Database Reference
In-Depth Information
which enables all to take ownership and act upon valid results, thus minimizing territorial-
ity and what some consultants call the “not invented here” syndrome.
Let's face it. The introduction of a Data Science initiative to the enterprise is a
tacit acknowledgment that existing methods of analytics, research, and innovation have
fallen short. Any professionals associated with these legacy methods are going to feel up-
heaved and threatened. If management does not astutely handle the introduction of the
new paradigm, giving all a sense of ownership, it is very easy for organizational infighting
over data ownership, decision authority, and other issues to arise, costing time and money.
On the other hand, when shared ownership and collaboration is made a top priority, the
near-term result is far more likely to be a flowering of data-driven innovation, a rich Data
Science culture, and clearly defined progress into new, previously unexplored and highly
productive terrains of BI.
At the outset, the vision for a Data Science-driven model of decision making must be
clearly enunciated and shared with all relevant staff, making them all stakeholders in the
process. Once legacy professionals understand the prospects, and internalize the respect
with which they are being consulted, they are far more likely to “buy-in” to the model and
cooperate. (Note that the “buy-in” will prove essential. Although tasks can be commanded
and responsibilities assigned,
enthusiasm
and
passion
cannot be. And these last two attrib-
utes are essential for success in Data Science, just as they are in most other things.)
Territoriality aside, most leading Data Science practitioners believe that the Data
Science team should be located organizationally within whatever group “owns” the data,
whether this be the IT department, the marketing/sales department, or some other division
of the firm. Per Greenplum's Annika Jimenez, this centralized structure is important be-
cause “there is an essential collaboration which must exist between the Data Scientist and
all the other owners of the 'value chain' of operationalized predictive and machine-learning
models.”
One last note on this topic: To build an effective Data Science team with the range of
skills necessary, the enterprise will invariably need to budget for and hire new profession-
als to come on board, not to mention the acquisition of new data tools and technologies. At
the same time however, it is critical to devise and implement training programs for any and
all legacy staff whom the enterprise sees as becoming part of the new equation.
