Data Stewardship (2014)

CHAPTER 2 Understanding the Types of Data Stewardship

Each of the types of Data Stewards is defined and the roles and responsibilities described. The types of steward include Business Data Stewards, Technical Data Stewards, Operational Data Stewards, Domain Data Stewards, and Project Data Stewards.

Keywords

Data Stewardship; business; operational; domain; project; technical; responsibilities

Introduction

The generic term Data Stewardship (as defined in Chapter 1) primarily encompasses two different types of stewards: Business Data Stewards and Technical Data Stewards. However, on occasion, there is a need to “fill in” with other flavors of Data Steward, including Domain Data Stewards and Project Data Stewards. These last two are support roles, picking up some of the load when the Business Data Steward is unable to handle everything. In addition, Business Data Stewards may get help from Operational Data Stewards, yet another support role. In this chapter we will discuss and detail each of the different types of stewards, when and why they are necessary, and how they work together to achieve the goals of Data Governance. In Chapter 3, we will explain the detailed responsibilities of each type of Data Steward.

Figure 2.1 shows how the different types of Data Stewards interact. Business Data Stewards represent their business area and steward the data owned by that area. Business Data Stewards get help from Operational Data Stewards, who are usually people who work directly with the data (e.g., by doing input) and can provide immediate feedback to the Business Data Stewards when they notice issues with the data, including declining data quality. Domain Data Stewards are the Business Data Stewards for data for which ownership must be shared across multiple business areas, and for which no one business area can have the ultimate say on what happens with the data. A typical example of a shared domain of data is “customer.” Project Data Stewards represent Data Stewardship on projects, reporting back to the appropriate Business Data Steward when data issues arise on the project or when new data must be governed. Finally, Technical Data Stewards are IT representatives with knowledge about how applications, data stores, and ETL (Extract, Transform, and Load) processes work. In the rest of this chapter we will look at each of these types of Data Stewards in more detail.

FIGURE 2.1 Interaction among the various types of Data Stewards ensures that questions get answered efficiently and correctly.

Business Data Stewards

So what do we mean by a Business Data Steward? A Business Data Steward is the key representative in a specific business area that is responsible for quality, use, and meaning of that data in the organization. Business Data Stewards make specific recommendations about the data to the Data Governors.

Business Data Stewards are typically individuals who know the data and work closely with it. People with titles like Data Analyst or Business Analyst are often good candidates to be Business Data Stewards. Of course, no one can know everything about all the data. The Business Data Steward should know who to talk to in his or her area to get the information needed to fulfill the role. And once a Business Data Steward has gathered that information, he or she (not the person providing the information) is responsible for stewarding the data. Having access to other resources who work with data means that the Business Data Steward has the authority to obtain some of the resource’s time to help in the stewardship effort. If requests for contributions from other resources go unanswered or are refused, escalation to the Data Governors is appropriate.

Choosing the Right Business Data Steward

Choosing the right person in the business area to be the Business Data Steward is crucial, not only for the success of the program, but also for the person’s job satisfaction. As stated earlier, the people chosen to be Business Data Stewards need to know the data and the issues with the data. They should be aware of where the data is not meeting business needs, as well as data elements of which the meaning is not well understood or is a “bone of contention.” Most business areas have people who care about the data and can even be said to have a passion for the data. These people are good candidates for the role. In fact, many business areas have individuals who everyone goes to in order to get questions answered. They have become practiced at answering questions about meaning, business rules, what data to use for what purpose, and so on. In fact, these local experts often spend a good deal of time answering the same questions over and over. In many cases, Business Data Stewardship simply formalizes this role and these responsibilities. Formalization of the role provides the following benefits to the Business Data Steward:

- Answer the question once. Once a question has been answered and documented in the business glossary (in whatever form that takes), the Business Data Steward can encourage the person asking the question to “look it up.”

- Fewer arguments. Since the Business Data Steward is a recognized authority on the data and has the ability to make decisions (with the backing of the Data Governor), there should be fewer arguments and less wasted time. Having this authority is great incentive for getting people to buy into the role of Business Data Steward. It can be very frustrating to people who care about their data if all they can do is advise others, who may then go off and do whatever they please.

- Repeatable process for changes. If the data decisions that have been made are not working for one of the stakeholders, having a repeatable process to engage with the Business Data Steward and other interested parties, followed by an update to the decision and documentation, should cause far less confusion than the typical “form a committee, have a bunch of meetings, never get resolution” method that seems to happen so often.

