Benjamin Haas - Fotolia

Kaiser's data governance model leans heavily on data stewards

The head of Kaiser Permanente's data governance program says data stewards hold the key to the initiative's success, and he offers advice on managing data stewardship processes.

Data governance programs are built on an intricate framework of policies, procedures and rules for using and managing...

corporate data. But the human element, in the form of data stewards, is what really solidifies a data governance model.

That's the view of Fawad Butt, chief data governance officer at healthcare provider and insurer Kaiser Permanente. "At the end of the day, the real operational work happens through data stewards," he said, pointing to their central role in creating, implementing and enforcing common data definitions and governance policies throughout an organization.

Butt joined Kaiser in September 2015 to lead the development of an enterprise-wide data management and governance strategy. He and his team identified 23 key data domains across the health system's operations, and he expects to have data stewards in place for all of the domains in the near future.

Kaiser, which is based in Oakland, Calif., wants to tap its growing data assets to improve care for the nearly 11 million people it serves -- and Butt said he's counting on the stewards to ensure data sets are maintained and used properly so the information they contain can be trusted.

In a presentation during the 2017 Enterprise Data Governance Online conference last week, Butt shared tips on managing data stewardship efforts, based on his experiences at Kaiser and several financial services companies. Here's some of the advice he offered to attendees listening via the web.

Think in multiple dimensions. Data stewardship "is not a one-dimensional construct," Butt said. Kaiser's data governance model includes data stewards who are responsible for particular types of data throughout the organization, but also ones who deal with data from a functional standpoint -- the latter are aligned with individual departments or regional operations, and their responsibilities extend across all of the data sets used by the business units they support. The combination of the two approaches is where "the real magic happens" in stewarding data, according to Butt.

The roles of the different data stewards are largely similar, "but they need to understand data in different ways," he said. Stewards assigned to data sets on a vertical basis must develop a deep knowledge of the data, while their functional counterparts working horizontally with multiple data sets have to provide context on how data is used in business processes.

Kaiser is also looking at adding a third dimension to its data stewardship plan: assigning stewards to oversee the data in specific IT platforms. For example, Butt said data lakes supporting big data applications are potential candidates for stewardship treatment because they're typically populated with a wide variety of information that's relevant to multiple business units.

It's just too difficult to train people to become stewards and keep them focused if they have other responsibilities.
Fawad Buttchief data governance officer at Kaiser Permanente

Stewardship should be a full-time commitment. Data steward isn't always a full-time role in companies, but it is at Kaiser. "It's just too difficult to train people to become stewards and keep them focused if they have other responsibilities," Butt said. He acknowledged that dedicating people to data stewardship might be difficult due to factors such as company size, resource constraints and allotted governance budgets. But he said a part-time focus affects the ability of stewards to get things done.

Make data stewardship a shared responsibility. Butt's team is part of the IT department at Kaiser, and it's initially managing the data governance program centrally. But, he said that as the data governance model matures, he plans to move to a more federated approach that gives the business side some control of the process, including data stewardship. Some of the data stewards, primarily the functional ones, come from business units. They all report to Butt for up to 24 months -- then, the idea is to switch to a hybrid organizational structure. That way, he said, business units "kind of have skin in the game, too."

Stewards need a blend of different skills. While data stewards don't have to be IT professionals, they should have some experience in dealing with data, Butt said. Otherwise, they could face credibility issues internally, he warned. He thinks stewards ultimately require skills not only in governing data, but in architecting and administering it. That's not all, though: Butt said they also need leadership, communication and problem-solving skills, as well as planning and project management know-how.

Training is a critical first step. Finding people with all of the required data stewardship skills upfront isn't always easy, so some of those skills might have to be built up via training, Butt said. In addition, he has made training on data stewardship roles and responsibilities a big priority at Kaiser. His team created a 100-page "playbook" that details internal standards underlying the data governance model and provides templates to use in stewardship efforts. It also runs a 2 1/2-day training program for data stewards, covering things such as Kaiser's organizational culture, change management techniques and the data requirements of different business units.

Another element of the training regimen is learning lessons when data stewardship work goes wrong. "Instead of running away from failures, I want to go find out what happened -- why things failed and how to avoid them," Butt said.

Next Steps

Get more tips on managing data stewards, from consultant David Loshin

How metrics on data quality gains help show the value of data governance

Key features to look for if you're looking to buy data governance software

Dig Deeper on Data stewardship