Given this definition, critical data elements will vary from time to time based on the selection criteria.
So what are some approaches that can be used to define critical data elements?
A common approach used to categorise and segment data is to divide it into domains – subsets of data that describe specific business areas such as Customer, Product, Employee or transaction.
This approach is commonly applied to master data management – allowing us to, for example, model and manage customer master data as distinct from other data domains.
However, two apparently contradictory challenges
By their very nature domains can be very large. If the customer domain contains hundreds of attributes not all of these are critical at any given time, or for any given problem. A domain based approach is too big
At the same, time domains restrict data to a particular subject area. Real business problems typically use data elements from multiple domains e.g. to calculate sales by customer I need product and customer data – data from two domains.
So domains are too big and, in many cases they are also too small.
Critical data elements are used to achieve specific business goals.
To identify these elements we need to look a t a business problem and how data supports it.
One approach is to look at key metrics that are used to measure success
Which data elements are used by this report?
What are the key performance indicators in the scorecard?
Which data is used to support a specific, published business policy?
Which data supports a specific regulatory requirement?
Which data is categorized as protected or sensitive in terms of legislation such as PoPIA or GDPR?
Which data is an input or output of a critical business process?
Which data must be present in order for us to operate?
These kind of questions produce a subset of data elements that can be regarded as critical within the context of a particular reporting requirement, regulatory initiative, or to achieve operational efficiency.