AI became a shopping channel this year. Here's the K-beauty it's discovering.

Two entries ago we argued the old moat was opacity; last time, what its collapse costs a brand that can't see. This closes the trio with the number underneath both. Per that essay, shoppers arriving from AI assistants converted about 40% worse than everyone else a year ago; today those same shoppers convert roughly 42% better, and AI traffic to US retail is up around 1,300% since late 2024 (Adobe Analytics). Mobile took a decade to earn that trust. AI took about twelve months.

The question changed

When a real share of buying starts with a question to an assistant, the old question — is my ad seen, is my page ranked — matters less than a new one: what does AI name when someone asks for K-beauty? That answer is a shelf of its own now, and it's the one a growing share of shoppers see first.

What AI is actually surfacing

We measure it weekly. The brand AI names most across our fixed question set — 124 mentions — is one that doesn't appear on Olive Young Korea's top 100 at all. Four of AI's top five names are absent from Korea's chart. The discovery layer and the shelf are not reading from the same list.

Two canons

There's a clean split underneath it. AI names the internet-famous — the brands that fill English-language routines and get written about. Korea buys the mask-and-basics volume — the everyday names the videos rarely mention. Neither list is wrong; they answer different questions: what's talked about versus what's bought. The mistake is assuming they're the same shelf.

Why we count both

Because the gap is the signal. A brand AI loves that hasn't reached the shelf is an early read; a brand deep on the shelf that AI never names is invisible in the channel growing fastest. We publish both, weekly, and say what each measures — AI mentions are recommendation surface, not sales; chart rank is shelf position, not units. Two canons, one archive, so you can watch them move toward or away from each other.