Evaluate alternative operating models.
The degree of centralization (or decentralization) of the overall analytics function is a critical factor in the feedback loop. A centralized organization provides greater control and economies of scale while a decentralized model provides greater insight into business unit and operational matters. Both are viable structures if there is standardization and an environment of control and accountability, but this is much more difficult to achieve in a decentralized model.
Align your structure to meet specific stakeholder needs.
Successful organizations structure themselves and their services to meet specific stakeholder needs. For the analytics department, this may be achieved through a three-tier structure consisting of front, middle and back office teams. With a focus on creating actionable insights, the front team provides decision support and analysis, such as management reporting, resource allocation and forecasting, and long-term strategic planning. The middle office provides specialty services, such as modeling, scenario simulation, and sensitive data masking & regulatory compliance in an effort to mine the data. The back office delivers raw data processing services, such as crawling & scraping, ETL, consolidation, and semantic model development in an effort to abstract the behavior of the business.