Four steps to data governance success

1. Lay a foundation
for consistency

Organizations generally start with different departments, each with terminology and descriptions for the same data assets.

This inconsistency hinders everyone since there isn’t a common language for discussing data.

One way to resolve this issue is to create a common business glossary that will allow your organization to:

Standardize business terms
Establish data standards
Improve overall efficiency and productivity

In a federated governance model, different lines of business may create their own definitions. It’s particularly important to link those definitions to datasets and data products so users can understand what the data means.

2. Establish confederation

Moving to a cloud-native hybrid data landscape can help companies become more agile and capable of scaling and pivoting in uncertain markets. Businesses can solve the most difficult business problems by sharing data, context and analyses.

But it doesn’t solve every problem. Enterprises are made up of different departments, teams and business units, each with its own operating style and needs. This diversity can make it challenging to consistently govern data.

A successful approach to data governance will unify access and control features — but it won’t be one-size-fits-all. Instead, it will provide a way for everyone to participate in a single unified system of data governance while allowing freedom to design a model that enables data owners to define roles, responsibilities and terms while integrating into the broader goal of becoming data-driven.

3. Enable certification

Establishing confederation is a significant milestone but maintaining it will require implementing higher-level business processes. These can include setting up a regular data standards refresh cycle, creating an update process for curated data and implementing best practices for handling certified assets. The main goal for this step is to separate raw data from data that’s been curated or certified, which is crucial for ensuring vetted, high-quality datasets that you feel confident about.

4. Execute continuous improvement

Once a certified set of well-monitored data products is available to data consumers, a data governance council can oversee the ecosystem of data standards, policies, definitions and new data product requirements. The council should consist of department and central data team members.

The council should use metrics to monitor data quality, availability and delivery to data consumers. Based on these metrics, the committee can approve new investments in projects that most benefit the organization. They can also monitor data incidents and faults and approve new investments to improve data quality or inefficiencies in particularly high-risk or complex data pipelines.

The council should conduct business reviews monthly or quarterly to understand the state of affairs and guide continuous improvement initiatives. In essence, they govern the data governance process.