Welcome to our blog. This is an archived post, most of our knowledge and advice remain valid but some material or links may be outdated. Click here to see our most recent posts.
This is the fourth blog post in the Data Management series. So far we have discussed Entities and the way they are realized in applications / information systems. This time we zoom in on Data Stewardship.
The Data Steward is a key role in successful Data Management. To put it more strongly, we will argue that without a Data Steward (or Steward, in short) most Data Management initiatives are doomed to fail.
The steward is an organizational role. The professional tasked with this role is typically responsible for data quality of the Entities in a Subject Area from a business perspective (in a recent talk at the Master Data Management/Data Governance summit in London, Analiese Polsky of SAS mentioned five models for stewardship. However, stewardship aroundinformation areas is still the predominant approach which is why we adopt it here).
In some cases a Subject Area is too large for one Steward to manage, and therefore it becomes a joint responsibility of a group of Stewards.
The Steward works with Data Professionals (i.e., IT staff) to make sure those requirements with respect to data are realized in an effective manner. Typical tasks for the Steward are: data modeling, data definition, data quality requirement specification and data quality improvement, reference and master data management.
It should be noted that a Steward is not a technical role. On the contrary: it is a prototypical business role. Deep business knowledge is essential to be able to articulate business needs with respect to information, and to negotiate the priorities with fellow stewards and IT, resolve conflicts in semantics, budget, and availability of staff etcetera.
In order to co-ordinate the work between data stewards, many organizations typically have a Data Council at the Enterprise level. This is where alignment issues are resolved and where strategic decisions with respect to data management are made. If necessary (especially in larger organizations), a level of stewardship can be created in between as illustrated by the following diagram that was taken from the DAMA DMBOK:
This topic does not require a lot of specialized modeling constructs. The most important one is to be able to represent the fact that a Steward is responsible for a Subject Area / Entity / Data Object. The standard way of modeling would involve a BusinessRole for the Steward, assigning it to a BusinessProcess that has access to the relevant passive structure elements.
In terms of visualization this is often too complex. We have found that a shorthand notation where a BusinessRole (the Steward) is associated to the BusinessObject (for SubjectArea or Entity) or DataObject works well for communication purposes.
The Data Council can be regarded as a body where stewards collaborate. This should be modeled using the BusinessCollaboration concept using standard ArchiMate notation. To avoid visual complexity, visual nesting (drawing stewards inside the council) is preferred over using graphical relations (i.e., drawing a line between the concepts).
In terms of analysis there are some interesting things we could do:
Select a Steward and generate a view with all Entities and associated Data Objects that s/he is responsible for
Select a Steward and highlight (color? Highlight view?) the Entities and Data Objects that s/he is responsible for
Generate a matrix-view that shows Stewards on one axis, Entities on the other axis and the names of the associated Data Objects at the intersection
Generate a matrix view that shows Stewards on one axis, Entities on the other axis, and Systems that manage (the C, U, D parts) a Data Object that realizes this Entity in between.
Select an Entity or Data Object and highlight (color? Highlight view?) the Steward who is responsible for it
Select a System and show with labels which Data Objects are managed in that system, and who the associated Steward is. A format could be “Steward: AAA – DataObject: BBB”
Doing this type of analysis is fairly straight forward in most modern ArchiMate tools If you are interested in a demo of this type of functionality in BiZZdesign Architect, please leave us a note!
The next posting will be about Master Data Management, one of the corner stones of the field of Data Management. It is often said that “You can do data governance without Master Data Management. You can also do Master Data Management without Data Governance, but that would be a bad idea”. An exciting topic, so stay tuned!
SUBSCRIBE TO BIZZDESIGN'S BLOG
Join 10.000+ others! Get BiZZdesign's latest articles straight to your inbox. Enter your email address below: