Case Study: Target’s Predictive Advertising Strategy

In 2012, a father walked into a Target store near Minneapolis, coupons in hand, furious that the retailer had mailed his teenage daughter advertisements for cribs and maternity clothes. He accused the store of promoting teen pregnancy. Weeks later, he called back to apologise. His daughter was, in fact, pregnant. Target had figured it out before he did. This moment became the most discussed case study in modern advertising, a story that sits at the crossroads of consumer psychology, data science, and ethics. It is also a cautionary tale for every marketer working in today’s algorithm-driven world.

The business problem behind the algorithm 🔗

Retailers have long understood that shopping habits are sticky. Most people buy the same toothpaste, soap, and cereal for years without thinking twice. Breaking those habits is extraordinarily difficult, which is why marketers have always hunted for life-event triggers, moments when routines collapse and consumers are open to forming new loyalties.

Pregnancy is the holy grail of these triggers. As Charles Duhigg reported in his now-famous 2012 New York Times Magazine piece, the period around childbirth is one of those rare windows when shopping patterns and brand loyalties are up for grabs as parents are exhausted and overwhelmed . A retailer that captures a customer during this window often keeps them for years, selling everything from diapers and formula to furniture and electronics.

The challenge for Target was timing. Birth records in the United States are public, which means that the moment a baby arrives, new parents are bombarded with promotions from every competing retailer. To get ahead of the competition, Target needed to identify pregnant customers before they gave birth, ideally during the second trimester, when most expectant mothers begin buying baby-related items.

Andrew Pole and the pregnancy prediction score 🔗

In 2002, Target hired a young statistician named Andrew Pole. His job, as described in the New York Times account archived by the University of Notre Dame, was to figure out how to use data to answer a provocative marketing question: could Target identify pregnant customers even if they had not told anyone?

Pole turned to an internal data infrastructure that most shoppers had never thought about. Every Target customer is assigned a unique Guest ID number that links credit card usage, coupons, surveys, emails, help-line calls, and website visits to a single shopping profile . This meant Pole could examine the buying behaviour of women who had voluntarily signed up for Target’s baby registry and work backwards through their purchase histories.

The 25-product pattern 🔗

Pole and his team discovered that pregnancy was surprisingly predictable through seemingly mundane purchases. According to the Harvard Business School Digital Initiative case study, women on the baby registry tended to buy larger quantities of unscented lotion around the beginning of their second trimester. A few weeks later, the same women often started purchasing supplements such as calcium, magnesium, and zinc. Other signals included cotton balls, hand sanitisers, washcloths, and extra-large bags of cotton swabs.

From these observations, Pole built a model that analysed roughly 25 product categories to assign each shopper a pregnancy prediction score. The algorithm could even estimate a customer’s approximate due date within a narrow window. That allowed Target to time coupon mailers to specific stages of pregnancy, from prenatal vitamins in the early months to diapers and baby clothes closer to delivery.

Consumer psychology at the core 🔗

It is tempting to see this as a story about statistics, but it is really a story about consumer psychology. Pole’s team was not just predicting pregnancy; they were exploiting a well-documented principle from behavioural science: habits are easiest to disrupt during major life transitions.

The habit loop and marketing windows 🔗

Duhigg’s broader work on The Power of Habit established that routines consist of cues, behaviours, and rewards. Most advertising fails because it tries to disrupt already-formed habits. But when a person’s identity changes, say, from working professional to new parent, the underlying cues change too. Suddenly, morning coffee gets replaced by midnight feeding, and the grocery list doubles in length. Retailers who show up during this reshuffling have a far higher chance of becoming the new default.

The psychology of being watched 🔗

There was also a darker psychological dimension. When Target began sending explicit baby-product mailers to women it had flagged as pregnant, some customers felt genuinely unsettled, as though an invisible observer was tracking their bodies. To soften this effect, Target deliberately mixed baby-related coupons with unrelated offers like wine glasses and lawn mowers to make the targeting feel less intrusive . The company realised that the perception of surveillance was as damaging as the surveillance itself.

The incident that made the story famous 🔗

Target’s predictive system might have remained an internal success story if not for the incident that turned it into a global case study. As academic commentators at Cambridge have recounted, a father in Minneapolis stormed into a store with a booklet of baby coupons addressed to his high-school-aged daughter, demanding to know why Target was encouraging teen pregnancy. The manager apologised. When the manager followed up days later, the father admitted, with some embarrassment, that his daughter was indeed pregnant. The retailer had found out before the family.

The anecdote exploded across media once Duhigg published it. It became, almost overnight, the textbook example of how far algorithmic prediction could intrude on private life. Even critics have pointed out that the specific mailer may not have been solely driven by the pregnancy score, but the underlying capability was real and the public reaction was telling.

