Two numbers in Amazon reporting both look like “sales”: ordered product sales from seller reports and attributed sales from advertising reports. They are often placed next to each other, subtracted from each other, or added together. Each of those moves can produce a figure that looks precise and means very little. This guide explains what each number measures and why they cannot be treated as pieces of the same total.
Two different measurements
Ordered product sales
Ordered product sales come from seller sales and traffic reporting. They add up units ordered multiplied by the item price for orders placed in the period, regardless of how the shopper found the product: search, an ad, a link, or a repeat visit. They are recorded when the order is placed. They are not settled revenue, they are not net of fees, and they do not reflect refunds that happen later.
Attributed sales
Attributed sales come from advertising reports. They are the sales that an ad report credits to an ad interaction, such as a click, when the purchase happens within a defined attribution window. Depending on the ad product, they can include purchases of other products from the same brand, not only the item in the ad. They answer the question “which sales does the advertising system give ads credit for?”, not “how much did the store sell?”
Why they are not interchangeable or additive
Overlap
An order placed after an ad click usually appears in both reports: once as ordered product sales and once as attributed sales. Adding the two counts that order twice.
Attribution windows
The window decides how long after an ad interaction a purchase can still be credited. A click near the end of one month can lead to a purchase early the next month. The seller report places that order in the new month; an advertising report can credit it to the day of the ad interaction instead.
Brand-level credit
If a shopper clicks an ad for one product and buys a different product from the same brand, the advertising report can count that as attributed sales, while the advertised product’s seller sales do not change at all.
Timing and late data
Attributed sales keep arriving for days after the click, so recent periods grow over time. Seller order data also updates as orders move through their statuses. Two reports pulled at different moments describe different stages of maturity.
Refunds, cancellations and settlement
Neither figure is net revenue. Settlement reports group money by settlement period and deduct fees, which puts the same order on a third timeline.
The illustrative table below follows one fictional order through three reports:
| Report (fictional order) | Date it lands on | Amount | Why |
|---|---|---|---|
| Advertising report | March 30 (day of the ad click) | 40.00 | Purchase fell inside the attribution window |
| Seller sales report | April 2 (day of the order) | 40.00 | All orders count, whatever brought the shopper |
| Settlement report | Mid-April settlement period | 27.80 | Net of fictional referral and fulfillment fees |
The same purchase lands in two months and appears as three different amounts. None of the reports is wrong. They answer different questions.
What goes wrong with combined figures
Here is a fictional month for one product, with all figures in the same currency and marketplace:
| Measure (fictional month) | Value | Comment |
|---|---|---|
| Ordered product sales | 12,000.00 | Seller report |
| Attributed sales | 4,800.00 | Advertising report |
| Ad spend | 1,200.00 | Advertising report |
| ACoS | 25.0% | 1,200 ÷ 4,800, one source |
| Spend ÷ ordered product sales | 10.0% | Two sources; needs explicit definitions |
| Ordered + attributed | 16,800.00 | Wrong: double-counts overlapping orders |
| Ordered − attributed (“organic”) | 7,200.00 | Unreliable: windows, timing and brand credit differ |
The subtraction in the last row is popular because it promises a clean split between “ad” and “organic” sales. It cannot deliver one. Some attributed sales fall outside the seller report’s period, some belong to other products, and some would have happened without the ad.
Defining a total-sales ratio
Spend divided by total sales, often called TACoS, can be useful for seeing how advertising cost relates to the whole business. If you calculate it, write down:
- Which sales: ordered product sales, gross, for which products and marketplace.
- Which spend: which ad products and campaigns, in the same currency.
- Which period: same dates and time zone, with the most recent days marked provisional.
- What is excluded: refunds, cancellations, other products credited through brand-level attribution.
TACoS = ad spend ÷ ordered product sales × 100% is a different metric from ACoS and should always be labeled separately.
Data permissions come first
Combining live advertising data with seller data is not only a math question. Amazon’s advertising data policies restrict combining Amazon Ads data with other data sources, or sharing it with third parties, without Amazon’s written approval. KEYXE treats seller data as another source for this purpose and keeps live joins between the two turned off. Neither integration is live today. In the public demo, synthetic seller and advertising datasets appear side by side in separate workspaces, so the definitions can be compared without merging real data. The Data Foundation page and the Ads API and SP-API integration pages describe the design.
Common mistakes
- Adding ordered product sales and attributed sales into one “total”.
- Subtracting attributed sales from ordered sales and calling the rest “organic”.
- Labeling spend ÷ total sales as ACoS.
- Comparing an advertising period that is still maturing with a seller period that is complete.
- Treating ordered product sales as revenue after fees and refunds.
Try it
Open the seller workspace and the advertising workspace of the synthetic demo in two tabs and compare the period totals and their “Explain metric” notes. The glossary defines attribution windows, settlement and ordered product sales. Related guides: What Is ACoS? and Inventory Data Freshness.