My articles from 2021

How does Target know What You Want Before You Do It?

note(involves pregnant women as examples sorry actually read this at a presentation)

Andrew Pole has just started working as a data expert for the company Target when a few colleagues from the marketing department stopped at his desk one day and asked the kind of question that Pole was born to answer: Can your machines figure out which customers are pregnant, even if they don't want us to know?

Pole has been a statistician his entire life, and it mostly revolves around using data to understand people he had grown up in a small town in North Dakota. While his friends were doing other stuff, he was gathering huge amounts of data to do experiments. For example, he threw a party and asked people their favorite jokes, which he listed and tried to see which one was the best one-liner. Also, he tried to find out the exact amount of beer he had to drink to build up the confidence to talk to girls at parties but not enough to make a fool of himself (that particular experiment never seemed to come out right).

But those experiments were just child's play, so when he heard about Hallmark, the greeting card company, was hiring statisticians from North Dakota towns, he applied and got in. It was like heaven for him, he was using data to determine if pictures of pandas sold more birthday cards or elephants. Or determining "What happens at grandma's stays at grandma's" was funnier in blue ink or red ink. But when he heard in 2002 that Target was looking for statisticians in Kansas City, he made the jump that enabled him to study one of the most complicated things —the American shopper.

Pregnant people who shop at Target would hand over their due dates and other valuable information, and Target will use them to seek pregnant customers. Now, why does Target want them to shop at Target? "Because if they start shopping here for diapers and infant clothes they will start buying everything else from us," Pole said. because it's Convenient and convenience matters the most to new parents.

For a time, Target would never hire statisticians like Andrew pole. They were hiring psychologists, which was how they used to predict customers' behavior. And the psychologists would use methods to try to make people spend more. Some are still used today by companies such as Walmart and Home Depot.

Target is now using something called the guest ID number, which keeps track of everything customers buy at Target. The problem with this is that the data needs a statistician to make sense of. For example, two people buying one bottle of juice each might seem like the same thing for regular people. But to a statistician one is buying juice for her kids, which means she may appreciate a Thomas the Tank Engine DVD. And the other person drinks juice after a run, which is why he may react to a discount on sneakers. After using their understanding of habits to predict customer behavior after a while, the program was made to predict who was pregnant. How? We are about to get into that.

Andreasen, who interviewed many people on the telephone about why they changed the things they bought. That's because their shopping habits were changed, and a lot of people he interviewed changed brands of products they bought like laundry detergent. His question was why have they shifted their habits. It turns out, it's often because they've just gone through a major life event. For example, what's most likely the reason why somebody bought silverware, pans, and frozen dinners? They probably moved or got divorced recently.

But the thing that changes most people's shopping habits the most is the arrival of a new child. Some studies show that new parents would spend 6,800 US dollars before a child's first birthday. So that was the reason why Target wanted to build the program that predicts who's pregnant.

Target also wanted to use the customer's data to predict if they're pregnant, but before they launched a national campaign, an employee asked how women are going to react to Target knowing their reproductive status. So they took a step back and considered their options. They thought it would be wise to do some polls before launching a worldwide campaign. Pole said they needed to be allowed to know what Target wanted to do with the data, and some women reacted to it badly. But why do companies want to predict who is pregnant? Because, for outlet companies, pregnant women are gold mines. When they are pregnant they are already on everybody's radar. And how did Target capture them? I'll explain it using another story.

In the early 2000s, a program called Hit Song Science was created, and it would predict which songs were going to be hit. A song called Hey Ya by Outkast was released that year and the program predicted that it would be a monster hit. It was a combination of all musical categories but when they put it on the radio, almost everyone switched the channel, why? The program had a problem, it can predict some behaviors, but other behaviors it predicts have not been embellished in the mind of individuals, so it was not that accurate. But radio DJs found a way to solve the problem. DJs would sneakily slip the theme song from Titanic right before Hey Ya, and that was the key to making Hey Ya a hit. If you listen to a pop radio you expect to hear pop songs, so they sandwiched Hey Ya between two sticky songs (a sticky song was what DJs call a song that a bunch of people liked).

Target used the same technique to grab pregnant customers. They put some products women usually wouldn't buy between baby products like toys and clothes and women products, and they named the area Mom and Baby. The sales from that department exploded. From 2002 to 2009, their revenue grew from 44 billion US dollars to 60 billion US dollars.

It's amazing that you can use all this data to do stuff like this. it is like mathematical mind reading. But sometimes it can get on people's nerves. This is the end of my article. Thank you for listening. by Arthur