ACoS Change Explainer

Why did my Amazon ACoS change?

Before changing bids, separate the ACoS movement into click cost, purchase rate and sales per purchase. Type comparable totals or import two reports to see which components changed and where to investigate.

Earlier period
Import earlier period report

CSV, TSV or XLSX · up to 10 MB and 100,000 rows · read only in this tab.

Later period
Import later period report

CSV, TSV or XLSX · up to 10 MB and 100,000 rows · read only in this tab.

Comparison basis
One currency for both periods, e.g. USD, GBP, EUR.

For example, compare the same accounts in the US marketplace, using 7-day purchases and 7-day sales in both reports. A 7-day window counts purchases for seven days after each click. Wait until the full window shown in your report has passed for the latest date, then allow for reporting updates. Recent totals can still change.

Use purchases, not units. Avoid mixing click-plus-view conversions with click-only metrics. Search Term Report totals describe that report’s coverage, which may differ from campaign totals.

Method and limits

ACoS = spend ÷ attributed sales = CPC ÷ (purchase rate × sales per purchase). Purchase rate = purchases ÷ clicks; sales per purchase = attributed sales ÷ purchases.

The bridge replaces earlier CPC, purchase rate and sales per purchase with later values in all six possible orders, then averages each factor’s contribution. The contributions add to the observed ACoS change. This splits interactions fairly across the three factors; it does not prove what caused them.

Account averages include changes in traffic, campaigns and products. This version does not isolate those mix changes. Compare one consistent product or campaign scope when possible. Different period lengths affect volume comparisons, so use equal lengths for spend and sales comparisons. Zero sales makes ACoS undefined; zero purchases prevents the three-factor bridge.

Can this tell me whether a price change caused the problem?

No. Price, discounts, competition, placements and traffic mix can all change these averages. Use the breakdown to choose what to inspect, then compare product-level evidence or run a controlled test.

Definition: Amazon’s guide to ACoS.

Same math, for developers: free JSON API · API specification. Calling the API sends the values you provide to the server. This page never calls it.

Explore all free tools

Learn the decision behind the numbers.

Work through the example, then apply this tool to your own reports.