The Automation Center of Excellence

20 min readNov 1, 2021

Executive Summary

Automation has seen a rapid evolution in the last decade. If you are a business leader worth your salt trying to make it big in the run-up to 2025 (by the time, World Economic Forum reports, more than half of all workplace tasks will be completed by machines), you would know automation is not just a ‘good-to-have’ today. It is the indispensable driving force behind key business imperatives — cost savings, operational efficiency, optimum resource allocation, and so on.

A decade back, it all started with task-level automation where some of the tasks in a process were automated. The tasks that were usually not automated were the ones that involved data capture. Gradually, we learn from our experience that even the data capture tasks were automated with the advent of hyper automation and with the help of machine learning. Once data could be dealt with seamlessly, the problem of data size came in. Automating segregation and storage of large volumes of data became the next big problem. This problem was solved using a lot of data engineering techniques. With cloud computing, this became an easier problem to solve. With the availability of a large amount of data, analyzing it (data science) and deriving meaning from the data (business intelligence) became the problems to solve. As time and technology evolved, the complexity of problems that automation has to solve kept evolving. Hence, automation also had to transform itself.

Today, automation initiatives in enterprises are mostly driven by their business units (BUs). BUs are able to customize the automation solutions to meet their specific business needs, and they work well. However, from an organizational point of view, it is probably not the best approach. Think about an enterprise where different automation solutions are being used to solve similar problems across BUs. This results in duplication of efforts, cross-functional integration siloes, inefficiency in resource usage and heavy maintenance cost involved in running these solutions.

CTOs and technology leaders are aware of these problems. And automation being a priority technology initiative at most organizations, they are looking to solve these challenges through logical integrations, governance and standardization across organizations. Many companies are creating the role of Chief Automation Officer to drive these overarching and strategic automation goals.

I have been working with insurance companies to solve these automation problems for over a decade. Based on my experience, I propose that every large organization set up an automation center of excellence (CoE) to standardize/ aggregate automation solutions across the organization, maximize the ROI of the automation initiatives and build automation benchmarks to ensure access to latest technological advancements across BUs — especially in the face of unavoidable change like the present pandemic situation.

In this whitepaper, I have tried to explain the need and mechanics of Automation CoEs, the various pillars you should consider and the approach to setting up your own Automation CoE.

While proposing this approach, I also understand that each organization is unique, and a one-size solution doesn’t always fit. My recommendations in this article are broad guidelines to help chalk your automation journey. Based on your organization’s culture and stage of technology evolution, you can customize these principles to fit your automation goals.

The key challenges of present automation initiatives

What is automation? The concept of automation is, of course, not new. It is as old as 762 BC; In Homer’s epic poem The Iliad, the God of blacksmiths and fire, Hephaestus, was depicted as inventing Automatons, self-directed metallic machines, to speed up the manufacture of weapons. These machines were perhaps the first instance of replacing human labor in the history of mankind.

Cut to the 21st century, and the world is seeing a Homeric reboot of automation, or better still the recreation of the widespread success of the industrial revolution. Not only it brings more efficiency to business operations but also leads to huge cost-saving opportunities. Due to the ongoing COVID-19 pandemic and its impact on businesses, it is expected that more and more organizations will shed off their overdependence on the human workforce for routine, human processes and adopt automation. This will not only build satisfaction and trust for the end-user but also free up employees to take up more crucial responsibilities, reduce costs and insulate businesses from such pandemic situations in the future.

Automation techniques like RPA and intelligent automation help bring accuracy, speed, service continuity, process efficiency, ease of use, workforce agility, scalable infrastructure, and strategic focus to an enterprise.

The International Society of Automation defines automation as “the creation and application of technology to monitor and control the production and delivery of products and services”

The efficiency factor

In a scramble to power their businesses with automation, most organizations fail to get the best out of their investment. In an enterprise setup, every business unit has large-scale requirements for automation. Think about it, there could be a host of automation solutions, in-house and external, used by each of your business divisions, without a single overarching solution that can interface with all these applications. When there are multiple automation initiatives in a single organization, they are bound to produce issues of efficiency.

