I would like to return to the definition of data governance given in Part 1 and explore some other aspects of the issue. The definition was data governance (DG) refers to the overall management of the availability, usability, integrity, and security of the data employed in an enterprise. A sound data governance program includes a governing body or council, a defined set of procedures, and a plan to execute those procedures.
My initial reaction to this definition was that it was some very technical folks creating another complex construct. However as we experience sharing of data throughout organizations, how does the organization deal with multiple versions of the truth? The solution is to create one version which everybody shares. The governance part is making this simple statement a reality.
If we take a case where several people in one division of an organization are developing performance data on many spreadsheets. The management in this organization realizes that all these spreadsheet require a lot of manual effort and could be done more efficiently by a data warehousing application. Different parts of this division report sales and revenue in quite different ways. In addition they do not align with the way other parts of the organization report sales and revenue. The reason is that each product line has quite different characteristics. When this data is on different spreadsheets, common definitions are not necessary.
We are asked to help this division develop this application on the warehouse and we recognize immediately the data governance issues. Currently the governance of the spreadsheets depends on the owners of the spreadsheets. One temptation is to create a structure that brings the division in line with all the structures of the larger data warehouse. However for this division, they have no reason to integrate with the warehouse and want to solve their productivity issue.
We then wonder about the appropriate structure to introduce some management to this process. Rather than creating a new process we looked at how the division deals with the cross-functional issues currently. We looked at the budgeting process and how the organization sets priorities. We found that most of these processes were relatively informal with the leader of the group determining the strategy and allocation of resources. Each group in the division competes for resources. We then looked at who was driving the process and who had the most to gain by improved productivity. Unfortunately the people who were currently creating these spreadsheets and their boss had the most to lose. The rest of the management group had the most to gain. We then looked who was supporting the project and their position in the organization. We then looked at the history of the management team and how they got things done.
After this assessment, we decided to ask for advice from the head of the division to determine the best way to proceed. We did not ask simply about data governance, but how to introduce the approach to the organization. We asked about what the best way to get agreement of definitions of specific things. We were careful not to use data warehousing terms but business terms.
We were surprised to find the division head understood the issue and was willing to take leadership on the issue. She appointed one of her key reports to work with people from the business and our project team to drive common definitions of terms. This was the beginning of the data governance process for this group.
We will continue to build on the structure created as issues arise. We will also modify the membership as our experience with the division and their management team increases. We will eventually develop a set of procedures but we must walk before we run. We must learn what works and what does not work in a specific organization.
This case is typical of many people who are beginning to realize the opportunities and challenges of data warehousing. The lesson is to introduce the disciplines of data governance in response to business issues as opposed to any technical requirement of the data warehouse.