Surveillance
Your Shopping Cart Knows You're Pregnant. You Just Don't Know It Yet.
In 2010 a Target statistician named Andrew Pole stood up at a data analytics conference and presented something that should have made headlines immediately.
It didn't.
The presentation had a dry technical title. The audience was mostly data professionals. Nobody outside that room paid much attention.
Two years later a New York Times journalist named Charles Duhigg wrote about it.
The world has never looked at a shopping cart the same way since.
What Target Actually Built
Target's analytics team had identified a problem.
New parents are one of the most valuable customer segments in retail. When a baby arrives, spending patterns change dramatically and they change fast. Diapers. Formula. Furniture. Clothes. The opportunity to capture a customer for life is enormous.
But by the time most parents announced a pregnancy, it was already too late. They had developed shopping habits with other stores. The loyalty window had closed before Target ever had a chance to open it.
So Target decided to find out before anyone announced anything.
Pole and his team went through Target's purchase database and cross-referenced the buying patterns of women who had previously signed up for baby registries. They worked backward. What did these women buy before they told anyone they were pregnant?
The patterns that emerged were not obvious.
They identified approximately 25 products that, when purchased together or in a certain sequence, reliably indicated a customer was pregnant. Unscented lotion bought in larger quantities than usual. Calcium, magnesium, and zinc supplements. Cotton balls. Hand sanitizer. Large bags. Certain combinations of vitamins and toiletries.
None of those products scream pregnancy on their own.
But together, cross-referenced against thousands of other data points, they painted a picture that was remarkably consistent.
Target's system assigned every female shopper a pregnancy prediction score. Not just whether she was likely pregnant, but how far along she probably was. The algorithm could estimate a due date within a narrow window. Which meant Target could time its marketing to arrive at exactly the right stage of the pregnancy.
First trimester. Second trimester. Right before the due date.
All calculated from what went into a shopping cart.
The Part That Got Buried
The story that spread around the world focused on a specific incident involving a father who received a coupon booklet addressed to his teenage daughter. The booklet contained coupons for cribs, maternity wear, and baby clothes. He stormed into a Target store furious. The manager had no idea why she had received it. He apologized. Called back a few days later to apologize again.
During that second call the father's tone had shifted.
He had learned his daughter was pregnant. She hadn't told him yet.
It is worth noting that some researchers have questioned whether that specific incident was actually the result of Target's algorithm or simply a coincidence from a broad mailing. That debate is fair.
But here is what is not in dispute.
Target built and deployed a real pregnancy prediction system. It used real purchase data. It assigned real scores to real customers. And when customers started to notice the targeting and feel uncomfortable, Target made a calculated decision.
They did not stop.
They hid it.
Target's marketing team began deliberately mixing pregnancy-targeted coupons in with completely unrelated products. Lawn mowers. Dish soap. Random household items. The goal was to make the targeting feel like a coincidence rather than what it actually was.
They kept profiling. They just made it harder to notice.
That decision reveals more about how these companies think than any single anecdote ever could.
The concern was not that the targeting felt invasive. The concern was that customers were becoming aware of it.
The solution was concealment. Not restraint. Concealment.
It Goes Far Beyond Pregnancy
The Target story became famous because pregnancy feels intimate. It touches something personal in a way that most retail behavior does not.
But pregnancy was never the only thing these systems were designed to predict.
Research published in academic literature has documented how purchase data and behavioral patterns are used to predict an enormous range of life events. Getting married. Getting divorced. Moving to a new city. Changing jobs. Running into financial trouble. Having a child. Even predicting the likelihood of a customer's death within a given time window.
One documented case involves a major credit card company that used purchase data to predict divorce among its cardholders, allowing it to get ahead of potential credit problems before they materialized. The company was not predicting divorce out of concern for its customers. It was predicting divorce to protect its own financial exposure.
A Canadian retailer was found to be using purchase patterns to assess credit risk. Customers who bought carbon monoxide detectors, premium bird seed, and felt pads for their chair legs almost never missed a payment. Customers who bought cheap motor oil and frequented certain types of establishments were flagged as higher risk.
Your shopping habits are being read as a behavioral biography. Every purchase is a data point. Every pattern is an inference. And every inference is being used by someone to make a decision about you.
The Data Broker Economy You Have Never Seen
Most people have a rough idea that companies collect data on them. What most people do not understand is the scale of the industry that has been built around buying and selling that data.