Scaling issues Businesses may encounter tech resources overwhelmed with the scale of operations.
Multiple vendor management They have to sift through multiple obligations that come with multiple vendors/ partners. Implementing them and managing them becomes a task.
Weak visibility Too many internal automation initiatives mean a lack of proper visibility into operations.
Lack of standardization A lack of proper central leadership, governance, and problem-solving strategies means a lack of standardized and common best practices or training and skilling methods.
Escalating costs Duplication of solutions to solve same problems across the organization rises licensing, maintenance and resource costs on an ongoing basis

This creates a roadblock to attaining faster time-to-market and also snowballs into a major issue in managing multiple automation initiatives properly. The outcome of efficiency issues? Low ROI, spiraling costs, and an aversion to adopting newer automation initiatives in the future. In order to meet this challenge, some companies set up automation teams that are most of the time without any binding principles and depend heavily on a single team, like IT. This traditional way of ‘fixing’ the problem is going only half the way. Not only does it not solve inefficiency issues but in a way creates more mess. This also impedes the larger goal of deriving greater ROI. Leaders also acknowledge that they have to reroute away from the model of building one or two bots to tackle an issue, to a more wholesome automation “factory” setup that can produce and implement scores and hundreds of bots when required. This creates an end-to-end automation factory that adds to efficiency and impacts the entire value chain. Any business knows that to meet the end goal of successful digital transformation, the movement needs to be from tactical to strategic.

The expectations factor

About a decade ago, automation would just mean creating an application that replaced a manual process. As it stands now, automation has evolved into a multi-disciplinary and multi-objective culture with multiple and new-age expectations.

Use case: Insurance domain

Let’s take a ‘Basic Policy Checking’ process in insurance. It involves checking policy and coverage details between the previous year and the current year’s policies. Consider a huge insurance brokerage firm that processes more than 100,000 policies a month. Expectations from the automation of the process a decade ago would be as follows:

Streamlining all the incoming policy checking tasks from various sources (FTP, Emails, etc.) into an organized application so that none of the incoming requests are missed out
Creating tasks for every person involved in the process and streamlining the flow of the process by taking the request through various stages to completion
Making sure that the process is digitized and people work on a unified application, completing the tasks assigned to them
Reporting to top management on the number of tasks completed, average turnaround time, etc.
Maintaining security and data privacy
Ensuring the ability to scale the application for 10% year-on-year growth for the next five years
If the above objectives are met, the process would be considered fully automated, and everyone would be happy.

To achieve this automation a decade back, one would have needed the following resources:

Business Analyst to document processes for automation
Developers to build web apps and persistence to manage transactions and UX engineering
Testers to test the application
Project Manager to manage the entire project from either side
Subject Matter Expert (SME) to be the POC from the process perspective and facilitate UAT

Fast forward to today, and consider the same process. What is the new age ask?

Basic requirements: All of the asks from a decade ago (obviously)
Enhanced UX and security: Enabling multi-channel/ applications for an enriched user experience; Enhancing security and data privacy due to the modern environment, mobile workforce, and multiple points of access to data for providing more control to users over their data
Extraction and analytics: Digitizing the incoming policy documents via Optical Character Recognition (OCR) technologies and extracting the fields required for further processing from the documents (converting semi-structured data into structured data), thereby reducing Dark Data; Storing all the incoming documents and their extracted metadata; Extracting intelligence out of the extracted structured data, which could give insights and open new business avenues
Integration and adoption: Integrating with multiple upstream and downstream systems for seamless data flow; Ensuring organization-wide adoption, which spells out a need for platforms rather than individual applications

For these new objectives to be met, a business needs expertise in a lot more areas apart from the ones stated in the earlier scenario

The new additional resources that can be added as a pod team include:

Solution Architects to architect the end-to-end enterprise system
Data Engineering team to manage the huge amounts of incoming data
Data Science team to analyze and derive patterns from the accumulated data
Business Intelligence team to slice-and-dice the data for actionable reports 8

And the list goes on…

The Automation COE Approach

So what’s the solution? What businesses need is a more strategic, top-down approach to address the complexities and rising expectations from automation. At the same time, they need to manage such multiple automation initiatives across the organization and reach each and every team.

