If a key concept of customer centricity is understanding relationship networks and any individual’s sphere of influence, it is critical that the organization of the data incorporate two different aspects of these networks.
The first is the concept of a relationship, which can bind a customer to some other entity or two customers together. We have already begun to consider how one customer can be linked to other entities, such as an employer, a place of residence or a contact mechanism. However, there is a qualitative difference between the knowledge that can be derived from those types of relationships and the relationships between and among customers themselves.
Because the interaction of social networks that effectively self-organize within a customer community shed light on aspects of a customer’s profile in terms of communal likes and dislikes within a subgroup, recommendation engines often rely on connection networks to help refine suggestions and promotions. Yet more complex networks are difficult to represent using a standard relational model, so different paradigms (such as graph database systems) may be better suited to organizing customer connectivity networks.
Instead of examining links between peers, the second aspect of connectivity looks at hierarchical organization. One common example is the concept of a “household,” which links a group of individuals together based on similar characteristics and under the guidance or control of a hierarchy of individuals. One manifestation is a “traditional family,” in which one or two adults oversee a collection of children living in the same location. Another manifestation might be a special interest group of individuals sharing some common interests (such as a motorcycle club) led by an elected set of officials.
The organization of hierarchy data is similar to that of standard relationships plus the addition of a “direction” or “inclusion” within the model. Members of that special interest group are linked together in a way that documents the interest, and individuals with leadership roles need to be organized by their level of influence (e.g. the president of the club can be seen above the vice president in the conceptual hierarchy). Again, the relational model may not work as well as a directed graph model might, since the latter might be able to capture both the connections and the hierarchical levels.
