How big companies like Target and Meta use Predictive Analytics

In talking to people outside of the field, there is a common misunderstanding that AI is limited to LLMs like ChatGPT. While Natural Language Processing is a big part of the Artificial Intelligence landscape, there are more subsections like machine learning and computer vision. Companies use AI for a multitude of reasons, some we are not aware of. So in this series we will explore how companies use AI and our data.

There was a news story about a father who burst into a Target upset because Target was sending his high school daughter pregnancy coupons. A few weeks later, he called into the same Target to apologize because his daughter was, in fact, pregnant. So how did Target know before him what was happening in his daughter's body? In the early 2000s, a statistician named Andrew Pole and his team explored customers who had signed up for a Target baby registry and looked at what they were buying months before. In this process, they were able to put together a list of 25 items, such as unscented lotion and certain vitamins, that seemed random but often were an indicator of pregnancy. This pregnancy prediction algorithm at Target was not only able to accurately predict if a woman was pregnant, but also approximate the due date.

While every company today does some form of predictive analytics, one company may know more about you than you know about yourself: Facebook, now Meta. Meta collects data from three places: what you do on its apps, what you do on other companies’ websites…