How we rank K-beauty

What the rankings measure

Each ranking shows relative search interest: how often people in one country searched for a product compared with the other products in the same category. A score of 100 is the most searched product in that list; a score of 50 means roughly half as much search interest.

Search interest is a signal of attention and curiosity, not of sales or quality. A product can be searched a lot because it is new, viral or hard to find. That's why we show reviews from real users separately and never mix them into a 'best product' claim.

Where the data comes from

For the United States, the United Kingdom, Canada, Australia, Japan and France we use Google Trends data, collected through the DataForSEO API.

For Korea we use NAVER DataLab search trends (through NAVER API HUB), because most people in Korea search on NAVER rather than Google. Korean rankings are therefore labelled 'NAVER' and are compared only with other products inside Korea.

We do not copy rankings, sales charts or reviews from online stores or other websites.

Local product names

The same product is searched differently around the world. COSRX's snail essence might be searched as 'cosrx snail mucin' in the US, 'コスアールエックス スネイル' in Japan and '코스알엑스 달팽이 에센스' in Korea. For every product we keep a list of the names people actually use in each country and add their search interest together.

Nicknames that could point to several products (for example just 'snail mucin') are used for site search only, never for rankings, so one product can't take credit for another's searches.

Making every number comparable

Search trend tools only compare a handful of terms at a time and rescale every request to 0–100, so numbers from different requests can't be compared directly. To fix this, every request includes the same reference term for that country (for example 'korean skincare' in the US or '화장품' in Korea). Each product is measured against that reference, which puts all products on one scale.

Smaller products can round down to zero next to a big reference term. When that happens we measure them again against a smaller reference, up to three steps, and convert the result back to the same scale.

Time window and updates

Rankings use the last three months of search interest, which smooths out one-off spikes. We update them once a month and keep every past month, so you can see what moved: ▲ and ▼ show the change in rank since last month, and NEW marks products that weren't ranked the month before.

Each month we also publish a trends report that highlights products popular in many countries at once, products that are big in only one country, and the brands that appear most often.

What we never do

Brands can't pay to appear in or move up the rankings. We have no paid placements.

If we ever add shopping links, they will be clearly marked as affiliate links, and they will have no effect on the order of any ranking.

Limits you should know about

Countries with less search volume produce noisier data, especially for lower-ranked products, so small differences near the bottom of a list don't mean much. Products with equal scores share the same rank.

New products appear once they have enough search history. If a product you love is missing or seems to be named wrong, please tell us through the contact page.

Reviews

Reviews on K-duck are written by signed-in users, who add their skin type, climate and skin tone so you can find opinions from people with skin like yours. Every review is checked automatically for spam and abuse, and new accounts' first reviews are checked by hand before they appear.