- Decentralized Operating Model
In a decentralized model, data management responsibilities are distributed across different lines of business and IT. Collaboration is committee-based; there is no single owner. Many Data Management programs start as grass root efforts to unify the data management practices across an organization and therefore have a decentralized structure.
The benefits of this model include its relatively flat structure and its alignment of data management to lines of business or IT. This alignment generally means there is a clear understanding of data requirements. It is also relatively easy to implement or improve.
The drawbacks include the challenge of having many participants involved with governance bodies and in decision-making. It is generally harder to implement collaborative decisions than centralized edicts.
Decentralized models are generally less formal and because of this, they can be harder to sustain over time. To be successful, they need to have ways to enforce consistency of practices. This can be difficult to coordinate. It is also often difficult to define data ownership with a decentralized model.
- Network Operating Model
Decentralized informality can be made more formal through a documented series of connections and accountabilities via a RACI (Responsible, Accountable, Consulted, and Informed) matrix. This is called a networked model because it operates as a series of known connections between people and roles and can be diagrammed as a ‘Network’.
A network model’s benefits are similar to those of a decentralized model (flat structure, alignment, quick set up). The addition of a RACI helps create accountability without impacting the organizational charts. The additional drawback is the need to maintain and enforce expectations related to the RACI.
- Centralized Operating Model
The most formal and mature data management operating model is a centralized one. Here everything is owned by the Data Management Organization. Those involved in governing and managing data report directly to a data management leader who is responsible for Governance, Stewardship, Metadata Management, Data Quality Management, Master and Reference Data Management, Data Architecture, Business Analysis, etc.
The benefit of a centralized model is that it establishes a formal executive position for data management or data governance. There is one person at the top. Decision-making is easier because accountability is clear. Within the organization, data can be managed by type or subject area. The drawback is that implementation of a centralized model generally requires significant organizational change. There is also a risk that formal separation of the data management role moves it away for core business processes and can result in knowledge being lost over time.
A centralized model generally requires a new organization. The question arises: Where does the Data Management Organization fit within the overall enterprise? Who leads it and to whom does the leader report? It is becoming more common for a Data Management Organization not to report to the CIO because of the desire to maintain a business, rather than IT, perspective on data. These organizations are also commonly part of a shared services or operations team or part of the Chief Data Officer’s organization.
- Hybrid Operating Model
As its name implies, the hybrid operating model encompasses benefits of both the decentralized and centralized models. In a hybrid model, a centralized data management Center of Excellence works with decentralized business unit groups, usually through both an executive steering committee representing key lines of business and a set of tactical working groups addressing specific problems.
In this model, some roles remain decentralized. For example, Data Architects may stay within an Enterprise Architecture group; lines of business may have their own Data Quality teams. Which roles are centralized and which stay decentralized can vary widely, depending largely on organizational culture.
The primary benefit of a hybrid model is that it establishes appropriate direction from the top of the organization. There is an executive accountable for data management and/or governance. Business Unit teams have broad accountability and can align to business priorities to provide greater focus. They benefit from the support of a dedicated data management Center of Excellence that can help bring focus to specific challenges.
The challenges include getting the organization set up, since doing so generally requires additional headcount to staff a Center of Excellence. Business Unit teams may have different priorities, and these will need to be managed from an enterprise perspective. In addition, there are sometimes conflicts between the priorities of the central organization and those of the decentralized organizations.
- Federated Operating Model
A variation on the hybrid operating model, the federated model provides additional layers of centralization / decentralization, which are often required in large global enterprises. Imagine an enterprise Data Management Organization with multiple hybrid data management models delineated based on division or region.
A federated model provides a centralized strategy with decentralized execution. Therefore, for large enterprises it may be the only model that can work. A data management executive who is accountable across the organization runs the enterprise Center of Excellence. Of course, different lines of business are empowered to meet requirements based on their needs and priorities. Federation enables the organization to prioritize based on specific data entities, divisional challenges, or regional priorities.
The main drawback is complexity. There are a lot of layers, and there needs to be a balance between autonomy for lines of business and the needs of the enterprise. This balance can impact enterprise priorities.
For Your Further Reading:
