We called next week's #1 in public. The first two landed.

There are two ways to talk about what's winning in K-beauty. One is to wait for the chart to move and then explain why it moved — the safe way, because you already know the answer. The other is to say what will happen next before it does, write it down where everyone can see it, and let the next week prove you right or wrong. Two weeks ago we started doing the second one. This week the scoreboard turned over its first results.

What we called, and what happened

Each week we name the single product we expect to be number one on a chart the following week, and we lock it publicly before that week starts. For 2026-W34 we had two live calls on our API-collected Olive Young charts — the Korea best-seller and the global one. On both, we called MEDIHEAL Essential Mask Sheet to stay at number one. When this week's captures came in, MEDIHEAL was number one on both. Two calls, two hits.

Why we could call it

The rule behind the call is deliberately dull: we look at how many weeks in a row the current leader has held the top spot, and if that streak is long, we predict it holds one more. MEDIHEAL had held Korea's number one for six straight weeks going into the call — a streak that rarely breaks in a single week. That's not a hunch or a hot take; it's a mechanical read of our own archive. The interesting weeks will be the ones where the streak is short and the call is a coin toss — and those are coming.

The number we are not counting

We also ran the same rule backward over past weeks to see how it would have done: 8 of 9, about 89%. It's tempting to lead with that number, and that's exactly why we don't. A backtest is hindsight — you already know how the weeks turned out. It tells us the rule isn't nonsense, but it is not a record. Our record is only the calls we locked in public before the week, and that record right now is small and honest: two for two. It grows one real week at a time, on the page, where you can watch it — including the weeks we get wrong.

Why this is hard to fake

A prediction is only worth anything if it can be proven wrong. Ours can: a call is marked incorrect the moment the next capture's number one is a different product — no reinterpreting after the fact. And we only make product-level calls on the charts we collect by API, where a product's identity is stable enough to check week to week; the screenshot-captured US charts are held out, because you can't fairly score a prediction against a number you can't pin down. The whole point of putting the call in public, early, and falsifiable is that it can't be quietly edited once the week disagrees.

This is the same idea the rest of this site runs on — the receipt matters more than the take. A prediction is just a receipt written in advance.