chris - Fotolia

Why CIOs need to establish an automation CoE

With organizations continuing to automate business processes across several departments at a rapid pace, it's time they consider implementing a Center of Excellence. Here's why.

The path to the cutting edge of automation is paved with failed efforts. According to Gartner, most enterprises struggle to scale past five robotic process automation (RPA) implementations.

An automation Center of Excellence (CoE) can help focus the attention, technology and processes required for successfully scaling automation initiatives.

"[CoEs] drive the organization's technological innovation initiatives and improvements, which accelerates the production of deliverables and improves the customer experience, whether that customer is another department, an internal business unit or a services client," said Deven Samant, global head of CoE at Infostretch, a digital engineering professional services company.

Additionally, sharing results, best practices and techniques across different business units aligns an organization's business and technical goals.

As an organization grows and becomes more complex, it's critical to develop expertise and set standards in different areas of practice that are part of daily operations. According to Samant, this requires a dedicated team that focuses on improving the quality of business processes, standardizing the execution of deliverables and bridging the gaps so the knowledge is shared across departments.

It's also important to drive enthusiasm across the company. Steve Miller, vice president of data, analytics and automation at Xerox said they launched an internal campaign that focuses on transparency and enrolling employees to be part of the change. "It's important to get employee buy-in, and the way to do that is to show them how automation can benefit not just the company but also them personally," Miller said.  

What does a CoE consist of?

"CoEs are typically competency centers that lead the way in exploring and adopting new technology tools, techniques and methodologies," said Prasanna Velayudham, IT transformation director of application management services at Capgemini North America.

Prasanna Velayudham head shotPrasanna Velayudham

Successful automation CoEs should have a cross-functional team comprised of change agents with key skills not only in digital technologies, but also in the areas of process, governance and organizational change to drive adoption in the enterprise.

Another critical function of an automation CoE is knowledge management and the ability to develop and maintain a catalog of reusable assets, standards, best practices and essential lessons that will enable business functions to be successful with automation programs.

"The automation marketplace is becoming overcrowded and CoEs need trusted sources to declutter the noise and focus on building core capabilities that will benefit their enterprise," Velayudham said.

Many types of CoE models

Ohad Barnoy head shotOhad Barnoy

Every organization has a different set of circumstances and they should structure the CoE in a way that works for them, said Ohad Barnoy, chief customer success officer at Kryon, an RPA vendor. At one end of the spectrum organizations adopt a centralized CoE, while at the other end, organizations adopt a more federated or distributed model.

"A centralized model defines a standard way of working, a governance model and method of interaction for implementation," said Manish Sharma, group chief executive at Accenture Operations.

A distributed model uses highly autonomous teams, which are housed within each business unit or function, with a delivery focus specific to that business unit or function. Sharma sees this model improving with a shift toward highly distributed ways of working along with better technology, tools and practice for citizen developers.

Rahul Talwar head shotRahul Talwar

Either way, these efforts generally start out small and then scale up as the automation CoE learns how to deliver success. A centralized model is more focused, while a decentralized approach can help to increase buy-in.

Rahul Talwar, solutions director at FortressIQ, has seen many organizations spend 12 to 18 months before getting a single bot into production. He finds that it is important to obtain buy-in from process owners and incentivize them accordingly.

Nainesh Ganu head shotNainesh Ganu

Scaling happens when business users see the value and they come up with new use cases and automation ideas, said Nainesh Ganu, intelligent automation advisor at Client Solution Architects, a management consultancy for the DOD.

"Business users should not feel threatened about their job or role to increase automation use," he said. He led a CoE at a retail client where they were able to scale the program to over a hundred bots across the globe because of the role the CoE played in communicating and coordinating with business users, IT and leadership.

Scaling and business user confidence encouraged the CoE to expand and enhance capabilities in AI, chatbots and intelligent document processing.

Characteristics of a CoE program

Deven Samant head shotDeven Samant

Deven Samant said some of the characteristics of CoE programs include:

  • Best Practices: standardizing the approaches, processes, tools, methodologies and knowledge centers;
  • Technology: identifying, innovating and accelerating reusable models;
  • Experience: creating proofs of concept and marketing materials for organizational growth;
  • Support: delivering the innovations and templates and providing continuous support and improvements to anticipate and prevent any challenges;
  • Training: training the teams and mentoring the organization to apply the standards and processes set by the centers of excellence; and
  • Governance: reviewing the deliverables and making sure they align with the organizational standards and best practices.

Different roles in implementing a CoE

It's also important to consider how new technologies like RPA are different from classic technologies in terms of the ability to reshape organizational structures.

Jon Walden head shotJon Walden

"The transition of work to a digital workforce requires management, which is why an operating model and CoE are critical," said Jon Walden, CTO at Blue Prism Americas, an RPA vendor. He noted that key roles in these efforts need to include everything from architectural technologist to evangelists.

