The Pregnant Pause
How the predictive power of big data got a little too revealing.
2 min read · from UNINTENDED by Mayank Mehta
Target's data science team, in the early 2010s, made a discovery that thrilled the marketing department. By analyzing purchasing patterns, they could identify with high accuracy when a customer was pregnant, often before the customer had told anyone. The signals were subtle but consistent. Pregnant women tended to buy unscented lotion around the beginning of the second trimester. They switched to supplements like calcium, magnesium, and zinc. They purchased larger quantities of cotton balls and hand sanitizer.
The algorithm worked. Target began sending tailored coupons and advertisements to customers it identified as expectant mothers: deals on maternity clothes, cribs, diapers, and baby formula. The company knew that if it could capture a customer during pregnancy, it could shape their shopping habits for years. New parents are among the most brand-loyal consumers in retail.
Then a man walked into a Target store outside Minneapolis and asked to see the manager.
He was holding a mailer full of baby product coupons that had been sent to his teenage daughter. He wanted to know why Target was encouraging his high school daughter to get pregnant. The store manager apologized, confused and embarrassed. A few days later, the manager called the father to apologize again.
The father's tone had changed. He had spoken to his daughter. She was, in fact, pregnant. Target's algorithm had identified the pregnancy before her own father knew.
The story, reported by the New York Times, became one of the defining anecdotes of the big data era. It illustrated not just the power of predictive analytics but the discomfort that power creates when it crosses from helpful into invasive. Target hadn't done anything illegal. It hadn't accessed medical records or hacked anyone's phone. It had simply observed what a customer was buying and drawn a conclusion that turned out to be correct.
In the aftermath, Target adjusted its approach. Rather than sending mailers full of baby products, it began mixing baby coupons in with unrelated offers, lawn mower ads next to diaper coupons, so that the targeting would be less obvious. The algorithm remained. The precision remained. The company just learned to be quieter about it.