we know for sure, agile delivery and RPA are not just introducing new methodologies for process execution and optimization. They are disrupting our mindsets of what constitutes work and how work is to be performed by introducing flexibility, freedom and value-creation mentality for the whole workforce. And that is the true benefit!