How to Keep Control and Balance in the RPA Race

Centralizing RPA delivery through an enterprise-wide RPA approach balances the need for speed with the need for control, helping organizations realize the promised value of RPA initiatives.

istock 646804268

Robotic process automation (RPA), a form of digital labor that automates simple transactional activities to improve speed, efficiency and accuracy, has sparked a bot revolution. Forrester predicts the RPA market will grow 41% each year until 2020, reaching $2.9 billion in 2021.

Thanks to the fact that it is quick and easy to implement, business units are eager  to race straight toward implementing  RPA delivery without thoughtful and planned enterprise-wide coordination and governance. However, this can become a serious problem that invites chaos — complete with unnecessary corporate risk, inconsistent quality, resource redundancy and, simply, too many bots.                                               

RPA implementation is a double-edged sword, explains David Kirk, managing director in KPMG’s Shared Services and Outsourcing Advisory group, whose new report, Governing the Bot Revolution,” details how companies can move beyond ad hoc RPA deployment and toward true transformation.

“The tremendous benefit is simplicity,” says Kirk, because you can quickly achieve tremendous results within weeks and even run on the desktop. But there is also the risk, as RPA expands across the enterprise and is adopted by different business units who launch hundreds of bots, that the implementation becomes disjointed and mismanaged.

It’s a balancing act of control and speed,” he continues. There shouldn’t be so much control that the organization becomes bogged down, as that defeats the benefits and simplicity of RPA. “At the same time, you have to control the environment and the risk,” he says.

Centralizing RPA delivery through an enterprise-wide RPA approach balances the need for speed with the need for control, helping organizations realize the promised value of RPA initiatives. There are three separate centralization models: 

1. Shared services model

With this model, everything is centralized, from the assessment of process to building, running, monitoring, documenting, best practices, and governance. The big plus is that the total centralization makes it easy to govern, thanks to a single, standard approach to technology. The drawback, however, is there is only one delivery place for business units competing for resources. “That tends to be the challenge  — keeping everyone happy, making them feel that their process is moving through the pipeline,” says Kirk.

2. Center of Excellence model

On the other side of the coin, with the Center of Excellence model, best practices are centralized, but beyond that businesses are on their own. “They come to the center for methodology, guidelines and tools, but they can run as fast as they want, with a pipeline unique to their business line,” says Kirk. The fact that business units can build at their own speed is a tremendous benefit, but the drawback is that it is clearly harder to manage and govern. In addition, it can lead to duplication of costs and tools, as well as underutilized resources.

3. Hybrid model

The hybrid model offers a bit of both from the Shared Services and Center of Excellence models, with a centralized organization that does all the best practices and governance, but also has business services capabilities that units can use or not use as they choose. The advantage is that this model allows the business to run quickly, but only if they have the appetite, knowledge, and  capacity to do so, says Kirk. “If the business unit is well-versed in digital labor, it may only turn to the center to follow corporate risk guidelines and leverage best practices,” he says. “But another may not want to have  digital labor capabilities in its unit and may want to leverage the fully centralized capability.”                               

To varying degrees, all three models centralize some amount of governance and some amount of RPA shared capabilities and knowledge. But while one size does not fit all, the hybrid model is most likely the end outcome for a typical organization because it provides the business units with a high level of knowledge, efficiencies, and governance while allowing them to provide as much of the RPA capabilities as they choose to support.       

No matter what model is chosen to implement RPA, centralization of digital labor is not always about saving money. Other enterprise-wide benefits include quality consistency, risk management, market intelligence, flexibility and scalability and empowering the business. Organizations that are on the path to adopting an enterprise-wide RPA solution model should keep in mind the following tips for success:

1.    Create a flexible plan: An organization doesn’t need to immediately move to full centralization, but there should be a vision of where it’s going to go. According to “Governing the Bot Revolution,” “A road map should be in place for evolving to a hybrid model in order to enable more rapid scaling while preserving core knowledge and standards.”  Today’s model might not be the best answer for tomorrow, Kirk points out: “There’s nothing wrong in saying right now we have a shared services model and this is good for today,” he says. “But can you put a hybrid model plan in place in advance, so you can design for possible end goal? Our message is to build a plan, but have flexibility built in so you can evolve.”

2.    Set development standards: A lack of scripting standards may lead to longer development time, inconsistent approaches, disparate results, quality issues, and maintenance nightmares. That’s where building standard approaches is essential, says Kirk: “We see this even at the coding level, where the bots look like humans, but it’s a good corporate standard not to hard code things like user IDs and passwords into the bot,” he says. “It’s about how you’re going to handle access, control and capability.” 

3.     Make sure to document: It’s tempting to focus on building bots without realizing that they can break and the business will need to return to a manual process. So, you don’t want knowledge to be lost. “The bot may break and no one remembers what they used to do and what the task used to look like,” says Kirk. “We advocate a detailed process that documents the current manual process as well as the resulting automated process — it may seem painful to record each step, but it pays off down the line.”


Copyright © 2017 IDG Communications, Inc.