The Algorithm Knows Your Bra Size and It Will Not Stop Emailing You About It

It started innocuously enough. You Googled a brand of running shoes once — maybe twice — and suddenly every website you visited for the next three weeks was wallpapered in that specific shoe in that specific colorway. Fine. Annoying, but fine. That was retargeting, the internet's most obvious party trick, and you learned to ignore it.

That was years ago. The technology has since gone to graduate school, gotten a PhD in you specifically, and moved into your phone.

From Targeted Ads to Targeted Feelings

Hyper-personalization — the practice of using behavioral data, purchase history, location, browsing patterns, and approximately ten thousand other data points to deliver marketing so specific it feels personal — has become the default mode of consumer capitalism. Not the ambitious mode. The default.

Spotify knows you well enough to name a whole playlist after your personality. Netflix's homepage looks nothing like your roommate's Netflix homepage. Amazon has an entire section called "Inspired by your browsing history," which is either a feature or a confession, depending on your mood. Stitch Fix sends you clothes chosen by a combination of human stylists and machine learning, based on a quiz you took about your lifestyle that somehow felt more intimate than a first date.

The pitch is irresistible: no more wading through things you don't want. The algorithm filters the world down to exactly you. Every recommendation, perfectly calibrated. Every ad, a thing you were probably going to want anyway.

The reality is that "exactly what you want" turns out to feel, at scale, a little like being watched.

The Intimacy Nobody Agreed To

Let me tell you about the email I got last spring. I had been browsing — not buying, just looking — at a particular style of linen pants on a mid-range clothing website. I didn't add anything to my cart. I didn't create an account. I lingered on the product page for maybe two minutes and then closed the tab to go back to doing something else.

Three days later, the brand emailed me. Not a generic promotional email. An email that said, in the subject line, "Still thinking about those pants?" with a photo of the exact pants I had looked at, in the color I had spent the most time viewing.

I had not given them my email address. I had not agreed to marketing communications. I did not know they knew who I was. And yet there they were: still thinking about those pants.

I was, for the record, still thinking about those pants. That's almost the worst part.

The Uncanny Valley of Consumer Experience

There's a concept in robotics called the uncanny valley — the point at which a humanoid robot becomes realistic enough to seem almost human, which makes it feel profoundly disturbing rather than charming. Hyper-personalized marketing has its own uncanny valley, and we are living inside it.

Generic advertising was easy to dismiss. A billboard for a car you'd never buy, a TV commercial for a medication whose side effects the announcer reads faster than human speech — none of it felt personal, so none of it felt intrusive. It was wallpaper. You tuned it out.

Personalized advertising, done moderately well, is useful. A recommendation for a book by an author you've previously enjoyed. A coupon for a brand of coffee you actually buy. These feel like services.

But when the personalization becomes granular enough — when the brand references something you didn't consciously register as a preference, when the timing is too precise, when the recommendation is for something you were only half-aware you wanted — it stops feeling like service and starts feeling like surveillance wearing a friendly font.

Meta's advertising platform allows marketers to target users based on inferred interests derived from behavior the user never explicitly declared. Google's ad ecosystem operates similarly. Data brokers sell profiles built from hundreds of sources, stitched together into a portrait of you that you have never seen and cannot fully access. The resulting ads don't feel like ads. They feel like someone finished your sentences.

Customization as Control (Theirs, Not Yours)

The language of personalization is always framed around consumer empowerment. We're giving you what you want. We're cutting through the noise. We're making your experience better. And there's enough truth in this to make the argument feel solid.

But personalization also narrows. When Spotify's algorithm decides what you like, it feeds you more of it, which reinforces what you like, which means the algorithm feeds you more of it. Your "Discover Weekly" playlist, which feels like it was made for you, is also slowly made by you — a mirror that gradually stops showing you anything you haven't already seen.

The same dynamic applies to product recommendations. Amazon's algorithm is optimized to show you things you are likely to purchase, not things that might surprise you, challenge you, or serve you better than you currently know. The algorithm is not your friend. It is a very attentive salesperson who has memorized your file.

And brands have learned that intimacy, even simulated intimacy, converts. Emails with personalized subject lines have open rates 26 percent higher than generic ones, according to Campaign Monitor data. Personalized product recommendations drive 31 percent of e-commerce revenue, per Barilliance. The creepiness is, financially speaking, working.

When Knowing Too Much Becomes Its Own Product

There's a specific flavor of modern anxiety that arrives when you mention something casually — in conversation, in a text, out loud near your phone — and then see an ad for it within the hour. Most of the time, this is confirmation bias. Our brains notice the ad because we were already thinking about the thing. But the feeling it produces — that low-grade, hard-to-articulate sense of being overheard — is real, and brands benefit from it even when they didn't cause it.

Because the feeling makes the ad memorable. The creepiness is the engagement.

We have arrived at a moment where the most effective marketing strategy is to make consumers feel slightly surveilled, slightly known, slightly seen in a way that skips consent and goes straight to familiarity. The brand that emails you about the pants you almost bought isn't your friend. It's not even particularly good at being a brand. It's just very, very good at knowing when you're wavering.

The promise of personalization was a world shaped around your preferences. What we got instead is a world that has studied your preferences so carefully it can predict them before you've had them — and then sell them back to you at a 15 percent discount if you use the code in the email it already knew you'd open.