Automation centers of excellence (CoE) are built on the premise of having a multidisciplinary team (the lines between IT, DevOps, and business teams ideally blur here) lead, govern, manage, control, secure, and support various automation initiatives or create a set of universal best practices to be followed by every business unit within an organization that has embarked on automation projects. Automation CoEs not only bring discipline to automation initiatives in an organization but also paves the path towards scaling effortlessly when required.

Automation CoEs take a stock of the business and ascertain its organizational, governance, and operational readiness before developing enterprise automation strategies aiming to a more evolved automation journey in the organization.

Automation CoE takeaways:

Hit markets faster
Maximize ROI
Prepare for a more evolved automation future

The key responsibilities of an automation CoE are as follows:

Assessment Auditing current automation initiatives, assessing risks
Planning Planning strategies, standards, guidelines, and best practices; Creating plans for issue identification and fixes
Management Managing end-to-end processes and automation CoE functions along with ensuring the security practices; Enabling testing, training, skilling, and change management; Aligning multiple partner/ vendor contracts; Implementing business continuity plans that factor in IT upheavals
Analysis and reporting Ensuring proper visibility through unified tools and dashboards; Establishing KPIs to measure the success of automation CoE; Monitoring, analyzing, and deriving insights from the performance of operation

The different models of an automation COE

An automation CoE gets modeled in one of two different ways depending on where it is started. It can either get initiated in a single division in an organization, get incubated, and then get adopted by other divisions if successful. Or it can be initiated from the top as an automation mandate and trickle down to the divisions of an organization. Let’s now see in detail how these two models work.

Decentralized or Siloed model

In this model, the automation CoE is formed in one of the business units (BU), which realizes that automation is necessary to take the next evolutionary step. This is usually a good way to start a CoE since:

The work scope is usually within a particular process and is more of a proof of concept (POC) than an actual implementation.
This also gives the automation CoE teams a lot of freedom to work on automation with the tools and skill sets available within the team

However, even though the POCs can be done faster, the siloed mode’s scalability is difficult since the best practices that are formed are usually BU-focused than organization-wide. For e.g., if there is a need for data extraction, it will only be designed and developed to work for a single process or processes within the BU instead of a centralized data extraction pipeline with various source However, even though the POCs can be done faster, the siloed mode’s scalability is difficult since the best practices that are formed are usually BU-focused than organization-wide. For e.g., if there is a need for data extraction, it will only be designed and developed to work for a single process or processes within the BU instead of a centralized data extraction pipeline with various source options and centralized licensing that can be used by many similar data extraction projects in the organization. In short we end up with the following:

Multiple teams work and create the same thing.
In BUs where the scope for automation of an identified process is less, the team is underutilized.
In the course of time, every BU might end up having its own Business Process Management tools, its own RPA tools, its own toolsets, processes, and best practices.
The overhead of managing it all efficiently and struggling with the problem of scaling turns huge.

All of this makes it really difficult for the siloed business units to see an ROI for the investment of efforts.

Centralized Model

In this model, the automation CoE is formed at an enterprise level with a directive from the top management. This is more of a top-down approach. The overall objectives are:

Collective intelligence
Cost reduction
Effective management of assets and resources

A CoE vision is first drafted that explains how the CoE will add value to the automation initiatives at an organization level, considering various business units that will benefit from it. The vision is submitted for funding so that a core team can be formed. This core team forms the basis of the governance team that will hold the following responsibilities:

Best practices — This refers to the formation of best practices that ought to be followed by all the implementation teams. This includes the process assessment templates, selection criteria, and prioritization techniques, ROI calculation templates, cost-benefit analysis techniques, and so on. These practices can either directly be adopted by the implementation teams or can be tweaked based on the needs of the process getting automated.