- Additional rewards and recognition. The data expert role is rarely recognized as being of value and compensated for. With the formalization of this role, the individual’s performance objectives (which usually are tied to compensation) can be adjusted and this expertise and responsibility formally recognized.

To summarize, it is quite appealing to people who care about their data to be in charge of it so that others can’t use it in inappropriate ways or corrupt it.

Being a Business Data Steward Can Save You Time!

Not everyone tapped to be a Business Data Steward is going to look on this assignment with joy. On the face of it, it is another “job” that is being heaped onto a usually already overworked individual. And indeed, there are new tasks to be learned, some tools to become familiar with, and meetings to attend. But to balance this out, quite soon the constant interruptions from colleagues and endless meetings about “your” data become a thing of the past as people learn to use the business glossary and engage with the Business Data Stewards using formally designed processes. A real-life example of this was the Actuarial Business Data Steward at an insurance company. She estimated that prior to Data Governance being instituted she spent about 40% of her time dealing with data questions and issues, rather than creating and running risk models. After approximately six months with Data Governance in place, she estimated that this had dropped to less than 10%. She became one of the biggest proponents of “getting stuff into the glossary” then making other analysts use the tools.

Characteristics of a Successful Business Data Steward

In addition to being interested (or passionate) and knowledgeable about the data, a successful Business Data Steward needs several other characteristics. These characteristics may well be harder to find than the data knowledge!

- Writing ability. Since Business Data Stewards are involved in writing definitions and business rules, it would be most helpful if they are able to write well. Further, Business Data Stewards have an active role in drafting procedures; so again, being able to write well is a huge plus.

- Committed to improvement. The best stewards are the ones who strive for excellence and aren’t satisfied with the status quo. It is extra work (despite the return on that investment) to drive to well-defined, high-quality data, but everyone is better off if this work is done. Obviously if a steward isn’t interested in an improved data environment, that steward is not going to be very effective.

- Good people skills. Stewards need to work together not only with each other, but with other stakeholders to try and reach agreements about meaning, location, and required quality of the data. This may mean setting up and leading meetings, collaborating with others, following up on issues, and publishing the results (writing!). All of this requires leadership, organizational, and facilitation skills, as well as a certain amount of political acumen.

Stewardship and Trust

One of the biggest “gotchas” in today’s business world is a tendency to mistrust the data. Mistrust causes a lot of extra work (and potentially wasted effort) as data analysts check the data, trace it back to the source, extract it into their own environments so they can manipulate it, and so on. And why do the analysts mistrust the data? It is because the data can so often be full of surprises! These surprises (as illustrated in Figure 2.2) can be caused by many variables, including undocumented changes to business processes, transformation issues, source data problems, uncertainty as to what the data means, and changes to the technical applications and processes. Without assigned stewardship, there is little formal accountability for researching the impacts of changes, making the changes, or communicating the changes. IT personnel may well want to work with the business to figure out the impacts of proposed system changes, but they are unable to find anyone in the business willing to step up and work with them. Data Stewards (both Business and Technical) take much of the uncertainty out of these processes, and can greatly reduce the number of data surprises, and thus increase the level of trust in the data.

FIGURE 2.2 Stewardship can prevent data surprises, and lack of trust.

Note

On occasion, there can be stewardship without a formal program. If there are critical processes that depend on the data, people will often “step up” and act accountable because the processes fail if they do not. This is especially true with finance data. However, without a formal Data Stewardship program, that accountability can vanish if the individual goes on vacation, accepts another assignment, or retires.

Business Data Stewards and the Data

As we said earlier, Business Data Stewards steward the data most critical to the business function they represent and that they are most knowledgeable about. This connection must be determined—that is, what data is the steward responsible for—and documented and available to everyone.

The simplest way to link a Business Data Steward to data is via a data domain, as shown in Figure 2.3. The approach of connecting the Business Data Steward to a data domain is much less complex than linking at the individual data element level, though also less flexible. In this context, the data domain refers to a grouping of related data that has meaning to the business. For example, a healthcare company might have groupings of data for drug, prescription, protocol, clinical indication, geographic region, patient, and so on. Notice that on occasion these data domains look much like master data. Master data (data for which multiple instances of the data exist across the enterprise and for which it is advantageous to have a single reconciled or “golden” copy) is often described as a data domain.

