Ask an account manager which of their SKUs lose money at current ad spend and you will usually get a considered guess. The guess is wrong about a third of the time, because the data required to answer it properly was never assembled.

What goes into the model

Start at net revenue — gross sales less returns and promotional discounts — and subtract, per unit:

  • Landed cost of goods, including freight and duty, not the ex-works price.
  • Amazon referral fee at the correct category rate.
  • Fulfilment fee at the correct size and weight tier — verify this rather than assuming it.
  • Storage cost, apportioned by the units' actual dwell time rather than averaged flat.
  • Returns processing and the value of units that cannot be resold.
  • Advertising spend attributed to that SKU, including a share of campaigns covering multiple ASINs.

What is left is contribution margin: the money the SKU generates toward fixed costs. It is the only number that should drive a budget decision.

The two costs everyone forgets

Returns

In categories with double-digit return rates, the difference between gross and net revenue is large enough to reverse the ranking of which SKUs are worth advertising. Returns also arrive on a delay, so a model that nets them off the same month understates the cost of recent growth.

Storage

Storage is charged by volume and escalates with age. Averaging it across the catalogue hides the fact that your slowest SKUs are carrying almost all of it.

Tier, do not enumerate

A per-SKU margin model is useful for analysis and unusable for daily operations. Group SKUs into three or four tiers by contribution margin and set an advertising efficiency target per tier. That is granular enough to stop misallocation and simple enough that the team will actually apply it.

What changes once you have it

Budget stops being allocated by revenue and starts being allocated by margin headroom. Some high-revenue SKUs get less spend and some quiet ones get considerably more. Expect total ad-attributed revenue to dip in the first month and contribution to rise from the second.

That dip is the moment the model gets abandoned in most organisations. Agree in advance that it is expected.