Manufacturing Taste

Netflix never learned your taste. It built the feeling of being understood, one data point at a time.

You collapse onto the couch and open Netflix. Before you've even decided what you want to watch, the screen already has an answer: pick up where you left off last night, another film from a director you loved, a title that people "with similar taste" are raving about. Your thumb finds the play button with almost no resistance — the way it would for a recommendation from an old friend. How does a screen come to know your taste better than your closest friend does? Sometimes better than you know it yourself?

This isn't luck, and it isn't intuition. It's data and artificial intelligence working together to produce the most sophisticated sizzle marketing has ever made. If advertising and UX design used to be a game of turning up the volume for an anonymous crowd, this new sizzle runs the opposite direction. Instead of shouting louder, it whispers one perfect sentence to one person at exactly the right moment. The marketer of the past painted a single mural for everyone. The marketer of today is closer to an architect, building a mosaic that rearranges itself in real time for each person's private cravings.

And that mosaic is no longer assembled by feel. Where a marketer's gut instinct used to sit, numbers now sit instead — and it isn't a person who arranges those numbers, but an algorithm. This assembly is evolving in two stages. First: picking the best-fitting thing from what already exists. Second: building, from scratch, something that never existed before — for one person alone. Most brands are still in the first stage. The furthest-out ones have already crossed into the second.

The line between the two is simple. Discovery means choosing one thing that already exists. Creation means making something that didn't exist, for one person alone. The first role is a librarian's; the second is a novelist's, writing a book on the spot for a single reader. Right now, both roles work the same screen at once. The librarian shows you to your seat — and by the time you sit down, the novelist is already waiting there for you.

Why does a recommendation feel like such a powerful sizzle?

Start with the recommendation engine itself — Netflix's film suggestions, Spotify's playlists, Amazon's "customers also bought." Each one reads the trail you've left behind and guesses your next want before you've named it. What makes this so powerful isn't just convenience. It's that it touches something deeper: the feeling of being understood.

The more information there is, the heavier a choice becomes. Decision fatigue — the exhaustion of not knowing what to pick — is one of the most common frustrations of modern life. Faced with hundreds of films, tens of millions of songs, hundreds of thousands of products, people freeze. A recommendation engine takes on that search for you, laying out things you're likely to love before you even ask. Cognitive ease peaks right here. Alongside it comes the pleasure of stumbling on a taste you didn't know you had. Spotify's weekly "Discover Weekly" mix feels like an album curated by a DJ friend who knows your taste perfectly. The labor of choosing disappears; only the joy of discovery remains.

Then there's one more layer: "People with taste like yours also loved this." That single line quietly plants a sense of belonging — the sense that your taste isn't solitary but part of some discerning tribe. At the same time, it delivers the feeling that your past choices have been noticed, analyzed, respected. It reads as the strongest possible proof that a brand sees you not as an anonymous customer but as a person with a taste of your own. This is why data, which should feel cold, ends up feeling warm instead.

It's strange, when you think about it. The thing recommending you things is a faceless server running equations. And yet the moment you receive its output, you feel warmth in it. That translation — cold data material becoming the warm sensation of being understood — is the real magic of hyper-personalized sizzle. The technology is cold. What it produces is not.

What a recommendation engine sells isn't content. It's the feeling of being understood.

When AI stops finding and starts making

Everything above was discovery — choosing the best-fitting thing from what already exists. Generative AI moves past that role. Now AI builds, on the spot, a sizzle that never existed before, made for exactly one person. It's the shift from curator to creator.

Ad copy is the clearest example. Search for dog food, and minutes later your Instagram feed shows a photo styled around a golden retriever with the line: "Start protecting your energetic pup's joints today." Your friend's feed, meanwhile, shows a shih tzu and an entirely different line: "Stop worrying about your sensitive pup's tear stains." Same product, different sizzle. An AI has taken each person's data as raw material and assembled, in real time, the most persuasive possible image and copy. The two of you have never seen the same ad.

Product design follows the same path. A running shoe brand's AI chatbot opens a conversation: let's design a shoe that exists nowhere else but for you. Drawing on the conversation and your purchase history, it generates a design on the spot, shaped to your foot, your habits, your taste. This isn't customization — a name stitched onto a stock product. It's a deeper act of creation, one where you participate in the moment a product is born.

Even the app interface isn't exempt. Frequently used menus quietly migrate to more reachable positions. Text automatically enlarges for users with weaker eyesight. The screen itself rebuilds itself, moment to moment, around who is looking at it and why. The screen you saw yesterday and the one you're seeing today are already two different things. The era of a brand having one fixed face is ending. Today's brand has as many faces as it has people looking at it.

This shift deserves a name. Call it the Genesis Sizzle — the moment an algorithm, not a brand, forges the sizzle itself, from nothing. There is no longer one right answer for the average customer. There are only countless different right answers, each one for the you standing here right now. "Genesis" might sound like overreach, but that is, in fact, what is happening. The brand's message isn't being adjusted toward you. It's being made, from the ground up, for you. In the age of discovery, the right answer already existed somewhere on the shelf. In the age of creation, the right answer comes into existence only at the moment it meets you.

