‹ All articles

How an outfit gets its colours

You pick one piece and the app builds the rest — and somehow the colours belong together. That's not taste, and it's not magic. Here's the actual recipe, including the part where a famous formula failed our test read.

Pick a piece in Sanago — say the olive skirt — and the outfits that come back tend to just... work. The colours sit together calmly. Nothing screams.

It would be easy to let you believe there's a secret palette behind that, some proprietary colour wisdom. There isn't. There's a short recipe, and every step of it is either backed by research or honestly labelled as craft. Here's the whole thing.

The piece near your face does the talking

One slot in every outfit matters more than the rest: whatever sits closest to your face — the top, the dress, the jacket if it's open at the collar.

That's the slot where your own colours go. Not "autumn" colours, not a stylist's hunch — the shades that stand clearly apart from your skin's depth, lighter or darker, so they read crisp instead of washing out. That's the one colour effect near a face with real research behind it, and it's the heart of how the app reads your colouring. [B] (We wrote up the whole story — and why we refuse to sort you into a season — in Why we don't do color "seasons".)

So when a look leads with navy against fair skin, or butter yellow against deep skin, that's the engine putting contrast where contrast counts.

Neutrals are the glue

The rest of the outfit — trousers, skirt, shoes, bag — leans on neutrals: the blacks, creams, camels, greys, denims.

We'll be straight with you: this part isn't a lab finding. It's craft — the oldest trick in every well-dressed person's book. Neutrals don't compete, so one interesting colour can be interesting in peace. We built the engine that way on purpose, and we'd rather call it what it is than dress it up as science.

The famous formula gets a floor, not a crown

Now the part we find genuinely funny.

There is real science on which colour pairs people find harmonious. A respected lab spent years building a formula for it — show it two colours and it predicts how pleasing people will find them side by side. On colour patches, it works. [A]

Then, in 2025, researchers tried it on actual two-piece outfits — real tops with real bottoms, judged by real people. The formula's predictions were worth almost nothing. Essentially a coin flip. [C] The tidy patch world and the world of worn clothes turn out to be different places — fabric, proportion, and bodies get a vote.

So here's what we do with it: the formula never picks our outfits. It's a floor, not a crown — a last check that rules out a genuine clash between two garments, never a judge that crowns a "best" combination. Using it as more than that would be pretending a patch result transfers to your closet. It doesn't, and we checked.

Nothing you own gets banned

One more rule, and it's the kindest one: a colour that sits quietly near your face isn't wrong — it's just better employed a little further down. That mocha top that blends into your skin? As trousers it's lovely. The engine doesn't throw colours out; it places them.

The receipts

Sometimes a look in the app carries a small line underneath — "Navy near your face: a strong contrast with your skin." Those lines only ever appear when there's research standing behind them. When a look has no line, that's not an oversight. It's the app declining to invent a reason.

That's the whole recipe: your colours where they count, neutrals doing the quiet work, one humbled formula guarding against clashes, and nothing banned. Less mysterious than a secret palette — but every step of it is one we can defend.

Sources

A note on the letters in brackets: they're our evidence grades. [A] means replicated, peer-reviewed research; [B] a single good study; [C] a reasonable idea that hasn't been tested; [D] folklore — which the app is never allowed to state as fact. [B‡] marks a published finding that later failed to replicate.