Most Amazon sellers keep overhead out of product-level maths on purpose. Software, VAs, accounting, agency retainers — none of it changes when you sell one more unit, so none of it belongs in a per-unit calculation. That reasoning is correct for a narrow class of decisions and quietly destructive everywhere else. A catalogue in which every SKU clears its contribution margin can still lose money every month, because contribution margin is the number that stops just before rent. This article separates the two questions sellers routinely collapse into one — “does this unit pay for itself?” and “does this SKU carry its share of the business?” — shows the arithmetic behind both, and maps which decisions need which number.
What actually counts as overhead for an Amazon seller?
The split that matters is not fixed versus variable in the accounting sense. It is whether the cost moves when the next unit sells.
Costs that move with the unit: cost of goods, referral fee, FBA fulfilment fee, storage, returns and refund costs, PPC, inbound freight, prep and inserts, per-order packaging, partner commissions tied to revenue.
Overhead: software subscriptions, staff and VA salaries, accountant, agency retainers, trademark and legal, business insurance, product research tools, office costs, and — the one that gets left out most often — the owner’s own salary.
There is a grey zone that causes more damage than the allocation method you eventually choose. Prep centre fees, insert printing and partner commissions arrive as one monthly invoice, which makes them look like overhead. They are not. They scale with volume, and they belong on the SKU that generated them. A variable cost smeared across the whole catalogue as overhead distorts every product except the one that caused it — and it usually flatters the exact SKU you were trying to evaluate.
Why is leaving overhead out of per-unit maths defensible?
Because for a genuinely marginal decision, it is the right answer. A $4,200 monthly overhead base exists whether you ship 3,000 units or 3,600. So when the question is “should I raise this keyword bid by fifteen cents,” the only costs that belong in the calculation are the ones that change as a result. Contribution margin is the correct tool, and loading an allocated share of the accountant’s fee onto that decision would produce a worse answer, not a better one.
The failure isn’t the logic. It’s scope creep. The same contribution figure then gets reused for reorder priority, price floors and keep-or-kill calls, where it is systematically the wrong tool — and where it always says yes.
What does overhead actually do to your per-SKU numbers?
Take a simplified four-SKU catalogue. Contribution per unit here is already net of COGS, Amazon fees, returns and advertising.
| SKU | Price | Units/mo | Contribution/unit | Contribution/mo | Revenue/mo |
| A | $12.99 | 1,600 | $1.45 | $2,320 | $20,784 |
| B | $49.99 | 260 | $6.40 | $1,664 | $12,997 |
| C | $14.99 | 700 | $1.05 | $735 | $10,493 |
| D | $24.99 | 1,100 | $3.20 | $3,520 | $27,489 |
| Total | — | 3,660 | — | $8,239 | $71,763 |
Every SKU is profitable. Overhead is $4,200 a month, so the business nets $4,039 — a 5.6% net margin on $71,763 of revenue. Now express that overhead as a recovery target:
- Per unit: $4,200 ÷ 3,660 units = $1.15 per unit
- As a share of revenue: $4,200 ÷ $71,763 = 5.9% of price
Apply each basis to the same four SKUs:
| SKU | Contribution/unit | Loaded, $1.15/unit basis | Loaded, 5.9%-of-price basis |
| A | $1.45 | $0.30 | $0.69 |
| B | $6.40 | $5.25 | $3.48 |
| C | $1.05 | −$0.10 | $0.17 |
| D | $3.20 | $2.05 | $1.74 |
Two things fall out. SKU C changes sign — profitable or loss-making depending on nothing but the allocation basis you picked. And SKU B, the premium slow mover that looked like the star of the catalogue at $5.25 a unit, loses a third of its apparent profitability the moment overhead is charged on turnover rather than volume.
Both tables reconcile to the same $4,039. Allocation does not create or destroy profit. It only decides who gets blamed for it.
Which allocation basis should you use?
Match the basis to what actually drives the cost.
| Overhead type | Examples | Allocate by |
| Driven by unit handling | VA order processing, 3PL admin, customer service | Units sold |
| Driven by turnover | Agency retainer, accounting, insurance, banking | Revenue share |
| Driven by catalogue size | Listing tools, photography, per-product compliance, research software | Active SKU count |
Most sellers have all three. The accurate approach is to split overhead into those buckets and allocate each on its own driver. If that is more precision than the decision warrants, use revenue share as the default and stay aware that it under-charges cheap high-volume SKUs.
One practical rule is worth more than the choice of basis: run the decision under two bases. If the answer holds under both, act on it. If it flips, the decision is not really about overhead — it is about something you haven’t measured yet, usually cash, storage or attention.
What happens if you cut the SKU that allocation made look bad?
SKU C loses ten cents a unit on the per-unit basis. Discontinue it and see what the business does.
| Before | After dropping SKU C | |
| Total contribution | $8,239 | $7,504 |
| Overhead | $4,200 | $4,200 |
| Net profit | $4,039 | $3,304 |
| Units | 3,660 | 2,960 |
| Overhead recovery per unit | $1.15 | $1.42 |
You removed a product that “lost money” and the business made $735 less. The overhead did not leave with it — the agency, the accountant and the software all still invoice the same amount. It simply redistributed onto the survivors, and the recovery target rose from $1.15 to $1.42.
Which puts SKU A at $1.45 − $1.42 = $0.03 a unit. It is now the next thing on the chopping block. Cut it and the recovery target jumps again, to $3.09, at which point SKU D turns negative too. Run the logic to its conclusion and you allocate yourself down to a single product.
