The flavor shown first was the one nobody wanted

A variation family concentrates a brand's reviews and ranking into one listing, which is exactly why it is worth building. What the brand had not considered is that a shopper does not arrive at the family — they arrive at one specific child, and something has to decide which. In this account the family had opened on the original flavor since launch — the one carrying four thousand reviews and, by now, ninth of the twelve by units — and nobody had ever checked whether the front door still matched the demand.

Two discontinued flavors were also still attached to the family, showing as unavailable in the swatch row. So the first impression of the brand's best-performing asset was a recipe most shoppers did not want, sitting next to two they could not buy — and every paid click paid full price for that first impression.

We made the family open on what people actually buy.

• Flavor-level demand mapped against the featured child, which is the comparison nobody had run: the default was ninth of twelve by units.
• The family cleaned up so the default stops being the wrong door — a seller cannot pick the featured child, it follows sales, so the discontinued children came out, advertising came off the fading original, and the current best sellers got the traffic that makes one of them the default.
• Discontinued children removed from the family so the swatch row contains only buyable options.
• Advertising pointed at flavor children directly, matching the flavor named in the search term instead of dropping every click on the parent's default.

Conversion went from 8.9% to 14.1% and revenue grew 55%, without changing a recipe, a price or the number of flavors on offer.

The cost of the old arrangement shows up cleanly in ACoS, which fell eight points: the ads were never badly targeted, they were just landing shoppers one click away from the thing they had searched for. Four fifths of visits now open on a flavor that is genuinely in demand, against under a third before.