FIGURE 2.3 Data Stewards are often responsible for a set of related data domains.

While the idea of linking Business Data Stewards to a data domain is relatively simple on the face of it, you have to be able to state who the Business Data Steward is down to the data element level. The main reason for identifying Business Data Stewards down to the data element level is that people who need to identify the Business Data Steward are usually doing so because a question has arisen about a particular set of data elements. In the case of assigning responsibility at the data domain level, the Business Data Steward is de facto responsible for all the data elements in that domain. However, you may also have the case where there is a split in the responsibility of the data elements in a domain, as shown in Figure 2.4. That is, some of the data elements in a domain are the responsibility of one Business Data Steward, while other data elements in the domain are the responsibility of other Data Stewards.

FIGURE 2.4 Data Stewards are responsible for a set of related data elements.

In the real world, even individual attributes of a data entity can have different stewards, and it is important to get the Business Data Steward right, as we’ll discuss later. For example, some of the attributes of an insurance policy may be owned by underwriting, while others are owned by marketing or sales. So, in the end, you really need to establish stewardship at the granular level of the data element.

The Key Role of the Business Data Steward

Although there are many other types of stewards (as described in the rest of this chapter), Business Data Stewards are the authorities on their data, in that they know what the data is supposed to represent, what it means, and what business rules are associated with it. Business Data Stewards coordinate their work with many others in the enterprise, including the Technical Data Steward and business stakeholders. The stakeholders’ concerns must be taken into account when decisions are made about that data. The relationship between Business Data Stewards, Technical Data Stewards, and stakeholders was shown previously in Figure 2.1.

Business Data Stewards work closely with Technical Data Stewards and business stakeholders to influence the data along the entire information chain (the flow of data from the source to various targets such as business intelligence) as shown in Figure 2.5. The partnership of Data Stewards and stakeholders makes it possible to manage the data as it flows through the enterprise.

FIGURE 2.5 Business Data Stewards work with others to manage the data in the information chain.

Note

An information chain is a supply chain for data. It shows where the data originates, how it is modified and stored, and where it is used. The information chain is an important concept because data must be managed across the entire life cycle. Many thought leaders recognize the concept of an information chain producing data as similar to a supply chain producing physical products.

Technical Data Stewards

Technical Data Stewards are IT personnel who have an important role in supporting Data Governance. But where Business Data Stewards are associated with data domains or specific data elements, Technical Data Stewards provide support and are associated with specific systems, applications, data stores, and technical processes, such as identity resolution (for Master Data Management, MDM), data quality rule enforcement, and ETL jobs. This relationship can be more clearly seen in Figure 2.6. That is, Technical Data Stewards are the people to turn to in order to understand how the data is created, manipulated, stored, and moved in technical systems. They can answer questions about how data got to be a certain way.

FIGURE 2.6 The relationships between Business and Technical Data Stewards.

As an example, consider a card membership system where data profiling revealed a significant anomaly in the distribution of birth dates. Although most days of the year had approximately the same number of members born on that day, the number of members born on December 31 was over 150 times higher than any other day. The Technical Data Steward for that system explained this anomaly: the previous system from which the data was converted contained only the birth year. During the conversion to the new system, December 31 was used as the default date. In another example, it was discovered that there were over 4,000 body types (e.g., four-door sedans, pickup trucks, etc.) listed for insured automobiles. The Technical Data Steward explained that the field in the application that captured the description of the body type was a free-form text field with no validation, so agents could enter whatever they wanted. Further, the Technical Data Steward revealed that it really didn’t matter that the data in the body type field was a mess, since this data was not used by the application—instead, the body type was extracted from the automobile’s VIN number by another part of that same application.

As you can see, this sort of information is not only valuable, but critical to a robust Data Governance or data quality effort. With additional knowledge about the system data, the data consumers can make better choices about which fields to use and which fields to fix. Having this knowledge increases the trust in the data. In many organizations there are IT personnel who can answer questions as they come up, but Technical Data Stewards are very different from these “casual” contributors because Technical Data Stewards are:

- Assigned by IT management. Working with Data Governance is considered an important part of their job, and not a “do it when you have time” task.

- Responsible and accountable for providing answers in a timely manner. They understand the importance of the Data Governance/Data Stewardship effort and their role in that effort.

