You will be able to get some idea of Data Stewardship in Data Quality Organization as well as in BI Roles and Skills (almost duplicate text). In-Brief- Data Steward is responsible for making sure that the data-elements within the organization (in whatever form) are in good-health, as they are used for various Data Management and BI applications. In other words, Data Steward is like the owner of the quality of cooking ingredients (like flour, vegetables) and intermediates (like Onion paste..) in a kitchen.
A data steward is not the custodian of the data, as that is owned by the rightful business owner. However, a data steward is responsible for the data quality and health deliverables of the data custodian. In a way, data custodian is responsible to the data steward.
Organizations have been facing challenge to institutionalize the role of a data steward as it is not a function like IT, Finance etc.
Here are some recommendations for making the best out of a data steward:
Make Data Steward accountable for defining the data quality goals
First things first- Defining data quality goal involves a fair degree of business and IT understanding. It requires a great degree of negotiations, facilitation and education with the stake-holders to agree upon the goals. Businesses typically want 100% data quality, but share-holder may look for business-case.
Make Data Steward accountable for the Data Quality KPIs
Data Steward should be orchestrating the laying down of the agreed KPIs for data quality. Data Quality has various dimensions like accuracy, completeness, consistency etc.. A data steward, with an understanding of the data groups and their respective criticality should drive the agreed KPIs and the expected standards.
Make Data Steward accountable for delivering on the Data Quality Goals
This is when rubber hits the road. A Data Steward should swim or sink with boat of data Quality. He should be made single point accountable for meeting the data quality goals.
Empower and rightly position the Data Steward
A Data Steward should be empowered, so that the stakeholders, which are responsible for delivering on the tasks to achieve data quality goals (process improvement, better controls, system automation..), should see data steward as their business customer.
A Data Steward can also be given a Data quality budget for driving the data quality initiatives.
A Data Steward should be positioned well in the organization in terms of levels and hierarchy. A Data Steward can be reporting to CEO, COO or to CFO.
Data Steward should be part of Business
ExecutionMiH.com has always professed that Data Quality as well as BI is mainly a business domain and most of the successes and failures are due to what we do or not do in business. Data Steward is a business role. Data Quality domain is not only limited to IT and also most of the Data Quality solutions are linked to processes or to the business rules (as defined by business), which are automated by IT.
Data Steward should be a subject matter expert on the businesses, processes and IT interface.
If we have to make a choice, we will choose business-expert data steward instead of IT Database and business systems analyst. A tech interface and back-ground is always a great plus, but it is the business and process knowledge, which is the key. Most of the root causes of data quality are linked with what data goes through before or at the time it is entered into IT systems.
Assign functional level Data Stewards
We need to assign functional level ( or even sub-function level) data stewards to support the enterprise level data stewards (who could be the head of the information management, or Head of BI etc..).. The reason is that not a single person can have the expertise of the entire business domain.
Assign distinct data stewards for distinct data group
Just like data custodians, its recommended that data stewards should have distinct ownership of data-groups. For example two data stewards should not have the ownership of leads data, or customer data or Vendor data etc..This means that there will cross-functional linkage (Data steward for the customer data, will have linkage with Operations, Marketing and sales.). This cross functional linkage can be well managed with appropriate procedures and empowerment. Given this cross-functional linkage, a function with maximum stake with a data group will have the data-steward from that function owning that Data Group. (for example customer data steward could belong to sales function, if it has the maximum linkage in terms of creating and changing customer data)
Avoid process-based data steward
I have heard suggestions that a data steward should be process based, as that is the direction in which a data flows. Execution-MiH recommends against that approach, The reason is that a data group is linked to many processes and a process is linked to many functions. This will create an unmanageable matrix.