Tools and techniques — Various tools like RPA, BPM, data extraction tools, and frameworks that align with the organizational technology stack are to be standardized for adoption across the board. The maintenance of these will be done by a central operations team.

Training and reskilling — Training of resources required for implementing and supporting the automation initiatives, and reskilling the existing resources for supporting a wide variety of tools and frameworks, are to be facilitated and funded by the central team.

Operations — The management of all the tools, their upgrades, maintenance, licensing, etc., are also to be managed by a shared services team, and the hours spent are to be billed to the appropriate cost centers

In the centralized model, there is very little flexibility for individual business units to make decisions at a framework level. The overall framework and methodology need to be adopted and tweaked to the needs of the process. Even the modifications have to be presented to the central team to get their blessings before implementation.

Hybrid model

Learning from the pros and cons of both the siloed and centralized models, many organizations are evolving to a hybrid or federated model. The hybrid model takes the best of both models and combines them into one. Here, the automation CoE functions are centralized within the core group but to enable scalability, their capabilities are federated out among BUs that need them. Mostly, such a model involves an execution engine that takes a stock of demand and delivers the capabilities to BUs that cannot manage with their own automation engines.

For e.g., the business team can readily adopt the best practices put together by the central team, thereby reducing the time for the initial POC and, on a successful implementation, can scale as easily by utilizing the knowledge base, the reusable components, and the framework set by the central team. The learnings of the individual teams are collated by the central team by a push mechanism so that there is an effective feedback loop that helps the central team improve on the best practices and expand the body of knowledge.

The hybrid model works best in the case of a globally distributed and diverse organization involved in a large number of automation projects and having decentralized decision-making. Some of its most pronounced wins are:

Ease of scalability, process, guidelines, and standards
Faster time to market
Higher independence of BUs in managing automation projects
An efficient way of reusing knowledge and learning from feedback

Here’s a snapshot of top automation CoE activities when handled by the three types of models

Composition of an Automation COE

The core strength of an automation CoE lies in its multidisciplinary nature. It’s a coming together of business analysts, developers, testers, solution architects, SMEs, and business intelligence, data science, and data engineering folks. In a CoE setup, these disparate teams contribute to fulfilling the ultimate objective of the organization in a structured and strategic manner. The CoE comprises multiple sub-teams, which work together with a unified goal of enterprise automation.

Governance team

The governance team is usually formed of SMEs from various business units, members from enterprise architecture, security, compliance and IT teams. The governance team plays a pivotal role and drives every other team of the CoE. This team usually reports directly to the CIO in most organizations.

Here are some of the key responsibilities of the governance team:
Does a capability assessment of the organization and recommends areas of improvement based on the organization’s maturity
Sets up the basic framework for automation that aligns with the enterprise vision
Sets up governance blueprint for process assessment templates, selection criteria, prioritization, ROI calculation parameters and cost-benefit analysis methods
Maintains the automation knowledge, in terms of case studies, from various automation factories, maintains a record of the lessons learned from automation initiatives in the organization, and creates an effective feedback loop for evolving best practices
Acts as a bridge connecting IT and business so that the automation initiatives are driven on IT principles for achieving business goals
Manages common infrastructure and central operations to effectively monitor all automation in production and to make sure that the ROI is met and business continuity plan is in place in case of a failed initiative
Audits for security and compliance
Evaluates vendors and their onboarding
Sets up SLAs for operation teams, maintaining and supporting the automation and the OLAs between various teams.

