In the data world, metadata is the ugly stepsister. Data is hot. Money and resources for data centric projects are flowing freely with hardware and software purchases. Data has moved to the top of the list of assets an enterprise possesses. The same cannot be said for metadata. Selling the value of metadata is difficult with most implementations being at the home-grown, grass roots level.
In late April I attended Enterprise Data World and a metadata panel discussion. Panelists reinforced the value of metadata and the issues that surround its implementation. I jotted down the following memorable quote by Danette McGillivray, one of the panelists.
“Creepy that they know something about me but even creepier that they know something about me that is not right.” Danette McGillivray MD Panel
Her quote underscored the effects of non-governance or poor governance of metadata. As data grows exponentially, the practice of data governance has become a necessity to assure the quality of data. Enterprises have placed high value on data as a competitive tool. It is critical that the integrity of this asset is managed to maintain its value.
Metadata is often overlooked as a data asset. It is often light or missing. Other times it is inaccurate or incomplete. Ask yourself, “Would I use a dictionary if I did not trust the definitions it contains?” The same logic applies to data. If definitions are of poor quality and cannot be trusted, you would shy away from using the data.
As corporate data grows in size so does the use of this data. Business intelligence and analytics has clearly demonstrated the value and shown good results. Analytics uncovers the deepest hidden details of the business and exposes patterns and trends that give that advantage. Many an executive are eager to tout the value data brings to their enterprise.
Poor, aka creepy, metadata is more likely to show up in areas beyond these analytics. They are often created by business data experts who are intimate with the data. They are light or non-users of metadata. They don’t need it. They know exactly how this data looks, acts and feels. Move beyond this inner circle of experts and problems arise. More junior and casual data consumers need metadata to succeed.
Metadata that casual data consumers encounter can result in the creepy results that Danette referenced. They rely on the metadata definitions, context, formulas, and relationships to make choices in reports and data analysis. If it is inaccurate or missing, the assumptions they make can be disastrous.
How do we take the creepy out of metadata?
- Implementing a data quality program for metadata is critical to assure its integrity and accuracy.
- Make sure data model reviews are frequent and include reviews of the underlying metadata.
- Get an enterprise perspective with a cross functional team in data requirements gathering.
- Sell the value of metadata to project managers and IT and business management.
- Make sure your metadata is visible and easy to reference and navigate.
- Train your enterprise on the use of metadata just as you would for deployed data.
Metadata is a complex creature. It is that complexity that makes it difficult to sell to the enterprise. It is hard to find an all-encompassing solution to deploying metadata. The best data professionals can do is look to the items listed above and work on implementing them. It can go a long way in taking the creepiness away.
I am presenting at Data Modeling Zone 2014 in Portland, Oregon. I hope that you will join me in my sessions: Is your data model a work of art? and Relationship versatility and the data modeler #DMzone