Sunday, February 19, 2012

Insights Come from Clues

What can we learn from product purchases?

In one of the most interesting cases of using data to create insights, the NYT reported and Forbes followed up on how Target figured out whether someone was pregnant. Using changes in product purchases the analytics team estimated the likelihood that a guest was pregnant and specifically in which trimester.

The marketing reason for such an analysis: identifying someone who is about to have a baby can be the basis for forming a long-term relationship. The business reason for doing so: If you can do that before other companies, the odds are much more in your favor of creating the habit that drives loyalty.

Those articles were shared by the boatload with lots of discussion.  The response was usually somewhere between cool and creepy; particularly when you start thinking about "selling the pregnancy score."  (Pic from that article.)

I'll leave the debate as to whether this goes beyond the boundaries of leveraging personal information to the ethicists.  From an analytic perspective, this a good example of how it should work.
  1. Identify a question that hasn't been adequately answered before.
  2. Attempt to infer the existence of a market based on what little we know.
  3. Create, execute and refine a plan based on the clues.    
No report could have produced this type of analysis. 

No comments: