Subscribe
About

Time to get serious about data stewards

Failing to appoint data stewards is a factor that will undermine the chances of data quality success.
Mervyn Mooi
By Mervyn Mooi, Director of Knowledge Integration Dynamics (KID) and represents the ICT services arm of the Thesele Group.
Johannesburg, 16 Apr 2008

How important is a data steward to the success of a data quality project? This is a vital question: a data steward may have been viewed as a nice-to-have function. But increasingly, companies which take the quality and integrity of their data seriously are coming to understand and internalise that the role of the data steward is vital to business success.

Gartner is unequivocal in this regard. It says: "Organisations striving to improve data quality must consider appointing data stewards."

"Data quality is a business issue, not an IT matter, and it requires the business to take responsibility and drive improvements," says Andreas Bitterer, research VP at Gartner.

As long ago as 1998, Ralph Kimball, the "father" of data warehousing, identified the need for a data steward: "The data steward is responsible for enterprise agreement on the warehouse's conformed dimensions and facts. Clearly, this is a politically challenging role," Kimball wrote in The Data Warehouse Toolkit.

Kimball identified the role as a clear and unambiguous one, alongside the business sponsor, business driver, business system analyst, analytic application developer and business subject matter expert, to mention a few.

Yet we still find many organisations which do not appoint data stewards, a factor which will undermine the chances of data quality success. Data stewardship is still often viewed as a secondary "background" process.

The fact that many organisations still do not appoint data stewards as a matter of course, points to an organisational culture that still views data as a necessary evil rather than a competitive asset, as Gartner neatly puts it.

It is the responsibility of the data steward to ensure that each data element:

* Is clearly and unambiguously defined and in context of its purpose.
* Does not conflict with other data elements in the metadata registry. Here the data steward must remove duplications and redundancies, and resolve potential overlaps.
* Has clear enumerated value definitions of the type "Code": in metadata, this refers to and is used in the name of data elements whose value can be represented as enumerated lists: for example, PersonGenderCode.
* Is still in use (the data steward needs to ensure inactive or unused data elements are removed).
* Is in constant and consistent use in various systems and applications.
* Is properly and adequately documented on its appropriate usage, with support notes.
* Can logically and accurately document the origin and source of authority on each metadata element.

It is important to note here that data stewards are the trustees of data, rather than owners of it. Because of the importance of the role, it is highly advised that the organisation has multiple data stewards, and that they are drawn from departments, where the person chosen is likely to be a subject matter expert.

It is important to note here that data stewards are the trustees of data, rather than owners of it.

Mervyn Mooi is director at Knowledge Integration Dynamics.

In this scenario, then, HR, marketing, sales, service and finance would each have their own data steward, and they would be responsible for maintaining the integrity of the data, keeping it comprehensive, accurate, consistent, current, and free of redundancies. They would also be required to ensure their department's data conforms to organisational data quality standards.

The data stewards must enjoy the determined and committed support of the executive. As data quality is a boardroom issue, the boardroom must lend the data stewards their total commitment. The data stewards must carry respect within the organisation and they need to be seen, and to be seen to have an impact. They need to set a vision, communicate it, obtain approval, evangelise each success, and generally run their data stewardship like a business.

The benefits for the organisation will be multi-faceted and significant. Business intelligence initiatives and other applications like CRM will have positive, more successful and more consistent outcomes, For example, postal costs will drop and the constant reengineering of data quality initiatives will slow down.

Furthermore, given the reported cost of poor data quality to businesses worldwide (billions a year), there will be a direct benefit to the bottom line.

So it's time to get serious about data, and to get serious about the role of the data steward.

* Mervyn Mooi is director at Knowledge Integration Dynamics.

Share