Technology team

The technology team usually comprises multiple practices such as the core engineering team, solution architects, data engineers, data scientists, cloud engineers, RPA engineers, etc. — alongside IT operations. The core team is formed of engineers from various automation factories as well. This ensures that the learnings from one automation initiative are consolidated and propagated organization-wide. This team usually reports to the enterprise architecture team, which, in turn, reports to the CIO (the reporting structure can differ from organization to organization). Here are some of the key responsibilities of this team:

Sets up core infrastructure for common systems
Incubates automation factories and reskills resources in technology areas
Reviews architecture and design for all automation initiatives to make sure it aligns with the enterprise architecture
Upgrades and does patching of tools and systems
Keeps abreast of the new features of the tools and frameworks used in automation initiatives, and continuously improves
Plans high availability and disaster recovery scenarios
Works with automation factories to extract reusable components and maintains an organizational marketplace of reusable processes, components and modules
Enables license management and support contract management
Carries out tool and framework evaluations
Runs innovation hubs and enables continuous improvement
Implements enterprise-wide workshops disseminating information on the pinnacle of what’s possible to increase awareness

Some of the sub-practices under technology will have their own charter by which they operate. In fact, in today’s agile world, a single small team is expected to be well-versed with a lot of these areas and, more so, to carry the learnings from one project to another. This is where a ‘pod’ comes in. Instead of assigning people to projects, we can assign projects to teams. This gives a great advantage where the time taken for a team to come up to speed on the protocols of work is completely eliminated. Running multiple such teams in parallel automation initiatives like a factory makes it a really effective way to run an automation factory.

Generic tech team charter followed by most practices

Improve the overall process to increase efficiency in current automation projects.
Keep abreast of current industry best practices and cutting-edge technologies.
Work towards garnering certifications, whitepapers, etc., to divulge information across the enterprise as a source of knowledge.
Participate in community events and host technology events inviting technologists from different business units and even other companies for networking and information sharing.
Support projects during initiation and during a crisis.
Display technical expertise via hackathons or roadshows.
Pioneer best practices in various areas and document the story.
Conduct frequent sessions and hands-on labs sessions to disseminate information.
Generate articles, whitepapers on technology, and abstract case studies from projects — to be used by the central team.
Build and maintain a reusable components repository based on project experience.
Work on getting strategic technical partnerships

Process team

The process team has SMEs as part of the team. Their main focus is on how to run the business, grow the business and transform the business. This team works towards achieving end-to-end process automation by seeing a process as a whole instead of just task-level automation. This team comprises people from multiple business functions and contributes to collecting information regarding processes and identifying projects that transform processes. These are usually teams that are a part of the actual project implementation as well. Here are some key responsibilities of the process team:

Evaluates the processes and prioritizes them based on their business value utilizing the process assessment and prioritization techniques
Pilots automation projects that have integrations with multiple enterprise core systems and thereby sets up the precedence and protocols for any automation that will use these core systems
Evaluates the cost-benefit analysis and arrives at when an ROI can be achieved for the given process and prioritizes it accordingly
Engages people from different business units and derives case studies from all the automation done by the BUs by measuring the results of the automation (either positive or negative)
Makes sure that the different automation projects are aligning with the enterprise strategic objectives
Sets up a repository of processes automated and derives reusable flows, which will help in reducing future automation

Operations team

Perhaps the most overlooked team, but certainly the most important one, is the operations team. In the case of RPA projects, the Robotic Operations Center (ROC) will also be part of the ops team. The operations team is generally responsible for making sure that the automation is running as expected in production and planning for contingencies when it is not. This team’s role in every automation initiative usually starts from the UAT Phase, where a handover is done from the automation factory to the operations team.

