Great article in The New York Times this past weekend, Can Social Media Sell Soap?, sheds some light on the war in marketing between the “humanists” of Don Draper’s world and the “quants” of the digital marketing revolution.
The essence of the story: data-driven marketing is incredibly promising, but it is not all-powerful. Those who polarize themselves into one camp or the other, rather than seeking the best of both worlds, find themselves on precarious footing.
Here is an excerpt that I found particularly compelling (emphasis added is my own):
While the rise of search battered the humanists, it also laid a trap that the quants are falling into now. It led to the belief that with enough data, all of advertising could turn into quantifiable science.
This came with a punishing downside. It banished faith from the advertising equation. For generations, Mad Men had thrived on widespread trust that their jingles and slogans altered consumers’ behavior. Thankfully for them, there was little data to prove them wrong.
But in an industry run remorselessly by numbers, the expectations have flipped. Advertising companies now face pressure to deliver statistical evidence of their success. When they come up short, offering anecdotes in place of numbers, the markets punish them. Faith has given way to doubt.
This leads to exasperation, because in a server farm packed with social data, it’s hard to know what to count. What’s the value of a Facebook “like” or a Twitter follower? What do you measure to find out?
In this way, marketing resembles other hot spots of data research, including brain science and genomics. In each one, scientists are combing through petabytes of data, trying to discern whether certain genes or groups of neurons cause something or simply correlate with it. It’s not clear, because these are immensely complex systems with millions of variables — much like our social networks.
Even as researchers swim in data that previous generations would have swooned over, they struggle to answer crucial questions regarding cause and effect. What action can I take to get the response I want?
