Contingency-based data governance model – Designing a framework that fits the operating environment
Updated: Jan 24
Practical and research-oriented data governance models in the past provided organizational structures that were assumed to be suitable for all companies. It has thereby been ignored as a fact that each company requires a specific data governance configuration that fits a set of contingencies. Contingencies are similar to context factors that impact the contribution of data governance to enhancing an organization's performance.
For example, IT governance - which is considered a subset of corporate governance - assists in the achievement of organizational success by efficiently deploying reliable and scalable technologies. As stated above, even for IT Governance the underlying assumption is that there is no universal IT governance design for all enterprises.
How does contingency theory apply to data governance in influencing company outcomes?
In organizational design, contingency theory describes similar dependencies. Some characteristics of an organization are influenced by contingencies that determine its overall effectiveness. Contingency theory traditionally addresses the fit between organization structure and the environmental conditions in which it operates. Later scholars expanded contingency theory by taking internal conditions into account, such as structural formalization and specialization in decision structures.
Companies are forced to continuously adapt to new environmental conditions through changing business models. Quality data is a prerequisite for meeting these changing business requirements and achieving enterprise goals. In addition to such strategic factors, some operational domains directly rely on high-quality corporate data, such as customer service management, strategic decision-making, business intelligence, marketing, regulatory compliance, etc.
Despite the business relevance of data quality management, it is surprising that responsibility for improving data quality and managing corporate data is often assigned to IT departments. In addition, many companies try to deal with data quality issues by simply implementing controls in data lake or data warehouse systems. Surveys on data governance reveal that organizational issues in managing data rather than technical issues are more critical to their success.
How different is a contingency-based data governance framework?
This framework suggests that contingencies affect data governance and that a data governance configuration is unique to an organization. Essentially, it proposes a flexible data governance model composed of roles, decision areas, and responsibilities. It also outlines which contingencies influence the company-specific configuration of the model. More specifically, the contingencies impact two main parameters of the model -
organizational structure of data management activities
coordination mechanisms of decision-making
A data governance model helps companies structure and document their data management contingencies and their impact on the model to show which configuration is best suited to their company. Neglecting it means that data management runs the risk of becoming an end goal in itself rather than data management impacting the effectiveness of organizational objectives.
An equally crucial aspect of a data function is managing capabilities outside the scope of data management, such as sponsoring, managing expectations, avoiding scope creep, and handling political issues. These activities will have to be factored into the accountability of data governance through formalization. In contrast to IT systems, which are mostly owned by IT functions, data is also owned by business units, shareholders, or customers of the enterprise.
In response to this contingency approach, data governance frameworks have evolved with patterns or models that reflect different contingencies. There have to be extensive studies to examine the impact of organizational context variables taken from classical contingency theory (such as organization size, competitive strategy, and locus of decision-making authority/control) on the functioning of data governance.
In my book Data management and governance services - simple and effective approaches, I have suggested three models: Centralized, De-centralized, and Hybrid models. The hybrid pattern distinguishes between two types of data functions:
Management of the use of data is decentralized to business units or assisted by a central function only to an activity or two
Governance of data is centralized
This is called assisted service, where the skillsets within a business division determine the nimbleness of a central function engaging in minimal activities like data profiling.
Figure-1: Contingency-based data governance framework
There needs to be a clear distinction between
data infrastructure and platform management
data usage management, and
data project management.
To understand this further, organizational contingencies determine the data governance configuration of a company. In order to understand how contingencies, affect the individual design of a company's data governance model, two design parameters should be considered:
organizational placement and
coordination of decision-making authority.
The value pairs range from two choices based on design parameters
centralized to decentralized, and
hierarchical to cooperative models.
A design parameter affects the configuration of a data governance model, that is, it influences how responsibilities are assigned in a RACI matrix.
Organizational placement of decision-making authority
For the data governance model, the first design parameter is the organizational structure of the Data Management activities. Centralized data governance leads to greater control over data standards and enables better monetization of information at scale. Decentralized IT governance allows greater responsiveness and flexibility with regard to business needs and customized solutions for each business unit.
However, the right model of data governance needs to balance trade-offs between
standardization on the one hand and
responsiveness to events on the other hand.
A digital function in a large financial services enterprise might be more offensive in its strategic direction and will have to be nimble to react to market conditions while other divisions may still be defensive in their strategy owing to higher regulatory environment conditions. e
Centralized data governance design
Decentralized data governance design
"A” in some decisions of major relevance
“C” (recommending, not commanding)
Many “C”, “I”, no “A” alone
Business & technical data steward
Many “R” and some “A”
A decentralized data model involves all decision-making authority being allocated to individual units, divisions, or lines of business. The centralized form is associated with smaller firms, defensive and conservative strategies, centralized control, and mechanistic decision-making. The decentralized form is associated with large firms, offensive or aggressive strategies, decentralized control, and organic decision-making.
Coordination of decision-making authority
The coordination of decision-making authority and influence relies on either hierarchical and vertical lines or collaborative and horizontal capabilities. Firms with distributed data management used these mechanisms to coordinate data sharing and decisions. Data governance design must reflect and support the decision-making style and the prevailing culture of the company to maximize its effectiveness.
Hierarchical Data Governance Design
Cooperative Data Governance Design
“A” in most decisions
“C”, “I”, few “A”
Business & Technical Steward
“R”, “I”, few “C”
Many “A” (conjointly) and “C”
Many “C” & “A” (Conjointly)
The hierarchical data governance model is characterized by a top-down decision-making approach. Either the data officer or the data council has decision-making authority for a single data management activity. Tasks are delegated to business and technical data stewards. However, they will not be directly involved in decision-making.
In a cooperative data governance model, formal and informal coordination mechanisms are used to reach decisions. A data council or data officer is complemented by working groups, task forces, and committees with members from multiple disciplines. No single role can make a decision on its own. New integrator roles, such as process owners or data architects that report to business units, establish a high degree of cross-unit collaboration.