Once, a furious father confronted a Target store manager for sending promotions to his teen daughter for baby clothes, cribs, and diapers. The store manager apologized for this, only to receive a call from the same father a few weeks later, saying that he had found out she was indeed pregnant!
Target knew this before she even told her parents!
My newsletter this week is inspired by a chapter from Eric Siegel’s book Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die. The chapter focuses on concerns about privacy and civil liberties that arise in data science.
While the Target story may be apocryphal, it is one of the most famous examples of using data science to make sensitive inferences about people.
Target used customer transaction data from loyalty cards, credit cards, and emails to identify patterns in their purchasing behavior. Purchases of certain items, such as unscented lotions, vitamins, or large amounts of cotton balls, were considered potential indicators of pregnancy.
Using machine learning, they created a predictive model that assigned each customer a “pregnancy prediction score.” Target then sent personalized coupons and offers for baby products to women who were predicted to be pregnant. The model like these can be scaled across millions of customers and continually refined based on new purchasing data.
A highly targeted marketing campaign can increase sales by reaching customers at a critical life stage when their buying habits are likely to change.
However, they can also raise privacy concerns if they reveal medically sensitive, unvolunteered data about customers they may not have wanted to disclose. The same concerns apply to predicting sexual orientation, race, health status, location, and an employee’s likelihood of leaving a job.
According to research by the University of Cambridge, even seemingly innocuous digital behavior such as Facebook Likes can lead to models predicting sensitive personal information such as race and ethnicity, religion, and political affiliation.
As marketers, we want highly targeted and personalized campaigns, and data scientists want to build as predictive models as possible. However, these goals need to be pursued with caution and ethics. With great power comes great responsibility.
That’s it! Thank you for reading, and see you next week.