Skip to the argument
nouyu
Sign in

Not the best.
The best for you.

nouyu reads what people with skin like yours actually said, and tells you what that means for you.

Every review out there was written about somebody else's skin.

Hers. Not yours. Which is why a five star average can be completely true and still tell you nothing.

So we start with yours.

What we take is a reading, never a diagnosis. Points on a surface, described and never named.

What a review leaves out

A review tells you what happened. Almost never who it happened to.

We labelled thirty store reviews by hand and counted what was actually written down.

0%

describe a result on the skin

0%

mention sensitivity

0%

mention skin type

0%

mention age, climate, or the other actives in use

That is the ceiling of the format, not of the method. No extractor fixes it, because the information was never written down.

nouyu research. Thirty store reviews labelled by hand.

One real product

Six thousand, then five hundred.

What survives each time you ask a harder question of the pile.

6,181

reviews found

2,271

carry any evidence at all

863

usable once quality is weighed

527

relevant to one specific profile

That product's public score is 4.21 stars, 78% of them four or five. We keep it in view on purpose. It is what the funnel exists to be compared against.

Aggregates computed on Amazon Reviews'23 (McAuley Lab, UCSD). No review text is reproduced.

47

efficacy milestones our research proposed, across six ingredients

25

survived, each with its citation, year, sample size and the part of the body it was measured on

23

fell, every one on attribution: the trial tested a finished formula, not the ingredient

A context, not a verdict

There is no best product. There is evidence relevant to a profile, and evidence that is not.

So nouyu never ranks products for everyone. A match is how relevant the evidence is to your profile, not a verdict on the product. What a review says is sorted into thirteen closed dimensions. Two of them, irritation and redness, are risks, and they are read in the opposite direction to the rest.

How close two people are, by weight

Skin type18%
Sensitivity18%
Acne tendency14%
Concerns14%
Goals12%
Actives already in use12%
Ingredient sensitivities8%
Climate4%

The weight of a single piece of evidence is three numbers multiplied together: confidence, relevance and quality. Multiplied on purpose. A zero in any one of them cancels the evidence entirely.

Your photo is yours.

It stays in your account so you can compare over time. Anthropic's AI reads it for nouyu, and you can delete it whenever you like.

Nobody else ever sees it.

Not your face, not your name, not your routine.

Yours joins everyone else's.

That is the whole trade. What you choose to share, how a product worked on your skin, is the only thing that helps the next person.