When the color is the search term and the swatch is called “Willow”

The brand's color names were part of its identity: Willow, Ember, Harbour, Clay. On its own website that works, because the customer is already looking at the swatch. On Amazon the color name is a search term — “sage green midi dress” and “rust wrap dress” carry real, purchase-ready volume — and the word “sage” appeared nowhere in the brand's catalog.

The damage was concentrated in a place nobody looks. A variation family's child listings each own their color field, and only the parent's title had been written with any care. So twenty-seven of thirty-one colorways existed as inventory, took up warehouse space, and were reachable only by a shopper who had already found the style and then opened the dropdown.

We kept the brand's names and added the shopper's.

• Color vocabulary mapped from search data — every swatch matched to the one or two plain-language color terms shoppers actually type for it.
• Two-part color values — the variation now reads “Willow / Sage Green”, so the poetic name survives and the searchable one indexes.
• Child listings written individually, each with its color term in the title, the first bullet and the backend fields, instead of inheriting the parent.
• Campaigns split by colorway for the twelve colors with meaningful volume, so bids follow demand rather than averaging across a style.

Sessions went from 71k to 105k a month, and revenue grew 70% on a catalog that did not gain a single new style.

The mechanism is unglamorous: twenty-nine of thirty-one colorways now appear for the color a shopper typed, against four before. Color-qualified searches went from 6% of revenue to 39%, and because those queries name both the garment and the color, they convert well above the style-level terms the brand had been paying for.