Automation CoEs also need to play a key role in uncovering automation opportunities through a combination of evangelism and building the technology infrastructure.

Michael Beckley head shotMichael Beckley

"An important part of this process is educating people about how to identify automation opportunities," said Michael Beckley, CTO and co-founder of Appian, a low-code tools provider. Once the automation opportunities are identified, the COE then needs to establish criteria for vetting and filtering them. The process of building reusable components and integrations can accelerate the adoption and deployment of automation. Reusable components also allow easier governance of the automation, thus ensuring their reliability and justifying their continued use.

Focus on the CoE process

It's tempting for executives to focus on the quick wins -- but this can come at a risk to the processes required to scale, which is the main point of an automation CoE.

Param Kahlon head shotParam Kahlon

"It's faster and easier to build and implement automation in silos, but it's difficult to scale and brings only short-term benefits," said Param Kahlon, chief product offcer at UiPath, an RPA vendor. Instead, there should be an organization-wide approach that looks at the key focus areas across the business that require automation. This can maximize the effectiveness and sustainability of the processes.

Augusto Barros head shotAugusto Barros

It's also important to think about how processes are performed. Many organizations start automation initiatives without fully understanding the task that needs to be automated, said Augusto Barros, vice president at Securonix, a security operations and analytics platform. "It is surprising to see how many different versions of a process can exist in an organization, even when there is a belief that it is a single, unified process," he said.

He believes a clear understanding of the task that will be automated should be the primary goal of an automation CoE. This requires experts with the appropriate skills to uncover, document and refine processes.

"Business process analysts with domain knowledge are hard to find, but they can dramatically improve the program's chances of success," Barros said. These experts can help developers understand the processes being automated, the expectations for the deliverable and assess the quality of the final product.

A successful CoE team can also bring an innovation mindset to the business in the hunt for new offerings that can be used across the organization. "This helps the development team to implement advanced solutions and increase automation reach," Ganu said. He worked on one project that involved reading a bar code from a very dark image on a document and was able to solve the problem with a set of Python libraries. After developing a small script to call this from an RPA tool for each image, the team was able to reuse it across the company for other bots.

Successful CoE practices

Manish Sharma head shotManish Sharma

Manish Sharma said some of the things that have helped companies succeed in their CoE efforts include:

  • Applying digital-first to the human-led operations and rigorously measuring outcomes such as productivity and utilization;
  • Employing process mining and automation diagnostics tools to create a steady pipeline of transformation opportunities;
  • Controlling automation scope with a smart 80:20 approach to help optimize return on investment by starting with an initial focus on productivity and expanding time to cover more complex scenarios; and
  • Establishing clear accountability and benefit realization ownership, targeting top quartile performance and using benchmarks to drive process maturity.

Measure the value of a CoE

A guiding principle of any large-scale automation effort is to measure what matters to prioritize future automation opportunities and prune away the seemingly good ideas that failed to deliver. This can also help identify the appropriate level of automation for a project.

Michael Sena head shotMichael Sena

"A vast majority of processes could be automated to a great degree, but very rarely, the highest automation level possible is optimal," said Michael Sena, founder and CEO at Senacea, a spreadsheet automation consultancy. The actual value lays in determining the appropriate degree of automation, beyond which additional benefits won't outweigh the cost of achieving them.

Sena recommended looking at the total value -- while labor is a good place to start, other factors can be even more significant. For example, a deeper analysis might reveal the value of reducing data discrepancies, which is directly contributed to losing many customers. "That's a game-changing factor deserving a value tag," Sena argued. Thus, keeping track of the associated financials is crucial and can allow reevaluation of the program scope in the later stages.

Jerry Ting head shotJerry Ting

Jerry Ting, CEO at Evisort, a contract management platform, has seen clients build efficiency gains in the form of time savings, better experience and increased quality of performance. It's important to be very use-case specific and to have a benchmarked approach to assess the key success criteria from a before and after state.

An example in terms of increased performance is that an employee could only track 10 data points manually but then 50 data points in the same time using an automated system. "There may not be a time saving here, but the overall quality of the performance has increased," Ting said.

Scaling a CoE for growth

A top challenge lies in making sure the automation is spread across the entire enterprise. "Scaling the programs should be done by raising awareness about the benefits of the program," said Daniel Cooper, managing director at, a digital transformation consultancy.

Daniel Cooper head shotDaniel Cooper

One good practice is to host internal DevOps days, like workshops, to showcase the results from early adopters. It's also good to create a monthly newsletter to present success stories. Cooper also found that publishing automation program code as internal open source helps build trust within the internal development community that drives adoption. He has also seen ChatOps tools help, particularly when they are integrated within the automation and are publishing the updates from automation projects at various points.

"It helps to get a community of practice started within the organization and drive adoption," Cooper said.

Dig Deeper on Digital transformation

Cloud Computing
Mobile Computing
Data Center
and ESG