Let’s take a closer look at the two most recently defined analytical roles of the data scientist and value role. Back in the eighties, the functions of these roles existed for many of the leading direct marketing companies. In direct marketing, there was the key individual or modeling analyst, the data scientist by today’s terms, building a direct mail response model. Yet, the use of this model within the business went nowhere without the involvement of the marketer, commonly referred to as the value architect in today’s terms. This required a “marriage” of the two disciplines in the sense that the data scientist needed to have some “marketing” knowledge while the value architect needed to have some “modeling” knowledge. Yet, this so-called cross-over knowledge would of course be irrelevant without the deep domain knowledge that would be paramount within each role.
To best understand this, it is useful to observe what this practically meant within the direct marketing world since this is the early history of how these two roles worked in building a solution. In the early days of analytics, the discipline was almost exclusively used in a tactical manner. Overall marketing strategy was determined by the key leaders in the marketing area. Specific tactics such as marketing campaigns were then designed and created in order to execute on this strategy. But to ensure that these campaigns were as profitable as possible, analytics were conducted in order to optimize efficiences. This was manifested in such performance indicators as reduced cost per order or reduced cost per customer saved
In this very tactical exercise, these two roles are quite distinct yet complementary in achieving their objectives. The marketer in most cases has already developed the campaign or initiative and of course its accompanying budget. Optimizing budgets was the primary driver of analytics in the early days. Based on some key learning, the analytics might have been to derive enough insights to develop some basic business rules that would reduce costs. For example in acquiring new customers, it was the identification of key lists or particular Stats Can demographic segments . At this level, the marketer was simply trying to understand the key characteristics that differentiates new customers from the general population.
Overtime the cost reduction from this learning would erode implying that we need a more advanced analytical approach beyond just basic business rules. Consideration would be given towards the development of a model and hence a meeting with the data scientist. The value architect or the marketer would explain the objective of the campaign and in almost all cases define the required modeling tool. Listening carefully to the marketer’s campaign objectives , the modeling analyst would then think about the data. For example, what data sources do I have in order to build an acquisition or retention model. Can I create the necessary information environment that will deliver meaningful inputs into a model Do I have the data to create the objective function or target variable . More importantly, do I have the software and an understanding of the software that allows me to create both the “information” environment as well as to leverage the various statistical routines.
The data scientist would then conduct the “analytics “ in delivering the required solution. Upon building the solution, the data scientist would arrange a meeting with the value architect/ marketer in order to demonstrate the solution in terms of its “business” impact. In simple business terms, this would comprise the identification of key characteristics of the solution as well as identifying the actual dollar benefit of implementing this solution. At this point, both the data scientist and the value architect would then discuss future initiatives of how this solution might be used. Once an initiative was identified, a marketing matrix would be designed by the value architect but in conjunction with the data scientist. The matrix would be designed to meet both business as well as learning objectives. The data scientist’s role would be to determine if they had the “right” data to build this matrix. Once agreement was obtained, the initiative would then be executed with the data scientist being mandated to construct the specific records that would comprise each cell within the marketing matrix.
After execution, measurement of the initiative is the next natural step in this process. Here again, the marketer and the modeling analyst confirm the business and the learning objectives of the campaign. The modeler will deliver the results vis a vis reporting with the marketer interpreting and deriving meaningful insights from the results.
Analytics in today’s Big Data World is a much more collaborative process between the two roles. This is best manifested by how marketing and in fact business strategies are much more data-driven. As a result, it is quite common today for many organizations to commence their foray into analyt ics with a data discovery. The data discovery represents a process that allows us to define a data strategy which clearly addresses the business needs of the organization. At the same time, this collaborative approach within the data discovery helps to outline the necessary tactics that are required for execution of the strategy.
As the world becomes consumed by topics and themes on Big Data, the roles of the data scientist and the value architect become ever more paramount. With social media and text data being readily accessible, the ability to understand context and what business insights might be meaningful will remain the purview of the value architect. Meanwhile, the ability to “work” the data in order to create the information necessary for these insights is the responsibility of the data scientist. These roles will continue to grow in importance within organizations as Big Data and Analytics become more entrenched within a company’s corporate culture. This will result in more collaboration and in a sense a more symbiotic relationship between the value architect and data scientist . Without this type of relationship, companies will be at a disadvantage in this new age of Information and Big Data.
image: data scientist/shutterstock