Power of Prediction?
How Analytics Companies Learn Your Secrets?
How to predict a woman pregnant or not?
In a February 2012 article for The New York Times titled “How Companies Learn Your Secrets,”
Charles Duhigg writes about a statistician named Andrew Pole hired by Target in 2002. Soon after he
started his new job, two employees in the marketing department asked Pole a strange question: “If we
wanted to figure out if a customer is pregnant, even if she didn’t want us to know, can you do that?”
Intrigued and up to the challenge, Pole said yes and began work on a “pregnancy-prediction model.”
He looked at products that potentially pregnant women bought, like lotions and vitamins, as well as
products that they didn’t buy. A woman four-months’ pregnant is probably not going to purchase baby
food and diapers and store them for five months.
The marketing folks weren’t just hazing Pole or giving him busy work. They had a very real
business reason for approaching him. Ultimately, they used Pole’s model to send custom coupons to
women that Target believed to be pregnant—or, more accurately, women who were likely to be
pregnant based upon Target’s data.
Don’t yawn yet. About a year later, the story gets really interesting. A man angrily walked into a
Target outside Minneapolis, Minnesota. He demanded to see the manager. He was clutching a mailer
that had been sent to his daughter. From Duhigg’s article:
“My daughter got this in the mail!” he said. “She’s still in high school, and you’re sending her
coupons for baby clothes and cribs? Are you trying to encourage her to get pregnant?”
It turns out that the manager didn’t know what the customer was talking about. (Retail employees are
often unaware of corporate marketing efforts.) He took a look at the mailer. The father wasn’t lying;
the mailer was in fact addressed to his daughter. It contained advertisements for maternity clothing,
nursery furniture, and pictures of smiling babies. The manager promptly apologized to his customer.
A few days later, he called the man at home to apologize again. On the phone, though, the father was
somewhat humbled.
“I had a talk with my daughter,” he said. “It turns out there’s been some activities in my house I
haven’t been completely aware of. She’s due in August. I owe you an apology.”
In a nutshell, Target knew that a teenage girl was pregnant before her own father did . Pole’s
pregnancy prediction model did not require individual names, and Pole certainly did not know the
identity of the woman in Duhigg’s story. None of that mattered. His model didn’t need that
information. Target’s own internal sales data was plenty.
In other words, Target could predict if a seventeen-year-old girl was pregnant (while her own
father didn’t know) way back in 2002.
Leave a Reply