The commercial pay-off 🔗

Financially, Target’s bet on predictive advertising was a blockbuster. According to the Harvard Digital Initiative analysis, the company’s revenue grew from roughly $40 billion in 2002, when the analytics department was founded, to nearly $72 billion by 2017. While not all of that growth came from the pregnancy model, analysts widely credit Target’s data-driven personalisation as a major contributor to its climb during that period.

The broader lesson for marketers was clear. Personalised, data-driven campaigns consistently outperformed mass marketing in terms of conversion, retention, and customer lifetime value. Predictive analytics moved from a niche curiosity to a core function inside every major retailer, bank, and consumer brand.

The ethical and legal reckoning 🔗

The Target case is now studied not just for its cleverness but for the questions it forces us to ask about consent, dignity, and the line between useful personalisation and manipulation.

Inferred data and the consent problem 🔗

A key insight is that the women flagged by Target had never explicitly shared their pregnancy status. The system inferred it from patterns in their ordinary shopping. Traditional notions of consent, where a consumer ticks a box agreeing to share certain information, do not cover inferred data. A person who buys unscented lotion has not consented to having her reproductive status guessed at and acted upon.

The Indian regulatory response 🔗

This question is no longer just academic. India’s Digital Personal Data Protection Act, 2023, along with the DPDP Rules, 2025, was notified by the Ministry of Electronics and Information Technology on 13 November 2025, with provisions rolling out in phases through May 2027. Among other things, the Act requires that consent be explicit, informed, and freely given, and that data fiduciaries clearly communicate how and why data is processed . Crucially, the Act prohibits behavioural monitoring and targeted advertising directed at children without verifiable parental consent. Had Target’s system been operating under this framework, sending pregnancy-related mailers to a minor would have faced serious legal scrutiny.

The Act also pushes advertisers toward data minimisation. As commentary in Law.asia notes, businesses that previously relied on broad-based data collection must now restrict themselves to information that is strictly necessary, use clear privacy notices, and explore techniques like anonymisation or zero-party data strategies where consumers voluntarily share preferences.

The creep line 🔗

Target’s own response to its controversy was revealing. Rather than abandoning the model, the company disguised it. By burying pregnancy-targeted coupons between unrelated offers, Target tried to keep the benefits of prediction while softening its perceived intrusiveness. This raises a hard ethical question: is it better to be creepy and honest, or effective and hidden? Many privacy scholars argue that hiding intent is actually worse, because it denies consumers the chance to object or opt out.

What the case teaches advertisers today 🔗

A decade and a half after the episode, three lessons stand out for anyone working in advertising, marketing, or public policy.

Prediction is not neutral 🔗

An algorithm that correctly guesses something personal about a customer is not simply offering convenience. It is making a statement of knowledge, and that statement has emotional and social consequences. A pregnancy coupon arriving at the wrong address can rupture a family.

Trust compounds more slowly than data 🔗

Advertisers often treat personalisation as a pure efficiency gain. But every interaction either adds to or subtracts from the consumer’s trust. Targeted campaigns that feel invasive can erode loyalty faster than clever targeting can build it. In the Indian context, where industry commentators now argue that compliance with the DPDP Act can itself become a brand differentiator, trust is emerging as a measurable asset, not just a soft value.

Ethics is now infrastructure 🔗

What used to be optional, things like privacy by design, impact assessments, clear consent flows, and vendor audits, is rapidly becoming legally required. The Target case, which once seemed like an outlier, has become a template for why regulators around the world, including India’s Data Protection Board, are tightening the rules.

A case study that refuses to age 🔗

The Target pregnancy prediction story endures because it captures, in a single narrative, everything that makes modern advertising both powerful and dangerous. It is a story about mathematical elegance, about the quiet genius of consumer psychology, and about the price of mistaking data access for the right to use it. It remains the cleanest illustration of a truth that marketers, regulators, and students of public administration all have to grapple with: the technology is far ahead of our shared ethical vocabulary, and closing that gap is one of the defining challenges of our time.

What do you think? If a retailer can predict something intimate about you from your ordinary purchases, should it be allowed to act on that knowledge even when the prediction is helpful? And where, in your view, should the line sit between smart personalisation and a quiet violation of dignity?

References

1. https://longreads.com/2012/02/16/how-companies-learn-your-secrets/
2. https://ethicsatwork.nd.edu/resources/how-companies-learn-your-secrets/
3. https://d3.harvard.edu/platform-digit/submission/big-data-and-retail/
4. https://andyneely.blogspot.com/2012/02/how-companies-learn-your-secrets.html
5. https://d3.harvard.edu/platform-digit/submission/target-you-cant-hide-that-baby-bump-from-us/
6. https://www.dlapiperdataprotection.com/?t=law&c=IN
7. https://www.cookieyes.com/blog/india-digital-personal-data-protection-act-dpdpa/
8. https://law.asia/dpdpa-advertising-compliance/
9. https://www.campaignindia.in/article/digital-personal-data-protection-act-marketings-new-era-of-trust/497259