- Part of the Data Stewardship team. They are not being asked random questions in a vacuum, they are kept up to date on Data Stewardship activities, and data management tasks are planned and integrated into their schedules. If the Technical Data Stewards are concerned about the level of their involvement and ability to get the work done, they have a chain of command (including the Data Governance Manager) to make those concerns known to.

Project Data Stewards

Ideally, there should be Data Stewardship representation on all major projects to provide support and make decisions about data meaning, usage, quality needed, and business rules. However, especially in companies with multiple projects going on at once, it simply isn’t practical to have the Business Data Stewards present at all meetings and work sessions on the possibility that they will be needed. To help take the load off the Business Data Stewards but still involve them where they are needed, you can use Project Data Stewards. A Project Data Steward is an individual who is trained to key in on issues and questions that require input from the Business Data Stewards, and who brings that information to the Business Data Steward. The Business Data Steward then provides input, which the Project Data Steward brings back to the project. Project Data Stewards are assigned to projects, and depending on the project’s size, may cover multiple projects.

Project Data Stewards can be analysts already assigned to a project (if they have the proper training), a member of the Data Governance Program Office (DGPO), or even a contractor. Major projects are expected to fund the extra cost of Data Stewardship involvement, so it is important that the project methodology include the Data Governance tasks so that the estimate and budget will include those costs.

But what about smaller projects, which typically have small budgets? Small projects can usually get by making use of an analyst on the project or a small to moderate percentage of a Project Data Steward attached to the DGPO. In fact, in my last Data Governance Management role, we funded a single full-time person to supply this service to small projects. She was well liked and well respected throughout the company, having worked there more than 30 years, and was welcomed onto each project.

Project Data Stewards are Not Business Data Stewards

There is a major difference between a Business Data Steward and a Project Data Steward: the Project Data Steward is not responsible for the decisions about the data. This distinction is important, because once a Project Data Steward gains some experience and gets to know the data, there can be a tendency to start making decisions on his or her own. This sort of behavior must be nipped in the bud, because any decision made by the Project Data Steward does not carry the authority of the Business Data Steward or the Data Governance program. The Project Data Steward is responsible for:

- Recognizing and recording issues and questions that require the input of the Business Data Steward.

- Coordinating the list of issues and questions with other Project Data Stewards, because multiple projects may well come up with the same issues and questions.

- Presenting the list to the appropriate Business Data Stewards and recording decisions and input.

- Providing the results as input back to the project.

Domain Data Stewards

One of the key Business Data Stewardship responsibilities is to assign ownership of key business data to the appropriate business area. In many cases the ownership should be clear. In an insurance company, for example, claims data is owned by the claims area, policies are owned by underwriting, and the risk models are owned by actuarial (or their equivalent). This determination (as we’ll discuss more in Chapter 6) mostly comes from which business function’s business would experience a significant change if the meaning or business rules for the data element were to change. But what about data that is so intrinsic to the operation of multiple business functions that it simply can’t be owned by just one? A classic example of this is “customer,” but other master data often falls into this category as well.

One possible way to solve this problem is to designate a Domain Data Steward. The Domain Data Steward is the official steward of the shared data, but is also responsible for keeping track of which business functions have a stake in that data and reaching an agreement with those stewards on any proposed changes. The Domain Data Steward must document the list of consulted business functions, and is responsible for convening meetings or finding other ways to reach consensus on changes, then recording those changes just as any Business Data Steward would do. A Domain Data Steward may be one of the “regular” Business Data Stewards (as long as everyone trusts that person!), but is often a member of the DGPO assigned to handle these duties.

Operational Data Stewards

Operational Data Stewards are “helpers” for Business Data Stewards. They can step in on some of the duties that mesh with their jobs. For example, they can help ensure that data creation rules are followed, or assist with researching issues. Oftentimes, Operational Data Stewards are people who work on the frontlines, and see early opportunities where improved data quality would benefit certain groups. Business Data Stewards can designate Operational Data Stewards to help them, though the Business Data Stewards remain responsible for the data owned by their business area.

Summary

There are a variety of different types of Data Stewards, including Business, Technical, Project, Domain, and Operational Data Stewards. Each type of steward has a set of responsibilities as summarized in Table 2.1. These stewards all need to work together (as detailed in Figure 2.1) to get the data fully stewarded.

Table 2.1

Summary of Stewardship Responsibilities