About The K-Beauty Data Desk

Built by betterdxd, a cosmetics-loving developer who wants data to make product research more useful—not more confusing.

What this site is

Korea's K-beauty retail rankings provide aggregate signals from skincare audiences. Each article here takes one product from a captured ranking snapshot and turns those aggregate ratings and polls, plus a capped sample of anonymous topic counts and metadata flags when available, into plain English. The output distinguishes ratings, poll representation, topic labels, and buying checks instead of treating them as product outcomes.

Where the numbers come from

The average rating, review count, star distribution, and displayed skin-type, concern, and gentleness polls come from the captured K-beauty retail product page. When deep analysis is available, the pipeline temporarily examines a capped sample of up to 500 helpful review records, discards their text, and retains only anonymous topic counts and metadata-flag shares. Those flags do not verify buyer identity, intent, satisfaction, or causation. I do not test products on my own skin, and I say so in every article.

Check me

Every post states its capture date and data limits so readers can distinguish a dated snapshot from current retail information. Prices are quoted only as dated anchors with market and date labels, and current listings must be checked before purchase. When data is missing, articles say "no data" instead of printing a zero, and a small captured review base is flagged as limited context rather than dressed up as a settled verdict.

About the illustrations

Article visuals are synthetic and labeled as such: data cards and charts drawn by code from the product's own aggregate numbers, plus generic format illustrations. They illustrate a format category, not the actual product, application, or skin result. You will not find before-and-after pictures, faces, breakout or irritation photos, or anything built from a reviewer's own photo here. I do not test products on my own skin, so it would be dishonest to show you a result, and a picture of a rare side effect would make a 1% number look like a 100% one. The reaction data stays a chart, drawn to scale.

Who runs this

This site is written and maintained by betterdxd, a cosmetics-loving developer and the editor of this site. I use code to line up aggregate review fields consistently, then add editorial judgment about what the numbers can and cannot support. My goal is to create practical context that helps readers ask better questions before buying.

What I never do

Start with a research question

Begin with the original 322,854-record rating-distribution report, then try the dry vs combination skin moisturizer guide, the PDRN vs hyaluronic acid data guide, the Anua vs Medicube PDRN comparison, or the soothing-redness review-poll guide.

Affiliate disclosure

Store links currently contain no affiliate tracking ID; this site earns no commission from them. Link configuration never changes the numbers, and an article telling you to skip a product is a perfectly good outcome here.