Need for an Enterprise-level Data Dictionary
• A manner of defining information and cataloging how it is assembled in a database, how fields are defined and how data relates to each other within the database.
• A crucial component to insure your business makes decisions based on good data.
If you’re going to become a data-driven organization, you better make sure you can rely on the data.
To make data-driven decisions, it’s imperative that organizational leaders speak the same language. Information behind decisions needs to allow apples to apples comparisons, not apples to oranges mistakes. Thus, there is the need for an enterprise-level data dictionary.
Within most databases exists a read-only file known as a data dictionary. It defines for a programmer how fields and views are established; who has access to the data, what types of information are recorded in fields, etc. These go by many names and are, without a doubt, necessary and valuable, but today we’re going to focus on the enterprise data dictionary.
In complex industries a business relies on several different databases, often administered by more than one department under different management. In such environments it’s a virtual certainty that, over time, data will be classified in different ways, leading to inevitable problems with pulling useful data alone, let alone high quality data to drive the business.
Proper use of a data dictionary simply ensures everyone is speaking the same language. It means when a report is generated, it will produce the needed facts
Take, for example, a situation several years ago at my business process outsourcing company. I was reviewing a client scoring report which assigned a number to a consumer’s likelihood to pay a debt. I noticed a big number in a category of consumers within a field simply marked “PP.” My heart beat faster as I thought I found a magic bullet–simply focus efforts on these consumers and watch revenues shoot through the roof! My moment of elation was short-lived, as the analyst explained the “PP” meant “promise to pay”—these consumers had already committed to pay. No wonder they scored so high.
Opportunities to generate bad data are many. An operations manager, an analyst and a programmer may all have slightly different thoughts on what a given term means. In another example from the call center space, a Right Party Contact may mean an agent has spoken to the intended party, but to another person it could also mean the agent simply left a voice mail. A Right Party Contact report generated to simply evaluate performance on completed calls could be skewed if voice mail contacts are included, possibly leading to bad decisions, bad strategy and inefficiency.
In addition to documenting common meanings for elements in a system, a data dictionary establishes features of a system and helps find errors. Creation of an enterprise data dictionary is never as easy as it seems. Various people across an organization may have very specific reasons why they define data elements in one way or another, even if it doesn’t match the department across the hall. In order to assemble a dictionary, a determined individual with top-level authority must drive the process.
Proper use of a data dictionary simply ensures everyone is speaking the same language. It means when a report is generated, it will produce the needed facts. It’s a building block for quality data. Once you add in an outside vendor, the importance of well-defined data only increases. Integration becomes much simpler if each organization can easily define data for each other.
Dynamic organizations live on data-driven performance. Starting off from a solid, defined base makes it easier to reach the outcomes you desire.