Better Beats Bigger Data, Every Time.
3. Skepticism vs. Materiality. Building on the above, there will be skeptics. Skeptics will challenge if data can or does create business value for the organization. A healthy degree of skepticism is helpful. A "Data Athlete" does bring a healthy curiosity about what data suggests, and how it is captured, transformed and considered. There’s unhealthy skepticism, even if it is innocent in its nature.
A common example of where skeptics combat a data-driven culture is finding examples of incomplete records, or inaccuracies. Another example is illustrating the gap between two systems. These all, at face-value, seem terribly problematic. However, those “gotcha” moments need to be considered in terms of context ― a 2 percent discrepancy between the core financial system and the marketing database is extremely important in the financial database ― and most likely immaterial in the marketing database.
4. Complexity Can Undermine Results. In the first example, an ambition to squash “silos” of disconnected systems was the justification for “biggering” a new database approach. To be certain, there were prior experiences where “siloed” data created frustrations in the CIO’s capabilities he/she extended inside the organization. However, none of these were directly aligned with the CMO’s objective of improving messaging, response and sales.
Bigger data doesn’t just mean more of the same data types ― it often means adding more types of data. The complexity and design of the underlying schema or data model is directly correlated with the sheer number of data fields being captured. So capturing more and more adds to the challenge of making it economically viable to create value with it.
The same can be said for the complexity of the data fields themselves. While leveraging transaction data can be done through reasonably well-understood statistical methods and models, incorporating social signals is typically more challenging. Marketers frequently cite correlation of certain observable social behaviors, (a common example being “likes”) with buying behaviors when causation is what’s needed to discern the economic value and business impact of these types of data.