You were going to buy the beer.

11:47 PM, Tuesday. You put Corona and ice in a cart on TaDa — AB InBev's Latin American delivery app — and then you closed the tab. Maybe you got tired. Maybe you decided you didn't need it. Maybe you just wanted the feeling of wanting it without the commitment of having it.

Twenty-four hours later, an email arrives.

Inside: a video. In the video, you're at a rooftop. The city glitters below. Friends you haven't seen in months are there, laughing, and someone has put on music you actually like, and the night is extending itself the way nights only do when they're good. You are holding a beer that is cold and perfect and the version of you that clicked "checkout" is living a life you didn't choose.

The tagline: momentos compartidos que no sucedieron… por ahora — shared moments that didn't happen… for now.

This is not a retargeting ad. This is a ghost story.

1. The Archaeology of a Forgotten Cart

Every abandoned cart is a small death.

You made the choice — you reached for it, selected it, moved it toward ownership. The psychological contract was almost signed. And then something interrupted the ritual. You closed the laptop. You fell asleep. You decided that wanting was enough.

What you didn't know: the algorithm watched you the entire time.

Cart abandonment is not a failure of commerce. It is a measurable psychological space — a gap between intention and action that e-commerce platforms have spent a decade studying. The Baymard Institute puts the average cart abandonment rate at nearly 70%. That means seven out of ten times a human being moves toward something they want, they stop before the finish line.

The industry standard response: the reminder email. "Hey, you left stuff in your cart!" with a gentle nudge and maybe a discount code. Standard. Forgivable. The digital equivalent of a tap on the shoulder.

AB InBev's TaDa, with its "Abandoned Nights" campaign created by R/GA, did something different.

Instead of reminding you what you left behind, it showed you what you left behind became.

2. How the Ghost Is Generated

The campaign launched in Mexico in April 2025, the first market for what AB InBev described as an "exploration of how AI can be used to generate hyper-personalized advertising at scale."

Here's what it does:

The system pulls your name, your location, and what's in your cart. That data feeds into a pipeline that also includes OpenAI for script generation, Replicate for image synthesis, ElevenLabs for voice narration, and Imposium (an R/GA Ventures portfolio company) for personalized video assembly. The result is not a generic "hey, come back!" message. It is a cinematic rendering of an alternate life that would have existed if you'd followed through.

The rooftop party. The gathering that materialized. The spontaneous night that shimmered into being in an AI server somewhere, calculated from your behavioral data, your social patterns, your likelihood to share a bottle with someone specific.

The campaign is opt-in — you have to be a TaDa customer, you have to have cart data — but the ghost it generates is specific in a way that generic retargeting never achieves. This is not "someone bought this item." This is "this is the version of you that existed at 11:47 PM on Tuesday, poised and ready, and the algorithm remembers every detail."

We're thinking about pain points. When we started, from 10 stories we had one that was, like, legally correct. It was like, "Oh my god, no, this is going to be a problem."

— Josefina Casellas, VP and executive creative director at R/GA SSLATAM, via The Drum

The early iterations pulled in so much data that the AI-generated narratives pushed against what could legally appear in an ad. The final version is stripped down — just name, location, cart contents — but the effect is the same.

The algorithm does not forget what you almost were.

3. The Uncanny Valley of the Almost-Self

Here is the part that should unsettle you.

Retargeting has always operated on a simple logic: you showed interest, we remind you of that interest, you maybe convert. The relationship is explicit. You looked at a thing. The thing follows you. You know why.

What AB InBev has built is something else.

The abandoned cart video does not show you the product you left behind. It shows you a life you would have had if you'd bought it. The friends who would have gathered. The night that would have unfolded. The self that followed through when you didn't.

This is the doppelganger effect, rendered in AI: the algorithm summons the version of you that clicked "checkout."

Research in psychology and marketing has identified a specific phenomenon at work here. When personalization feels too intimate — when the system seems to know things about you that you didn't tell it — discomfort rises even as engagement increases. Studies in the creepiness of personalized advertising show that the gap between "relevant" and "violating" narrows when the system demonstrates knowledge of your internal states, not just your behaviors.

The abandoned cart video knows you wanted to buy the beer. It also knows what you almost had after the beer. The night that the purchase would have enabled. The version of you that was one click away from being real.

That's not retargeting. That's resurrection.

4. The Attention Economy Finds Its Liminal Form

The word "liminal" means existing on a threshold. Between one state and another. Between here and there. Between the click and the arrive.

The gap between "want" and "buy" has always been where the attention economy lives. That's where the friction is. That's where the optimization happens. E-commerce has spent two decades trying to close that gap — better checkout flows, one-click purchasing, friction reduction, desire compression.

"Abandoned Nights" doesn't try to close the gap. It inhabits it.

The campaign takes the liminal space — the space between intention and action, the moment between click and arrive, the self you almost was — and makes it into content. Not a reminder to complete the purchase. A narrative about what would have happened if you had.

This is a different theory of attention. Not "capture it" or "hold it" or "redirect it." Haunt it.

The ghost of the abandoned cart becomes a product. The liminal space between wanting and having becomes a content format. The algorithm learns to monetize not just your behavior, but your almost-behavior — the version of you that existed in the moment of intention before you pulled back.

What the TaDa campaign reveals is that the attention economy has found a new form: not the ad that follows you, but the ad that remembers who you almost became.

5. What the Algorithm Remembers

The video ends.

You're not on the rooftop. You're at your desk, or in bed, or sitting on the couch with the tab still open from 24 hours ago. The friends aren't there. The beer is unpurchased. The night is un-had.

But somewhere, on a server, the ghost persists.

The algorithm doesn't forget. It can't. It exists in the space between your data and your desire, and it has built a memory of every version of you that almost existed — the nights you almost had, the purchases you almost made, the selves you almost were.

momentos compartidos que no sucedieron… por ahora
The shared moments that didn't happen… for now.

6. What We Can Learn From the Ghost

The "Abandoned Nights" campaign is corporate surrealism at its most precise: a real ad that functions as a horror film, a retargeting strategy that doubles as existential art, a piece of marketing that shows you exactly what it knows about the gap between who you are and who you almost became.

The lesson is not "this is creepy." The lesson is more specific than that.

The algorithm has learned to monetize the liminal space itself.

For two decades, the attention economy optimized for capture and retention — get the click, hold the attention, earn the conversion. The next phase is different. The next phase inhabits the gap. It doesn't just know what you want; it remembers what you almost wanted, what you almost did, who you almost were.

The abandoned cart is not a failed conversion. It's a data point about a version of you that exists only in the space between intention and action.

And now that version has a video.

The question the campaign leaves unresolved — is this haunting, or helpful? Creative, or invasive? A clever piece of marketing, or something genuinely new under the attention economy's increasingly strange sun?

The answer is probably all of these at once.

But the liminal space persists.

Between the click and the arrive. Between the self you were and the self you might have been. Between what you almost bought and what it would have become.

The algorithm remembers.

It always remembers.

References