The fee tier that turned six SKUs negative
A fourteen-brand agency, home and kitchen, $310k a month in blended spend. In August a size-tier change moved six SKUs into a higher FBA fee band. Ads kept running against them until the end of September. It is Case File 01, and it is the clearest argument we have for why true profit per ASIN needs an agent of its own.
Nobody was careless. The problem is that the failure is invisible to every view an agency actually looks at.
Why the dashboards stayed green
Spend was on plan. ACOS sat inside target, because ACOS is measured against revenue and revenue did not move. Sessions and conversion were stable. Account health was clean. Inventory was in stock. Every one of those is a real metric and every one of them was telling the truth.
The number that changed was contribution per unit after referral fees, FBA fees, storage and returns. That number is not on a dashboard because it is not in any single report — it has to be assembled from settlements, fee estimates, storage charges and return rates, per ASIN, and rebuilt whenever Amazon changes a tier.
Break-even ACOS is downstream of fees
The second-order problem is worse than the first. A break-even ACOS target is derived from margin. When fees rise and the target is not recomputed, the ads team goes on optimising toward a number that is no longer break-even. They hit their target and lose money doing it.
This is the argument for agents that hand work to each other rather than tools that sit beside each other. Margin recomputing a floor is only useful if Spend is obliged to bid under it. An ads-only tool has nothing to hand off to, and no way to know its target went stale. How that hand-off runs, minute by minute, is the shortest explanation of the difference.
What changed
$17,400 of spend redirected off negative-contribution SKUs, and blended margin up 2.1 points. Not because anything clever happened, but because the ads stopped pointing at products that lost money on every unit. The pause itself was staged rather than applied — the three permission modes decide which of those two happens on your account. If you want the same five weeks read back on your own catalogue, that is what a Decision Audit is.
Case File 01. Account shape and figures are illustrative and anonymised.