As sure as spring signals new fashions from Paris and Milan so too does it prompt the world of marketing to reveal its next “hottest thing.” This year some of the macro marketing types are absolutely breathless about the recently coined concept of “Big Data.”
Big Data proponents from places like IBM and the worlds of management and IT consultancy tell us that 90% of all data the world has ever known has been created within the last 24 months. Seemingly every twit, tweet
, idle thought, innocuous consumer measurement, rumor and other data artifact whose volume has exploded in the digital world is worthy of the same care destined for this season’s baby robins.
Never mind that the paradigms of marketing research, basic R&D, strategic planning, marketing strategy and such have long harnessed (what we might comparatively call) “small” data for effective marketing and general business use. Big Data, so the claim goes, is the necessary deep data mining of all the newfound descriptive and numeric information to unlock unknown meaningful statistical patterns and the business opportunities they represent. Somehow by simply spending more time, money and energy on analyzing all bits and bytes that document human behavior in the digital sphere our knowledge banks will be reformed to be smarter and more profitable.
But the operating premises of Big Data largely fly in the face of established research and scientific principles. For instance virtually every facet of human inquiry buoyed by statistical analyses is driven by systematic sampling approaches. Virtually nowhere in the name of analysis is it thought that in order to “know” about something that entire object must be scrutinized from bow to stern. Thus, a sample. (An obvious and highly controversial exception to this is the United States Census). It’s this principle that allows your doctor to correctly analyze your blood from a mere two-ounce vial or a pollster to accurately predict Presidential election outcomes from 1,200 voters or after talking to 500 mothers allow a market researcher to know a certain box of cereal will move market share at $3.29 but will fail if priced a dime higher. By appropriately assessing just a fraction of a whole, what’s happening can be accurately determined without measuring everything about what’s happening!
Or also consider that when deeply mining data it should be guided by some reasonable hunch, hypothesis or premise. Any data analyst must have some idea of what they are looking for, otherwise, how will they know when it’s found? But the Big Data analyst is not concerned with this. He is simply looking for something—anything—that seemingly represents a statistical corollary. That the relationship of the “connected dots” might defy any plausible explanation is secondary. If the analysis infers something is associated then 1) it is, and 2) it must be important otherwise it wouldn’t be associated in the first place!
This sort of circular thinking would never let an undergraduate escape from the most basic of research and statistic classes. And yet Big Data proponents are shifting into high gear to convince the worlds of marketing and management that guided by this philosophy significant new and larger investments in data storage, management and analysis are needed lest one’s competitor seizes the business advantage.
It is an idea that, at least for now, is simply too half-baked and runs the risk of leading the marketer who is not a researcher or statistician to chasing his proverbial tail in an endless data swamp. Heed what is suggested here. When encountering Big Data in the coming months do it carefully—with appropriate skepticism and consideration for the fundamentals of data measurement and analysis that have effectively worked for eons.
_____
John Durante is senior marketing associate for WordWrite Communications.
