Blog · Board and management reporting
Amazon marketing analytics compares advertising spend with sales, refunds, seller fees and product costs. Follow a seller-export reconciliation that separates attributed revenue from the contribution remaining on matched sales.
Amazon marketing analytics compares advertising spend with sales and the contribution left after refunds, product costs and seller charges. Start with a defined sales population, then reconcile the exports before deciding which campaign deserves more budget. Advertising-attributed revenue and recorded seller revenue need separate labels because they answer different questions.
Save the campaign spend export, order-line export, refund adjustments, seller charges and product-cost file. Record the marketplace, currency, reporting dates and extraction cutoff. Use USD throughout this example; exclude sales tax collected for others from the merchandise-sales base.
The campaign report tells you what advertising receives credit for. The seller records tell you what was sold, refunded and charged. Amazon Ads defines advertising cost of sales and ROAS using advertising spend and advertising revenue. Those ratios do not establish the contribution on accounting transactions. The broader distinction belongs in marketing ROI versus ROAS.
Do not assume a campaign report contains order identifiers. Use an existing, documented mapping when available. Otherwise, compare campaign attribution separately and calculate contribution at a defensible SKU-period level. An allocation across campaigns is an allocation, even when its totals reconcile.
For the worked example, an illustrative US home-goods seller supplies a mutually exclusive mapping of product sales groups to campaigns. Each group appears once. The mapping organizes the comparison; it does not prove advertising caused the purchases.
The seller compares an order cohort from September 1 through September 30, 2026, with known adjustments through October 23, 2026. The following grouped rows summarize the included order lines. Refunds and costs are positive deductions; fees are net of any fee credits. Product cost already reflects the seller's treatment of returned inventory.
In a worksheet named Lines, put these inputs in columns A through G:
| Campaign | SKU group | Gross merchandise sales (USD) | Refunds (USD) | Seller fees (USD) | Fulfillment (USD) | Product cost (USD) |
|---|---|---|---|---|---|---|
| Kitchen | K1 | 6,000 | 300 | 900 | 600 | 2,100 |
| Kitchen | K2 | 4,000 | 200 | 600 | 400 | 1,400 |
| Storage | S1 | 6,000 | 900 | 900 | 900 | 2,400 |
| Storage | S2 | 4,000 | 600 | 600 | 600 | 1,600 |
| Total | 20,000 | 2,000 | 3,000 | 2,500 | 7,500 |
Use actual posted charges instead of applying a blanket fee percentage. Amazon's selling-fee guide distinguishes referral fees from selling-plan charges and explains that referral fees vary by category. Keep fixed selling-plan costs outside this variable contribution comparison, or show a separate, clearly labeled overhead allocation.
Add net sales in H and contribution before advertising in I. Enter these Excel formulas in row 2 and fill through row 5:
H2 = C2-D2
I2 = H2-SUM(E2:G2)
The first Kitchen row therefore leaves $5,700 of net sales and $2,100 before advertising. The next section groups those calculated rows without repeating any charge.
The campaign export records $2,000 of advertising spend for Kitchen and $2,000 for Storage for the same comparison period. Keep this spend in a separate campaign table. Copying the full campaign expense onto every SKU row would duplicate it.
Create a summary worksheet with the layout below. Percentages are rounded to one decimal place; sales multiples are rounded to two decimals. Calculations retain unrounded values.
| Campaign | Net sales (USD) | Before-ad contribution (USD) | Ad spend (USD) | After-ad contribution (USD) | Contribution margin | Net sales / spend |
|---|---|---|---|---|---|---|
| Kitchen | 9,500 | 3,500 | 2,000 | 1,500 | 15.8% | 4.75 |
| Storage | 8,500 | 1,500 | 2,000 | -500 | -5.9% | 4.25 |
| Total | 18,000 | 5,000 | 4,000 | 1,000 | 5.6% | 4.50 |
With campaign names in column A and spend entered in D, use:
B2 = SUMIFS(Lines!$H$2:$H$5,Lines!$A$2:$A$5,A2)
C2 = SUMIFS(Lines!$I$2:$I$5,Lines!$A$2:$A$5,A2)
E2 = C2-D2
F2 = IF(B2=0,"",E2/B2)
G2 = IF(D2=0,"",B2/D2)
Fill through the Storage row. Microsoft documents SUMIFS as adding values that meet specified criteria. Here, campaign is the criterion, and the summed ranges contain already reconciled net sales and costs.
The last column is a matched-net-sales multiple, not native advertising ROAS. Both campaigns have gross merchandise sales of $10,000 against $2,000 of spend, a gross-sales multiple of 5.00. Their refunds and variable costs differ enough to leave opposite contribution results. For the distinctions between cost layers, use gross margin, contribution margin and net margin.
The source bridge must close:
Gross sales = Refunds + Seller fees + Fulfillment + Product cost + Advertising + After-ad contribution.
Here, $20,000 = $2,000 + $3,000 + $2,500 + $7,500 + $4,000 + $1,000. Net sales also reconcile independently: $20,000 less $2,000 equals $18,000.
On the summary sheet, the total margin is =SUM(E2:E3)/SUM(B2:B3), giving 5.6%. Compare each input-column total with its filtered source export. Retain an exclusion list for unrelated sales, non-order charges and adjustments outside the cutoff. Do not compare the included merchandise subtotal directly with a bank deposit; settlement timing and other charges can differ.
If an additional $500 refund belongs to this cohort, with other costs unchanged, net sales fall to $17,500 and contribution falls to $500. The revised margin is $500 / $17,500 = 2.9%. Refresh the cohort through the same adjustment age before treating its result as final.
Storage needs a review of returns, product economics and fulfillment costs before a budget increase. Kitchen leaves positive contribution under the stated cost scope, but that alone does not establish that additional spend will produce the same result. Check inventory availability and whether campaign assignments remain comparable.
Keep new-customer measurement separate: customer acquisition cost needs an identified new-customer denominator, which this sales table does not provide. If the business also buys app traffic, mobile marketing analytics compares acquisition cohorts rather than seller SKU economics.
Covirage is not affiliated with Amazon or Microsoft.
Bring a sample of your own seller exports and campaign-spend file to Covirage, with the mapping and reporting window documented. Covirage maps the columns, its tools compute and check the contribution comparisons, and the external AI model explains the results with every figure traced to its rows. Create a free Covirage account; account creation takes about a minute, after which you set up your business and bring your data.
See customer profitability analysis for the supporting business context.
No. ROAS compares attributed sales with advertising spend. It does not deduct product costs, refunds, fulfillment or seller fees. A campaign can show strong sales per advertising dollar while leaving little contribution after those costs. Use a separate contribution calculation with a documented sales base.
Do not assume that it can. Check the identifiers and reporting detail in your exports. If you cannot establish a defensible campaign mapping, calculate contribution for a product and period, keep campaign attribution separate, and label the remaining uncertainty rather than inventing a match.
Choose whether the comparison follows an order cohort or a posting period. For an order cohort, attach known refunds to the original included sales through a stated cutoff. For a posting-period view, include that period's postings and show any refunds relating to earlier orders separately.
Include the net sales, product costs and variable seller charges associated with the included transactions, then deduct the corresponding advertising spend. State the treatment of fulfillment, refunded fees and returned inventory. Contribution excludes overhead unless you explicitly allocate it, so it should not be labeled net profit.