If you decide to build an analytics team, there are three primary approaches to locating the data analytics professionals within your organization. Making this choice is another important part of how organizations can get started with advanced analytics.
Decentralized, i.e., located within specific functions, such as Operations or Sales
In this model the analytics professionals are integrated within various functions and work on the projects specific to that function. The main benefit of this structure is that the analytics professionals have a deep understanding of the business processes and analytical problems the department is facing. They are ideally positioned to identify problems and projects that could benefit from an advanced analytics solution and they have a close relationship with the business owners, which helps build trust and acceptance of the analytical solutions. This structure is typically present in companies where only some of the functions are mature enough to recognize the need for advanced analytics. This structure is also beneficial when a function has relatively complex processes and detailed modeling requires deep understanding of the operations processes. A decentralized approach is also useful if there is a long-term assignment to a large project in a specific area, or many projects in one specific area. Finally, in a decentralized approach, data scientists may engage with a self-service approach to analytics, where they follow best practices, but take the initiative to acquire and leverage analytics tools appropriate for the initiatives within the functional unit they are supporting.
Centralized
In this model, analytics and data science professionals are located in a standalone business unit and work on projects across multiple business functions or government components. A group could exist within Information Technology, or within an Office of a Chief Data Officer, Chief Analytics Officer, or other similar analytics leader. It is common for the centralized team to be located within one particular department but operate globally. This typically happens when a local functional team expands its scope to another component. There are two main benefits of the centralized model. The analytics professionals have visibility of data, processes, and requests across multiple functions and are in position to propose and create integrated solutions that are more strategic than specific dedicated requests. In addition, the centralized teams can establish company-wide processes and standards for analytics project inception, development, and integration. This structure is typically present in more mature organizations where executives recognize the need and have a vision for enterprise-wide analytically-driven decision making.
Center of excellence
This model combines decentralized and centralized approaches. There is a centralized analytics team (center of excellence) that establishes company-wide analytics strategy and processes and has governance over all analytics capabilities and projects across multiple business functions or government components. There are also analytics professionals integrated within individual functions who are dedicated to projects within those functions. This approach combines benefits of the centralized and decentralized models and is typically present in very mature organizations with critical mass of analytics professionals and enterprise-wide strategic analytically-driven decision making. This level of maturity is often found in big data organizations, such as technology companies, which are used to the idea of integrating data collection and analysis into processes at every level. Reaching this level of maturity makes the development of business cases much simpler, because this level of integration promotes close collaboration, increasing the chance that an employee is able to, for example, identify a business opportunity and quickly bring the business case to an analytics/business intelligence professional on his or her team to brainstorm possible technical approaches.