The Skan Score, explained honestly
The Skan Score is a single number with real inputs behind it - fit, shape, and how a piece sits with your taste. Here is what goes into it, how recommendations improve over time, and where the accuracy stands today.
A number with reasons behind it
The Skan Score is not a like button. It combines how an item fits your proportions, whether its shape balances your features, and how it sits with the looks you have already kept.
That is why the same jacket can score differently for two people - the render is personal, and so is the score.
Four things behind every score
No single factor decides the number. The weighting shifts by category - fit matters most for clothing, shape balance for eyewear.
Proportion fit
How the item's real dimensions sit against your measured proportions - the largest input for clothing and accessories.
Shape balance
For eyewear and silhouettes, whether the shape complements your face and frame rather than working against it.
Style coherence
How well the piece sits with the looks you have kept before, so the score reflects your taste, not a house style.
Context of the look
Whether the item works with what it will be worn alongside, judged on the same photo rather than in isolation.
The more you skan, the sharper it reads you
Recommendations start broad and narrow as you use the app. Every item you try, keep or pass on is a signal - so the feed moves from generic to genuinely yours without you filling in a style quiz.
- Keeps are the strongest positive signal
- Passes quietly tune the feed away from a fit or shape
- Your data trains your recommendations, not a shared model of everyone
Where accuracy stands today
The Skan Score is a strong guide, not a guarantee. It is most accurate for eyewear and structured clothing, where geometry is well defined, and improving for soft, draping fabrics and fine jewellery.
Category coverage grows as more brands join. When an item falls outside what the engine models well, we would rather show a lower-confidence score than pretend to a precision we do not have.
Get a score that learns you
Skan yourself and start building the signal that makes every recommendation sharper than the last.