Breaking one over-merged listing into the four products it really was

Someone had put four products under one parent listing on the reasonable theory that a shared review count helps everything. The four were a vent mount, a desk stand, a bike mount and an adhesive grip — related only in that a phone touches them.

The merge cost more than it returned. A search result shows one child from a family, and it is usually the best-selling one — so three of the four products were represented in search by a photograph of the fourth, and their own pages were reached mostly by accident. The reviews, pooled as intended, mixed complaints about adhesive with praise for a bike clamp, leaving every shopper reading feedback about a product they were not buying. And four unrelated products are not a variation family at all — that is a listing-policy problem waiting to be enforced.

We separated them and let each one be specific.

• Four standalone listings, each with its own title, image stack and keyword set built for one situation.
• Review equity handled deliberately — the parent retained by the strongest performer so its history was not thrown away, the other three relaunched with a review-generation plan.
• Campaigns rebuilt per product, which exposed that two of the four had been quietly subsidising the other two inside one blended ACoS.
• Cross-links between the four in A+ comparison modules, keeping the merchandising benefit the merge was supposed to deliver.

Revenue grew 127% from the same four products — $71k to $161k a month — with no new SKUs and no price change.

Page-one keywords went from 38 to 210, and the arithmetic behind it is simple: four specific listings index for four vocabularies where one generic listing indexed for a compromise between them. Conversion rose 58% because shoppers now arrive at a page about the thing they searched for, and ACoS fell fourteen points once each product's advertising was judged on its own economics.