An allocated loss is not a causal loss. The keep-or-kill test is not “is loaded profit negative.” It is: is contribution positive, and is there a better use for the cash, the storage and the attention this SKU consumes? A negative loaded figure is a prompt to look, not a verdict.
How does overhead change your break-even ACOS?
This is where the two numbers most often get confused, because both ceilings are real and they answer different questions.
Take SKU D: price $24.99, contribution of $5.50 per unit before advertising.
| Basis | Calculation | Break-even ACOS |
| Contribution | $5.50 ÷ $24.99 | 22.0% |
| After overhead recovery | ($5.50 − $1.15) ÷ $24.99 | 17.4% |
Run that SKU at a 20% ACOS and the numbers look fine: $5.00 of ad spend per unit, $0.50 of contribution left over. Charge the $1.15 of overhead recovery and the same unit is $0.65 underwater. Contribution-positive, portfolio-negative.
Neither ceiling is wrong. They answer different questions:
- 22% — “does the next click pay for itself on a SKU that is already running?” Yes, anywhere below 22%. Use this for tactical bid changes, defensive spend, and launches you are deliberately subsidising.
- 17.4% — “can this SKU hold this ACOS as a standing position and still carry its share of the business?” Only below 17.4%. Use this for the steady-state target you hold a mature SKU to.
Sellers who only ever compute the first one end up with a catalogue where every campaign is technically profitable and the P&L is flat.
Which decisions need which number?
| Decision | Use | Why |
| Raise a bid on an existing keyword | Contribution | Overhead doesn’t move with the click |
| Run a 7-day Lightning Deal | Contribution (floor: stays positive) | Temporary; the capacity is already paid for |
| Liquidate aging stock | Contribution plus storage avoided | COGS is sunk; the question is cash recovery |
| Which SKU gets the next reorder | Fully loaded | You are allocating scarce cash and capacity |
| Keep or discontinue a SKU | Fully loaded, then re-run the portfolio without it | Allocation is not causation |
| Price floor on a new product | Fully loaded | A new SKU carries its share, or an existing one carries it twice |
| Open a new marketplace | Fully loaded, plus the incremental overhead it adds | New VAT, compliance and accounting come with it |
| Hire a VA or sign an agency | Contribution created vs. overhead added | Straight payback question |
| Am I hitting my margin target? | Fully loaded | It’s the only figure that matches the bank balance |
How sellerboard helps
Separate the variable costs out first. Profit Analytics → Variable Expenses lets you build rules calculated as a percentage of sales, a fixed fee per order, a fixed fee per unit, or a percentage of ad spend — with conditions by product, marketplace, channel (FBA/FBM) and fulfilment status. This is where prep fees, inserts and partner commissions belong, so they land on the SKU that caused them instead of being smeared across the catalogue as overhead.
Then record true overhead properly. Profit Analytics → Expenses handles one-time, weekly and monthly costs, with categories, and lets you assign an expense to specific products and marketplaces where it genuinely belongs to them. Two options here matter more than they look:
- Amortize expense daily spreads a cost evenly across the period instead of dropping it as a lump. An annual subscription booked in one hit makes one month look terrible and eleven look better than they are.
- Assigning to multiple marketplaces means one shared cost is entered once rather than duplicated per region — no double counting, no forgotten updates when the amount changes.
Read it at both levels. At account level, Profit → Dashboard → Tiles → More expands the full breakdown, where Net Profit = Gross Profit − Expenses. At product level, Dashboard → Products → More shows the expenses attributed to that specific product. The gap between those two views is exactly the general overhead you have not assigned to anything — and that total, divided by units or revenue for the month, is your recovery figure. It is one line of arithmetic, done once a month, and it is the number most catalogues are missing.
Common mistakes
- Comparing a product-level profit figure against a business-level net margin target. They are not the same shape, and the SKU will always look better.
- Filing prep fees, inserts and per-unit 3PL charges as general overhead instead of variable expenses.
- Booking an annual invoice as a single-month hit rather than amortising it.
- Leaving the owner’s salary out of overhead entirely. The model then says the business is profitable when what you have actually bought is a job.
- Choosing an allocation basis once and never revisiting it as the product mix changes.
- Treating an allocated loss as evidence that a SKU should be discontinued.
- Loading overhead into clearance and liquidation maths, then holding stock that is quietly accruing storage fees.
FAQ
Isn’t overhead a sunk cost I should ignore? Fixed and sunk are different things. Sunk means spent and unrecoverable — last year’s photoshoot. Fixed means recurring and avoidable over a longer horizon — the agency retainer you can cancel with 30 days’ notice. Sunk costs stay out of every decision. Fixed costs stay out of the marginal decision and belong squarely in the structural one.
What is the single best allocation basis? There isn’t one. Revenue share is a reasonable default. Driver-matched buckets are more accurate. The discipline of checking whether the answer survives a change of basis matters more than the choice itself.
Should overhead be in my break-even ACOS? Not in the contribution break-even, which is the correct ceiling for marginal spend. Yes in the target ACOS you hold a mature SKU to. Compute both and know which one you are quoting.
How often should I recalculate the recovery figure? Monthly, and any time overhead or product mix moves by more than about 10%. A recovery target built on last quarter’s volumes is worse than no target, because it looks precise.
Does this matter if I only sell one product? More, not less. Allocation is trivial — 100% of overhead sits on that SKU — but there is no second product to absorb the shortfall, so the gap between contribution margin and real profit is the entire business.
Contribution margin tells you whether a unit pays for itself. Overhead tells you whether the business does. Sellers who only ever look at the first number can spend a year optimising a catalogue in which every product is profitable and the company is not.
