The 26× problem: why parent-level reporting lies to you

Children on a shared parent receive identical traffic and convert at wildly different rates. Parent-level reporting averages them into one number: a number that describes no child, and hides where the margin lives.

The standard dashboards aggregate at parent level because that is where the report is easy to read. The problem is that the decisions worth making happen at the child.

Inside one five-variation family we measured a 26× velocity spread between the best and worst child: same listing page, same traffic, same price band. One child was doing almost all the work; the parent-level report made the family look uniform.

Advertising at parent level spreads budget evenly across the spread: the weak children got clicks they couldn't convert, and the strong child got fewer than it had earned.

The variation that converts is the one whose listing answers the query. We isolate it into its own campaign: its own bids, its own budget, its own negatives. The rest of the catalogue builds baskets, not acquisition costs.

Concentration is not neglect of the catalogue: it is where the margin is spent deliberately instead of evenly.

Child-level performance review is part of the weekly cadence, alongside search-term harvesting and negation. Titles, bullets, and backend keywords refresh per child as ranking data accumulates.

If the 26× problem sounds like your account, the fix starts with a question the parent-level report can't answer: which child is actually earning the margin?