The operations team maintains a checklist that it uses to validate and acquire data about the automation. It also has the power to revert to the automation team for any changes that need to be done to the automation before a successful handover. This team coordinates with all other teams to make sure the automation is a success. This team can also be subdivided into L1, L2, and L3 support, where the L2 and L3 can be centrally managed as well, while the individual BU handles the L1. Here are some of the key responsibilities of the operations team:

Facilitates handover meetings and coordinates with teams for the handover of the automation from UAT to go live and post-that for the lifetime of the automation
Maintains a checklist of best practices that will help in supporting automation like scalability, logging, audit trails, documentation, SLAs, and so on for the handover and makes sure that the various automation teams adhere to the standards
Sets up protocols that define the SLAs and the OLAs for various teams and stakeholders
Enables capacity planning to make sure that the team is adequately staffed to support and maintain the incoming automation
Tracks continuous progress of the automation and reports on how and when an ROI is attained and makes course corrections accordingly
Plans upgrades, disaster recovery tests, work with the automation team for enhancements and maintenance of licenses and coordinates with multiple integrating applications

Steps for creating an Automation COE

Once your teams are in place, what do you do next? The following approach to creating an automation CoE takes into account the vision behind it, its incubation, execution of pilot, introduction in the company automation initiatives, and its subsequent democratization.

The automation CoE team helps identify, assess and prioritize processes for automation. The first job is to form a vision for automation that will drive the entire automation CoE, and which must tie-up with the company goal. The vision is also dependent on the existing capabilities of the business. It is detrimental to go ahead without an audit of the existing capabilities. The vision is submitted to the executing committee with an aim to get funding for the automation CoE.

The next step involves forming an extended team of automation champions who evangelize the CoE across the company, executive sponsors who are tasked with maximizing ROI and monitoring savings, and a steering committee that can arrange necessary resources, review automation projects and plan for future ones. Incubation of an automation CoE also involves planning for design best practices and creating governance and processes.

To start off by piloting a CoE, it is imperative to assess it thoroughly before executing a proof of concept. Setting up the infrastructure for core systems, documenting case studies, and forming a feedback loop are also key parts of the process. During the pilot phase, the COE sets up one or more automation factories in the business unit so that the automation initiatives can be handled by the units using their own preferred vendors. The automation CoE-ratified tools, techniques, and practices need to be adhered to by the individual automation factories. The operations can either be handled by the BU or the BU can utilize a central shared services operations team with chargeback to the BU.

Once the automation CoE is in place, the team needs to introduce it across the company BUs and also look for internal projects that need automation solutions from the CoE. The next steps involve zeroing in on a particular model to start off with, kickstarting the automation initiative, documenting learnings, and scaling up easily when required.

With a target to democratize the CoE, governance and chargeback models need to be set up. The automation CoE is also responsible to support large-scale deployments, and standardize operations with an eye on continuous improvement.

Steps involved in creating an automation COE

Automation COE Overview

Automation must take into account the end-to-end view of the overall process or the portfolio of sub-processes that together create business value.

Conclusion

Centers of excellence are the new temples of management in organizations. An automation CoE tackles one of the ubiquitous techniques in modern organizations — automation — in the most organized and value-driven way possible. Apart from maintaining a body of policies, guidelines, governance rules, and best practices, it ensures consistency on automation initiatives across an organization, helps reduce risks, and removes any redundancies that affect efficiencies and create dents in the company wallet. An automation CoE impacts the bottom line by increasing speed to market, savings costs and enhancing the ROI.

Automation CoE models are not only heavily customized but also proven to deliver tangible and sustainable results for multiple clients looking to transform their automation experience.

Every day, every hour, every minute you run multiple automation initiatives in your organization without an overarching and governing CoE, it impacts the business bottom line — especially in this high-impact pandemic situation. Why run such huge costs and deal with inefficiencies at a large scale? Partner with the right vendor at any stage of your automation journey to brainstorm and create the perfect automation CoE model based on your requirements and provide the additional support of maintenance, scaling, analysis and so on.

The approach is never formulaic. It’s not a one-size-fits-all approach. Your company must have its own targets, dependencies and priorities. The proposition outlined in this article/whitepaper serves as a framework for you to understand the mechanics of an automation CoE, its optimum usage and its proven benefits. You must tweak this approach, the models and the compositions to create a best-fit automation CoE for your organization. I hope that this article helps you weigh the market realities, consider our approach and brainstorm with us to create the perfect automation CoE for your business.