In my last post, I suggested that there is a difference between data attributes used for unique identification and those used for attribution to facilitate customer segmentation and classification. An example of some attributes used for segmentation are those associated with location, such as home address or package delivery address. A customer’s home address can be combined with regional profiles of purchasing patterns of other individuals living nearby, and that can guide your company’s decisions of the types of products or promotions to offer customers of the same ilk.
In other words, attribution elements are those data elements whose values can be used for generally describing pools of individuals. These attributes contribute to describing a cluster or a segment, as well as those data elements that can drive the iterative analysis of customer clusters and segments. In general these demographic data elements are descriptive, and span the range of geographic, psychographic and historical data elements. Some more discrete examples include location of birth, number of cars owned, level of education attainment, net worth, home ownership, etc.
In some cases these data elements are inherent to the individual and do not change (such as location of birth), while others can change over time (such as the customer’s employer). Inherent data elements can be added to the core customer entity data model. However, you may want to manage multiple sets of attribution data elements (such as a permanent residence as well as a vacation residence) as well as track the history of more transient attribution relationships (such as maintaining the customer’s employment history). In turn, each of these attribution elements might represent its own entities (such as a location, organization or contact mechanism).
This suggests that one approach to organizing attribution data is to first determine whether any of those associated attributes are individual data elements or if they are entities in their own right. If they are individual data elements, they can be directly added to the customer entity model. Otherwise, the new entity concepts (such as “employer”) can be modeled independently. In turn, each unique customer can be conceptually linked to these attribution entities in ways that capture both the association and the duration of the association.
Next time: thoughts about relationships and hierarchies.
