In this data-driven era, organizations are realizing the significance of governing their data. Poor governance can lead to low data quality and lack of visibility.
Deploying a data stewardship or distributed data stewardship model aids governance, fueling positive business outcomes.
With large amounts of data -- data storage on cloud and otherwise, complex data pipelines and ecosystems, higher scrutiny and more consumption of data -- organizations are moving toward the democratization of data. This enables everyone in an organization to work with and feel confident talking about data. They use observability tools to monitor, track and improve the quality of data in real time. Data stewardship models, armed with customer experience information, have become a hub of data-information decision-making.
Data decision-making depends largely on the use and consumption of relevant data across an enterprise. This often leads to conversations about data ownership, accountability, visibility, quality and the like. It also raises questions about what constitutes a data stewardship model and creates confusion around terminology.
The data stewardship and distributed data stewardship model lingo is frequently mixed together, with rudimentary understanding between the two. This article aims to dispel any misunderstandings, clarifying how these two terms are different and why organizations should consider using one of these models for their data operations.
Data stewardship vs. distributed data stewardship
Data stewards are a central point of contact. They enforce accountability of the data lifecycle, and oversee data governance and visibility. In many instances, data stewardship is a centralized business or IT function. These settings require enterprise data governance or expertise in data management and governance execution.
Distributed data stewardship is a model or framework that allows teams closest to the data to manage access and permissions. Data management is decentralized and resides within the business unit.
To be or not to be a data steward organization
The promise of data stewardship is usually not fulfilled because many organizations don't clearly define the role of data stewards. Formalizing the role of stewardship is essential to preempt confusion or disillusionment.
Effective data stewardship programs begin with an organization's data management and governance. Key components include the trifecta of data stewardship: people, technology and processes.
- People means the data-centric skills of staff, data analysts and modelers.
- Technology encompasses knowledge of data systems.
- Processes covers comprehension of where relevant data resides and what is available for different projects.
The core component of a distributed data stewardship program is similar to a data stewardship one. The success of such a model depends on how well a decentralized IT, governance and distributed access management model works.
Because a distributed data stewardship model delegates data management responsibilities throughout the enterprise, the fundamental difference between a data stewardship model and a distributed data stewardship model is in shifting an organization toward decentralizing data access. This requires time, effort, cadence and key stakeholders who agree and adhere to such a framework.
In either case, data-driven organizations adopt a customer-focused, measurement-driven, quality or lean six sigma data improvement initiative for their data stewardship model or program.
How does data stewardship and distributed data stewardship work?
To implement a successful data stewardship program, three core elements must be in place: organizational culture, data management and measurement.
- Organizational culture. Data stewards can only be successful if enterprise executives support and help incorporate their vision. Clarity on data ownership, breaking down internal silos, reusing data and sharing it across the enterprise gives data stewards staying power to realize successful enterprise outcomes.
- Data management. Defining, managing, tracking and improving data quality generate attention on data origin, its architecture, how it's administered, and its security and access rights.
- Measurement. Data stewards must align their work with clear success metrics. These metrics may include financial, quality and human development; labor productivity; and alignment with business objectives. Metrics to track progress in an organization include reduction in complexity, lower cost and execution time, higher performance metrics, optimized value from data assets, improved speed of execution and elimination of redundancy, increased collaboration and improved results.
Without these core elements, data stewardship models may not yield an organization's executive vision about the value of data. This includes what data is relevant, utilized, consumed, reusable and scalable across the organization.
Various real-world examples of data stewardship models indicate how organizations fulfill their data value vision and get the benefits of a data steward. In some organizations, data stewards are subject matter experts, owning and managing a discrete data subject area. In others, functional data stewards focus on a line of business and affiliate with the business objectives of the department. They may also oversee discrete business processes and oversee multiple data domains or application/systems.
More often, data stewards are IT centric, assigned to systems that generate the data they manage. Project-driven data stewards end their role when the project ends. In all these cases, ownership boundaries should be clear with defined business rules and usage environments.
In some scenarios, a project may devolve into poor data decisions, even if a data steward was involved initially. If a data steward goes off a project before the project is complete and without established data quality boundaries, it disrupts continuity. For example, project team members could unintentionally manipulate data while generating a report to support business decisions, and in doing so create inaccuracies.
Data is often not reusable or scalable enterprise-wide, which does not help achieve broader business goals such as customer retention. With a solid foundation of the three components, you can reach the desired outcome under any data stewardship model.
Distributed data stewardship model implementation
Similar to data stewardship models, distributed data stewardship models also warrant the use of the three core components. Key centralized IT functions such as data infrastructure, management and governance of data; appropriate data access for users; and regulatory compliance are potentially rearranged over time.
When implementing a distributed data stewardship model, these phases may be one of the following:
- Centralized IT and governance team with distributed data access management to a line of business for everyday user access.
- Centralized IT and decentralized governance team with distributed data access management, enabling domain-based access across the organization.
- Decentralized IT and governance team with distributed data access management.
Development of a data stewardship framework starts by selecting a model framework that best aligns with your organization. Then, define the core components of an operating model, and develop a transition plan to incrementally migrate to the target operating model. This will assist data governance, data sharing, and regular use of relevant data, while improving an organization's data ecosystems.