The room without a door

But a world built for you alone is also a world where you alone are locked in. A screen that shows you only what you're likely to love is comfortable — and it also quietly isolates you from what you don't want to see, from opinions unlike your own. This is what's commonly called the filter bubble. A shopping app that keeps showing you the styles you already like is, without announcing it, withdrawing your chance to try something new. The room the creator builds is perfectly comfortable. It has also quietly erased its own door.

What's unsettling is that this isolation doesn't feel like isolation. It feels like comfort. Things that don't match your taste never appear before your eyes in the first place, so the sense of being confined never arises either. Diversity is traded away for comfort, and no one notices the trade taking place. However pleasant the room the creator builds, the fact remains: there is no door.

Think of someone who's shown, over and over, only videos that match their politics. They gradually come to believe their view is common sense shared by everyone, and slowly stop listening to anything else. A brand's hyper-personalization runs on the same logic. A screen that keeps repeating familiar tastes back to you looks like respect for your taste, but is actually confining that taste inside a narrow pen. The world the creator draws is a world made for you. It is also, only, a world made of you.

This is why well-designed personalization can't stop at simply satisfying you. It has to miss the mark, pleasantly, once in a while — leaving room for a taste you didn't know you had, a discovery you weren't looking for. Without this margin for what might be called serendipity, personalization stops being a window that widens your taste and becomes a wall that encloses it. Personalization without that margin ends up as nothing more than a comfortable cell.

The filter bubble and the manipulation that follows might look like separate problems, but they share a root. The algorithm, having become a creator, has gained the power to decide, on its own, how far it will take you. Point that power toward narrowing the range of your taste, and you get a filter bubble. Point it toward shaking your judgment itself, and you get manipulation. Different directions, one origin.

Where persuasion turns into manipulation

A more fundamental problem follows. Something that understands you perfectly can also shake you perfectly. Understanding and manipulation stand on the exact same data, the exact same technology. What separates them is only the intention behind the use.

Imagine an AI that watches not just your browsing but your emotional state, your finances, your history of impulse purchases. It pinpoints the exact moment you're loneliest and most vulnerable — right after a breakup — and puts an ad on your phone: "Forget the heartbreak. Treat yourself to something sweet." Or the moment your portfolio crashes and panic sets in, it pushes: "Talk to an investment expert now" — leading straight to an expensive paid consultation. Both moments are the result of data landing exactly on target. Except the target this time isn't your need. It's your weakness.

Past this point, it isn't persuasion anymore. It's the deliberate selection of someone's most vulnerable moment in order to extract an irrational decision. Data was originally a tool meant to make a customer's life better. The instant that tool turns to aim at a customer's weakness instead, the creator stops being a helper and becomes a manipulator. Persuasion leaves room for the other person to decide for themselves. Manipulation erases that room in advance and makes the decision for them. Both wear the same label — "recommendation" — on the outside. One opens a door. The other pushes you through it.

What makes this risk hard to catch is that it rarely arrives wearing an unpleasant face. It usually arrives wearing the friendliest face imaginable — the face of something that knows you best. In the moment, the consumer doesn't feel violated. They feel like they've just run into exactly what they needed. Only the party who designed the system knows where that line actually sits. Which means policing this line isn't really something a consumer can do from the outside. It falls, in the end, to the brand's own restraint — a line drawn from the inside, invisible from anywhere else.

On the surface, the two sizzles are nearly indistinguishable. "We understand you" and "we know your weakness" arrive on screen in the exact same warm tone of voice. The only difference is what the brand aimed at, behind that sentence. Aim at a need, and it's still good sizzle. Aim at a vulnerability, and no amount of polished copy changes what it already is: manipulation. You can't tell the two apart by reading the sentence. The only thing that gives it away is the timing of its arrival.

The weight of creating

There is one question a brand has to answer here: is this data being used to make a customer's life better, or to find the crack in it? The same precision produces two entirely different outcomes. One leaves behind the memory of being understood. The other leaves behind the memory of being used.

The fuel behind all of this is personal data. The more accurate the recommendation, the more perfectly tailored the experience, the more data it demands. And the consumer lives permanently inside this contradiction — worried about being watched, yet unwilling to give up the convenience personalization offers. This tension, often called the privacy paradox, will only pull tighter as hyper-personalization keeps getting more precise.

On this uneasy balance, the least a brand can do is refuse to hide the terms of the exchange — being clear about what it's taking, how much, and whether it can be undone. However sophisticated the data, without trust it's a house built on sand. Transparency isn't a drag on personalization. It's the minimum ground personalization needs to survive.

The marketer of the past picked up a brush. The marketer of today holds data, and creates by borrowing an algorithm's hand. But the seat of creator is both power and obligation at once. To build someone's taste is also, inevitably, to build your way into the cracks of their day and their mind. Where will that power be pointed? A brand that can't answer this question honestly will, however elaborate its personalization becomes, eventually lose something far larger: the customer's trust.

No brand chose to become a creator. As data accumulated and technology advanced, brands arrived at this seat whether they wanted to or not, carried there by momentum. But not choosing the seat doesn't excuse what comes with it. If anything, the opposite is true — the more unintentionally a power lands in your hands, the more deliberately you have to ask what you'll do with it. Becoming the thing that understands you most perfectly, and becoming the thing that protects you most perfectly — closing the distance between those two is the next assignment waiting for sizzle, now that it wears the shape of data.

The most personal message is not always the most honest one.