Data stewardship: Essential to data governance strategies

Data stewardship vs. data governance

Every company has its own distinctions between data stewardship and data governance. Most experts agree data governance is a broader concept and data stewardship is a specific role to put it into practice. There are different ways of framing the distinctions between them.

Broader versus supporting controls. Data governance is a comprehensive set of controls -- strategies supported by policies, procedures and technologies -- to help control data, McGivern said. Data stewardship, by design, is a specific supporting set of controls to help data governance act.

"Data stewards are often where 'the rubber meets the road' for data governance controls," McGivern said. "[They are] helping to provide the necessary context to the data and other knowledge about required controls, proper usage and the current state of quality."

Framework versus role. Data governance establishes a framework for how an enterprise provisions and stores data, while data stewardship is a role within the organization that advocates for effective uses of that data to create value, said Ed Murphy, senior vice president of data science at 1010data.

In this regard, governance focuses on the essential infrastructure for data-driven enterprises:

• Roles.
• Access controls.
• Storage.
• Availability.
• Metadata integration.
• Disaster recovery.
• Legal and regulatory compliance.

In contrast, stewardship focuses on wringing value from data:

• Data set discoverability.
• Documentation and metadata.
• Change management.
• Quality advocacy.

Policies versus tasks. Data governance is the ownership and management of the policies, processes and procedures for data, said Claire Thompson, chief data and analytics officer at Legal & General Group, a financial services company. In contrast, data stewardship is undertaking tasks needed to ensure adherence to and compliance with the data policies and frameworks. The data steward interprets and implements required processes on a day-to-day basis.

Outcomes versus inputs. Data governance is the outcome while data stewardship is the input required to achieve it, said Graeme Thompson, CIO at Informatica. Data governance includes what someone is trying to achieve and the mindset necessary to achieve it. Data stewardship involves defining ownership and responsibility for different pieces of data and for the processes that generate the data.

While they're different, they are not separate.

"Most organizations that are committed to data governance invest in a data stewardship practice," Graeme Thompson said.

This is often handled by a dedicated team, or at least a team that has the responsibility, established processes and supporting technology to manage the data.

Procedures versus standards. Data governance establishes procedures while data stewardship ensures reliable standards, Graeme Thompson said. Procedures include classifying data, tracking its lineage for transparency, and enforcing rules for appropriate data access, usage and retention. The standards protect critical data and ensure compliance with appropriate conditions and objectives.

Why is a data steward needed?

In theory, an organization can get by with a data governance process spelled out by the CIO or chief data officer. However, experts see several benefits of a data steward over and above a more general data governance practice.

Apply subject matter expertise. Data governance applies to the broader field, while data stewards can use it in a specific business domain.

"Data stewardship roles are usually undertaken by people within a specific business area so they would have the SME [subject matter expert] knowledge of what that area does and what systems and data it uses to undertake the key business processes," Claire Thompson said.

Data stewards work with business leads and data owners to ensure regulatory, risk and policy compliance. They must identify, define and document the key data points, including the flow and usage of data. Data stewards can also monitor and analyze data to check quality, identify problems and escalate potential data risks.

All analytics-driven organizations will be slower without a steward as they try to reverse-engineer a data set for lack of documentation or quality.

Manage assets that manage the data. Data stewards protect and secure data because they manage the assets that manage the data, said Naveen Kamat, vice president and CTO of data and AI services at Kyndryl, an IT infrastructure services company. They can also provide insight into which assets are working and which are not. Data stewards can make recommendations on assets to remove or utilize more within the organization.

Facilitate between IT and business users. Data stewards are sometimes referred to as the data ambassadors within an enterprise because they are the lynchpin between the technical IT operations, data management teams and line of business users who consume business-critical data, Graeme Thompson said. When data is safe, secure and appropriately managed throughout its lifecycle, workers can trust it to make strategic, data-driven business decisions.

Create value with data. Data stewards make it easier for companies to treat data as a product, said Bret Greenstein, data and analytics partner at PwC. Data product ownership is all about creating value with data. The data product owner can be the data steward, adding to their responsibilities. The data product owner focuses on the use and value created from data.

Product success is typically measured by the value of the product, its usage and user satisfaction. Data product owners focus on driving data requirements that make a difference for the business. This helps companies achieve more value from their data.

Pains of no oversight

Not having a data steward can hurt the organization in the following ways:

Inconsistent naming. Data stewards can ensure consistency. Greenstein recently encountered a company struggling with customer data. They had it stored in multiple places that did not follow consistent naming. The same customer showed up under different customer IDs for various functions in the business. Customers got different pricing and discounts depending on the channel. The business leaders did not have a holistic view of a customer's buying history, preferences and needs.

Reduced speed. Data teams lose speed without stewardship because low-quality data requires more cleaning, more dev time, more code and more testing than high-quality data. Data issues slow the entire development cycle.

"All analytics-driven organizations will be slower without a steward as they try to reverse-engineer a data set for lack of documentation or quality," Murphy said.

Poor data quality. Lack of data stewards can lead to poor data quality that inhibits adoption of new technologies and infrastructure, Claire Thompson said. Without data stewardship, organizations can lack effective screening for financial crime obligations and decision-making around business and customer initiatives. The result is inaccurate reporting and inefficient service delivery and outcome for customers.

Lost opportunities. Not having a data steward reduces proactive uses for data when companies are trying to enable new business outcomes, Graeme Thompson said. For example, if a customer experience program fails to understand consumer buying preferences across multiple channels, it will miss opportunities to improve products or recommend similar offerings. The lack of a data steward can also make it harder for enterprise data consumers to find data that they can trust.