“Why does my campaign look profitable by ACOS but lose money after fees, refunds and promotions?”

This question cuts to one of the most common traps in Amazon advertising. A seller watches their Advertising Cost of Sale fall inside what feels like a safe range, keeps spending, and then reconciles the month to find the business is underwater. The problem is not the metric itself — it is what the metric cannot see.

Direct Answer

ACOS measures one relationship: ad spend divided by ad-attributed revenue. It does not know your Amazon break-even ACOS, because that number requires inputs — cost of goods, FBA fees, referral fees, refunds, and promotional discounts — that live outside the advertising dashboard entirely. A campaign can display a “good” ACOS while every sale it generates loses money on a contribution-margin basis. The fix is not to chase a lower ACOS in isolation; it is to calculate the break-even threshold first and then judge your ACOS against that threshold.

Break-even ACOS formula (manual)

Break-Even ACOS = (Revenue − COGS − Amazon Fees − Refund Reserve − Promotions) ÷ Revenue

Every variable in the numerator must come from real account data, not estimates. Only then does the ACOS figure in your campaign reports mean something actionable.

What to Verify

Before you diagnose your campaigns, confirm that you have clean, complete data for each cost layer. Missing even one layer produces a false floor.

Revenue layer

Confirm which revenue figure your ACOS report uses. Amazon attributes revenue to an ad click for a defined window; organic sales on the same order are counted differently.
Check whether your attributed revenue includes VAT or marketplace-collected tax. The Amazon Finances API (v2024-06-19) surfaces transaction-level financial events, including tax components, that the advertising dashboard does not disaggregate.

Fee layer

FBA fulfilment fees, storage fees, and referral fees vary by category, size tier, and season. Retrieve these from Seller Central’s Fee Preview report or from the Finances API financial event groups.
Fee changes mid-period silently shift your break-even threshold. If Amazon announced a fee adjustment during the measurement window, your historical break-even ACOS is not valid for current bids.

Refund and chargeback layer

Refunds reduce net revenue but appear on a delay. A product with a high return rate can show a healthy in-period ACOS that deteriorates when returns settle.
The Finances API RefundEventList provides the data needed to adjust revenue figures for returns. Reconcile refunds against the same attribution window you use for ad revenue.

Promotion layer

Percentage-off deals, coupons, and Subscribe & Save discounts reduce the effective revenue per unit. If a campaign drives traffic to a promoted listing, the ACOS denominator is inflated relative to your actual pocket revenue.
Retrieve promotion adjustments from the Finances API ShipmentEventList and AdjustmentEventList objects.

Diagnostic Sequence

Work through these steps in order. Each step either confirms a layer is clean or surfaces the specific gap distorting your ACOS interpretation.

Step 1 — State your margin target. Write down the maximum percentage of revenue you can afford to spend on advertising before the unit loses money. This is your break-even ACOS ceiling. Do not proceed to campaign data until this number exists.

Step 2 — Pull a complete fee schedule for the SKU. Use the Fee Preview tool in Seller Central or the Finances API to retrieve current referral and fulfilment fees for the exact ASIN. Verify the size tier has not changed.

Step 3 — Calculate net revenue per unit after fees. Subtract COGS and all Amazon fees from the selling price. If you run promotions, apply the average discount depth for the period you are measuring.

Step 4 — Apply a refund adjustment. Retrieve the refund rate for the ASIN over the same period. Reduce net revenue by that proportion to reflect what you actually keep.

Step 5 — Compute your break-even ACOS. Divide the remaining contribution margin by gross revenue. The result is the highest ACOS at which the product breaks even on advertising spend.

Step 6 — Compare your actual campaign ACOS to the threshold. A campaign ACOS below the break-even ceiling is contributing margin. A campaign ACOS above it is consuming margin. A campaign ACOS that equals the threshold is neutral — growth with no profit.

Step 7 — Check attribution window alignment. Ensure the revenue window used in your ACOS report matches the window you used to collect fee and refund data. Misaligned windows produce a structurally false comparison.

AdsPilot is designed to keep these layers — advertising efficiency, fees, refunds, promotions, and product economics — distinct as separate evidence streams, so that the break-even calculation is assembled from verified sources rather than estimates. The AdsPilot feature overview describes how the intended workflow structures profitability evidence alongside campaign data.

Unsafe Shortcuts

Using category benchmarks as a personal break-even target. Benchmark ACOS figures are averages across products with different cost structures. Your break-even threshold is a function of your specific COGS, fees, and return rate. A benchmark that reads as “safe” for a high-margin product is loss-making for a low-margin one.

Treating a single week of data as representative. FBA storage fees, promotional periods, and refund settlement patterns are not evenly distributed across a month. A week that excludes a mid-month fee debit or a batch of settled refunds will produce an artificially healthy break-even calculation.

Optimising bids before verifying the fee schedule. If your fee schedule has changed — due to a size-tier reclassification or a category fee update — the break-even ACOS you used to set previous bids is no longer valid. Bid changes made against a stale threshold may accelerate losses.

Relying solely on the advertising dashboard for revenue. The Sponsored Products and Sponsored Brands dashboards report ad-attributed revenue. They do not deduct fees, refunds, or promotions. Using dashboard revenue as the denominator in a contribution-margin calculation produces a figure that overstates actual returns.

The Amazon SP-API integration that AdsPilot is designed to use is intended to retrieve transaction-level financial events — including the fee, refund, and promotion data described above — to avoid exactly this class of error.

Decision Checklist

Use this checklist before acting on any ACOS figure in your campaign reports.

I have calculated a break-even ACOS for this ASIN using current COGS, current Amazon fees, and a refund-adjusted net revenue figure.
I have confirmed the fee schedule (referral fee, FBA fee, size tier) reflects today’s rates, not rates from a previous period.
I have retrieved refund data for the same attribution window used in my ACOS report and applied it to net revenue.
I have accounted for any promotional discounts active during the measurement period.
I have verified that the revenue figure in my ACOS report and the revenue figure in my break-even calculation use the same attribution window.
I have compared my campaign ACOS to the break-even threshold — not to a category benchmark — before making a bid decision.
If any of the above data is unavailable or stale, I have paused bid changes until clean data is available.

Data-quality gates of this kind — checking freshness, attribution alignment, and coverage before surfacing a decision — are a core part of how AdsPilot’s campaign intelligence workflow is intended to operate. See the Amazon seller knowledge base for related guidance on campaign evidence standards.

This article describes the intended completed AdsPilot workflow. Its practical scope depends on the connected Amazon account, marketplace, granted API roles and the evidence those sources actually provide.

Sources

Amazon Developer Documentation — Finances API v2024-06-19 Reference: https://developer-docs.amazon.com/sp-api/docs/finances-api-v2024-06-19-reference