Using Data Governance To Drive AI at Scale

This article was written by Rana Abbaszadeh, a Senior Associate in Metis Strategy’s West Coast Office

As companies look for ways to harness data and AI to deliver on business outcomes, they first need to develop the foundational governance capability that enables them to do so effectively. Data governance requires significant time and resource investment, to be sure, but it ultimately enables organizations to realize the long-term value from their AI and analytics initiatives. 

At a high level, data governance refers to the development and management of information about an organization’s data. It includes maintaining a catalog of a company’s data from lineage to definition and utilization. When done well, data governance creates a single source of truth that can be used to unlock trusted insights, inform strategic decision making, and enable personalization at scale.  

Companies that implement data governance can:

• Develop a reliable source of data to improve business outcomes and performance
• Enhance data security and quality
• Improve compliance with data protection laws and regulations 

Metis Strategy takes a strategic approach to data governance and recommends that organizations start with the data that drives significant value. For example, a retail company could focus first on the governance of customer and product data, as this information is core to the company’s growth. Focusing on high-value data helps generate buy-in from key stakeholders and builds momentum for governance initiatives. After that, organizations can turn to other data until governance becomes embedded into the company culture. 

This article will outline how to develop a data governance program within your organization, including the different roles and stakeholders involved.

Identifying governance opportunities

Using the Metis Strategy methodology, organizations can quickly realize value while improving overall data maturity. We recommend developing a cross-functional steering committee consisting of senior leaders across business and technology units who will guide the governance process. The steering committee is responsible for setting strategy, direction, and prioritization for the data governance program.

The committee’s primary responsibilities include: 

• Identifying high ROI use cases. Create a framework for determining the value of a governance project versus its expected cost. High ROI use cases may include governing customer and product data, financial data, or any sensitive consumer data. Once the use case has been chosen, identify the data required.
• Developing an MVP for the organization. After selecting an initial use case, create a governance council to implement the program. Leaders should ensure the program aligns with the organization’s business strategy and requirements. Once aligned, data cataloging can begin. This involves providing consistent definitions for data and managing data’s lineage. After this process is complete, the data is ready to be used by the organization.
• Monitoring progress and repeating with new use cases. The steering committee will oversee the MVP in action, monitoring adoption and measuring return on investment. These metrics help frame the business case for future governance initiatives, with the ultimate goal being an enterprise-level data governance capability. 

In addition to the responsibilities above, the committee also will evaluate the business case for specific initiatives, approve funding and resource requests, and guide program adoption throughout the enterprise.

Building the Governance Council

In addition to the steering committee, the data governance program should include a governance council that will scope, document, and monitor data assets and lead governance operations. The council should consist of individuals across different business units to provide varied perspectives across domains. Members take on roles such as data owner, steward and custodian to ensure accurate data sets for their respective business units. A high-level overview of this is shown below.

The Data Governance Council consists of several roles with varying responsibilities. Metis Strategy recommends the council have at least the following three roles:

• Business data owners are responsible for providing guidance and recommendations as to how data is used within their domain and have ultimate authority over business unit data. They have a domain-specific understanding of how data is used and manage data definitions, utilization and cost.
• Data stewards and custodians ensure data-related work is performed according to operating procedures set forth by the business data owners. They are responsible for defining usage permissions and policies, and working across business and technology teams to identify, define and standardize data at the implementation level, managing issues as they arise. 
• Delivery execution teams are data or IT teams responsible for data modeling, usage tracking and access management defined by the data stewards and custodians.  They ensure data is available, accessible, well performing and recoverable.

Business unit end users

Business unit end users will have access to trusted data based on their business unit needs and role requirements. They will collaborate with the business data owners to ensure maximum utility of the enterprise data.

Conclusion

Data governance is critical to ensuring the success of strategic data projects across any organization. Having the right structures in place will enable a faster return on investment and allow the governance capability to scale throughout the organization. As more high-value use cases come to life, analytics and AI teams will be empowered to use trusted data to improve business performance, enhance the customer experience and improve operational efficiency.

Companies have had great success in initial governance efforts, unlocking the utilization of customer and product data to help drive product design and improve sales outcomes. For example, after developing a governance program around its consumer and product data, one retailer improved the personalization of a merchandising ad unit by 17% through an enhanced understanding of user engagement and behavioral patterns. Success in this area helped the company make the business case for future analytics and AI use cases. In this case, a strong data governance capability built confidence and momentum for the organization as it continued to scale its analytics efforts. 

To learn more about developing a robust data governance program, please contact us at information@metisstrategy.com