You Think You're Picking Products. It's the Algorithm Picking You.

A few days ago, a friend vented to me.

She said she'd chatted with her best friend on her phone about candles — a few sentences, nothing more. Then she opened Instagram, and right there on her feed was a Yankee Candle ad.

She asked me: are they listening in on me?

I said the answer to that question might be more interesting than you think. In 2025, a group of marketing scholars from several universities in the UK and Denmark ran a study. They recruited 20 ordinary consumers and asked each of them to draw out the complete path of their most recent online purchase — where they first saw the brand, which ads they clicked, what prices they compared, where they finally placed the order — and then interviewed them one by one, question by question.

When the results were published, the paper became one of the most-cited articles in advertising research that year.

Why did it get so much attention? Because it punctured a long-standing illusion:

You think you're browsing the store. It's the store that's browsing you.

What Does It Mean to Be "Browsed by an Algorithm"?

Let me tell you three little stories from the study.

The first — let's call her Ellen. She was scrolling Instagram when she came across an ad for Bondi Boost, an Australian haircare brand she'd never heard of. She didn't buy. She just browsed the brand's page and its comment section. A few days later, a retargeting ad with a discount found her, and she placed an order.

The second, Paul. He saw a carousel ad for sneakers on Facebook and was drawn to one pair of Nikes. He went to the Nike official site — out of stock. He searched retailer after retailer — all out of stock. In the end, it was a shoe store called Schuh that precisely targeted him with an ad, and he bought from Schuh. The interesting part: afterward, Paul couldn't remember at all whether that first ad had been placed by Schuh. The Nike logo was so bright that it reduced whoever paid for the ad to a piece of background scenery.

The third was a bride shopping for a wedding dress. She searched around, and the wedding dress ads the algorithm served her were, in her own words, "99.9% styles I absolutely hate." Her conclusion: this AI doesn't understand me well enough yet.

Three stories, three endings. Someone was gently led by the algorithm to a new brand; someone was chased by the algorithm until the loop closed; someone was annoyed by the algorithm into rolling her eyes.

This is the real situation of doing business today: the algorithm can help you close the sale, and it can also help you push customers away.

Consumers Do the Privacy Math Better Than You Do

There was one finding in the study that stopped me in my tracks.

Almost every participant said they worried about privacy. But guess what? The vast majority, the moment a cookie pop-up appeared on a webpage, would click "Accept All" with their eyes closed.

Why?

One participant put it very plainly: I want the pleasure of "press one button and find what I want." Paying a little data for that is worth it.

Another participant, who lives in a small city, put it even more bluntly. Asked whether she feared data leaks, she said: no, because the things on Amazon simply can't be bought in the stores near my home.

You see, consumers do the math on privacy. In the first half of the journey (discovery, comparison), the benefit of convenience is tangible and real, while the cost of privacy is invisible and untouchable — so people pay up gladly. But the moment they reach the checkout, the math flips — by then the convenience has already been enjoyed, and any hint of "being manipulated" makes people turn on a dime.

One participant said she deeply resented those discount pushes that "can influence me within seconds," and felt the design of this whole purchase journey was "off." Another said it straight: whatever the brand does, just say so honestly. Don't dress yourself up as eco-friendly or conscientious, or I won't spend a single cent on you.

The same group of people: in the first half of the journey, they trade privacy for convenience; in the second half, they abandon discounts for dignity. If your ad budget lands on the wrong stage, the money is burned for nothing.

The Most Wronged Losers: Loyalty and Repeat Customers

The study had another finding, one I think every brand builder should copy onto their wall.

Brands pour enormous effort into using AI for acquisition, traffic, and conversion — but the last two stages of the customer journey, loyalty and advocacy, are almost a blank space. People buy, and that's it; very few go on to write reviews or recommend to friends. It's not that customers are unwilling — it's that brands use the wrong methods: frenzied retargeting at people who just bought, chasing them with ads, until customers feel they are being strong-armed into buying more. So they block, unfollow, or even blacklist the brand.

Meanwhile, it's the "clumsy old methods" that are working. One participant's contact lens supplier cut his order because of a stockout — and not only did he not leave, he said he'd buy again next time. Why? Because they sent him email after email: out of stock, there will be a delay, here's when it will arrive — they kept him in the loop the whole time. Another participant received an email with shipment tracking that "read like a human wrote it," her goodwill soared, and she went back and gave them a serious five-star review.

Good grief. The thing that defeated the sophisticated algorithm was an honest, human-written email.

And there was one counterexample. A participant saw a product in an ad, clicked in excitedly — out of stock — and the page even graciously suggested, "You might also like…" Her reaction: no. What I liked was the one in your ad. You had no stock and still lured me in — I'm not placing this order.

This routine of "hook you at the door with a desirable item, then hand you off to a substitute" has an old name in sales psychology: the low-ball technique (an old sales trick: win you over with one offer, then quietly switch it). It existed fifty years ago. The algorithm didn't invent a new trick — it just delivers the old trick faster, more precisely, and in a far more infuriating way.

Three Bottom Lines

So what should brands actually do? The researchers offered a framework — three E's — and honestly, it couldn't be more straightforward.

First, govern yourself. Where does the data come from, where is it used — get clear on that yourself first, before the regulators come knocking. EU GDPR fines are not for decoration.

Second, hand control back to the customer. Make cookie management simple; when a customer asks "why are you showing me this," honestly tell them what data you used; go easy on the retargeting — don't push people all the way to blocking you.

Third, proactively cultivate the relationship. Loyalty programs, convenient entrances for viewing and editing your data, a customer service line an actual human answers — these "heavy" chores are precisely what the algorithm cannot replace.

The researchers also cited a number: 92% of customers, once they doubt data security, will uninstall the app and disappear from the brand entirely.

92%. That number is worth remembering for a lifetime — more than any ad-optimization technique.

Back to my friend frightened by the candle ad. Whether devices are actually "listening in" — scholars are still arguing about it. But the more piercing conclusion of this study is this: most people, while feeling they're being surveilled, keep trading privacy for convenience — and even if they really are being listened to, as long as the recommendations are accurate enough and delivered gracefully, many people will keep cursing with their mouths while tapping with their fingers.

Customers forgive technology. They never forgive arrogance.

The stronger the technology, the more humility it demands. That may be the most counterintuitive — and most valuable — piece of common sense for building a brand in the age of AI.

May you never mistake your customers' tolerance for the algorithm's achievement.