User Not Found
| June 06, 2011
I know that many of my colleagues despair of me when I question the value of marketing based on passively collected, digital data. They either think I don’t get it, or I am being hopelessly old fashioned. So maybe I need to make my case more clearly.
It is not that I am anti-listening, anti-targeting or anti-social – media that is – it’s just that I believe there is far too much hype and hope surrounding the use of digital data. However, there is incredible value to be gained, but only if we approach it in the right way.
Let me try to explain the root cause of my concern. Every time we use the Internet on our mobile phones, we leave behind a trail of digital clues as to our needs, wants and interests. This data exhaust is vast and free, but it is far from exhaustive. What is missing is any insight into our underlying motivations or any understanding of the influence of context and cues, the triggers that cause us to behave in certain ways or make certain decisions.
Data mining will tell us what people did, but we can only infer why they did it. And all too often that inference will be wrong. The beliefs, expectations and objectives of the people charged with making sense of the data will shape their interpretation, just as much as the quality of the raw data.
But there is a solution to this problem. We can ask questions to clarify why people behave the way they do. Don’t take my word for it. Listen to this interview on “Here and Now” with Professor Alex “Sandy” Pentland of the Massachusetts Institute of Technology (MIT) Human Dynamics Laboratory.
In the course of the interview, we learn that mobile phone data can be used to predict flue epidemics, the onset of obesity and changes in voting behavior. How? Sandy answers:
We look at the cell phone data and we also pester people with questions.
It turns out that the predictive power of digital data really lies in its integration with active questioning. Sandy and colleagues essentially conduct surveys to gain insight into why people behave the way they do. They ask questions by phone. They ask people to complete diaries. They ask people to weigh themselves. This additional data gives them tremendous insight into what the patterns in the behavioral data really mean. It gives them what Sandy refers to as, “a god’s eye view of human behavior.”
If we want to make effective and efficient use of passively collected digital data for ad targeting or making sense of online word of mouth, then we need to be prepared to emulate Sandy and colleagues.
We need to dig deep to find out why people behave the way they do. Then, once we have established reliable and predictable patterns from the behavioral data, we can downplay the role of direct questioning to the odd confirmatory text or tweet.
So what do you think? Can we make sense of digital data without integrating it with active questioning? If so, how?