As of 2024 the global data broker market was valued at over 300 billion dollars. That number is projected to nearly double by the early 2030s. There are estimated to be as many as 5,000 data brokers operating globally, companies whose entire business model is collecting personal information, analyzing it, and licensing it to other companies.
These companies are not household names. They operate entirely in the background.
But they know your name. Your address. Your income bracket. Your purchase history. Your location patterns. Your health conditions inferred from what you buy. Your political leanings inferred from what you read. Your relationship status. Your employment situation. Your credit behavior. Your social connections.
All of it packaged, scored, and sold to whoever is willing to pay for it.
Retailers use it to target you. Insurance companies use it to price you. Banks use it to evaluate you. Employers use it to screen you. Law enforcement agencies use it to track you, often purchasing the data directly rather than obtaining a warrant.
The data that feeds this entire ecosystem has to start somewhere.
It starts with your behavior.
This Was 2010. Think About What Exists Today.
Target's pregnancy prediction score was built from loyalty card data and credit card purchase history.
That was considered sophisticated over a decade ago.
Today every app on your smartphone is collecting data that makes Target's system look like a sticky note.
Your location history updated in real time. Your search patterns across every query you have ever typed. Your browsing behavior across every site you have visited. Your social connections and how often you interact with each of them. Your sleep data from your fitness tracker. Your heart rate and activity levels. Your spending habits down to the individual transaction. Your emotional state inferred from what content you consume and when you stop consuming it. Your political leanings inferred from the publications you read. Your health conditions inferred from the products you search for and the symptoms you look up.
All of it being fed into systems far more powerful than anything that existed in 2010.
And unlike a coupon booklet that arrives in the mail, you never see the output.
There is no moment that tips you off.
The targeting happens invisibly. In your ad feed. In the content served to you. In the prices you are quoted online. In the job listings you are shown or not shown. In the insurance rates you receive. In the loan terms you are offered. In the opportunities that appear in front of you and the ones that quietly never do.
All of it shaped by data you generated without knowing it was being collected, fed into models you will never see, producing decisions that affect your life in ways you will never fully trace.
The Question Nobody Is Asking
Most conversations about data privacy focus on whether companies should be allowed to collect this data in the first place.
That is the wrong question. Or at least it is too late a question.
The data is already being collected. The industry is already worth hundreds of billions. The infrastructure is already built. Regulatory efforts have slowed some practices in some places, but they have not stopped the underlying machine.
The more practical question is this.
How much of your data is actually being generated?
Because the data broker ecosystem depends entirely on data existing in the first place. It cannot profile behavior that was never logged. It cannot sell location history that was never recorded. It cannot build a pregnancy prediction score from purchases that were never tracked.
Target's algorithm worked because its customers handed over their data in exchange for coupons and the convenience of a loyalty card. The same exchange is happening on your smartphone right now, at a scale that makes Target's operation look modest.
Every app you install is a potential data collection point. Every permission you grant is another window into your behavior. Every search, every purchase, every location ping, every message is a data point being added to a profile that shapes how you are seen, what you are charged, and what you are shown.
What You Can Do About It
Start with your smartphone. It is the hub from which most of this data flows.
Audit every app that has microphone, camera, and location permissions. Remove access from any app that has no clear reason to need it.
Review the terms and conditions of apps before installing them. Research has found that 91% of people agree to terms without reading them. That agreement is what grants legal permission for the data collection that follows.
Use a browser that does not track your searches. Consider a VPN for your regular browsing activity.
Be deliberate about loyalty programs. Every swipe of a loyalty card is a data point added to a profile that will outlive your membership.
And consider the operating system your smartphone runs on.
The Ghostphone runs GrapheneOS, which removes Google entirely from the operating system. Apps are sandboxed, meaning they cannot share data with each other or access sensors they have not been explicitly granted permission to use. Location tracking is off by default. Microphone and camera access require your explicit approval every single time.
The data that feeds these systems has to start somewhere.
Target's algorithm needed a shopping cart.
Today's systems need your smartphone.
On a Ghostphone, they have nowhere to start.
The Ghost Phone
No data to collect. No profile to build. No algorithm to feed.
Ed Warren is a Digital Privacy Consultant with over 15 years of experience in the surveillance and data security industry.
