Amazon PPC Metrics That Matter: And the Ones That Mislead

Last updated October 23, 2026 · First published November 5, 2026 · By Chad Norris

Disclosure: the author runs an Amazon advertising agency (Tide Flow) that reports these metrics to clients and builds Amazon software (AdSpark). Nothing below depends on those products. See the disclosure.

Most Amazon PPC reports lead with ACoS and stop there. The short answer: judge every campaign against that product’s break-even ACoS, watch total sales and TACoS direction to see whether ads are growing the business, and treat click-through rate, impression share and raw ROAS as supporting evidence, not goals. Each metric below answers one question well and misleads on another.

The core definitions

Amazon’s advertising cost of sales guide (read October 2026) defines ACOS as ad spend divided by ad revenue times 100, and ROAS as ad revenue divided by ad spend, with the example that $50 spent for $100 of sales is a 50% ACOS or a ROAS of 2. Amazon’s Sponsored Products best-practices guide defines impressions, clicks, sales, cost per click (total spend divided by total clicks), new-to-brand orders and ROAS (total sales divided by total ad spend).

CVR and CTR are computed the standard way; Amazon’s console may label or round them differently, so check the column definition in your account. TACoS is not defined in the Amazon pages we read; it is computed from advertising reports and Business Reports, as covered in our ACoS and TACoS guide.

The ones that matter

1. ACoS against break-even, per product

Amazon says there is no single good ACOS: it depends on industry, company size and campaign frequency, and break-even ACOS is tied to profit margin. So the useful reading is the gap between each product’s ACoS and its own break-even. An account at 30% ACoS is fine on a product that breaks even at 45% and losing money on one that breaks even at 20%.

2. Conversion rate, by query type

CVR tells you whether traffic is the problem or the page is. Compare branded and category terms separately: they convert very differently, and pooling them hides both. If conversion is far below the market on the same queries, no bid change will fix it; see the funnel-shape method in the Search Query Performance guide.

3. TACoS direction, over weeks

TACoS answers a question ACoS cannot: whether ad spend is growing faster than total sales. Judge it over three to four weeks, not days, because sales timing and attribution make daily TACoS noisy. Falling TACoS with steady or rising total sales is the healthy pattern. Rising TACoS with flat sales means you are paying more for the same business, or inventory or price changed. Published “healthy TACoS” ranges differ across sources, so use your own history as the benchmark.

4. Spend and sales by campaign purpose

Report discovery, harvest and branded campaigns separately, each against its own target. Discovery campaigns are allowed to run hotter than harvest campaigns; the test is whether discovery produces terms that later convert profitably. See how campaigns are structured for the segments.

The ones that mislead

Blended ACoS

One account-wide figure averages products with different margins. A strong branded campaign can hide a category campaign that loses money on every sale. Always show ACoS by product and by campaign segment.

ROAS in a report that also shows ACoS

Because they are inverses, showing both adds no information and invites a client to compare a 4.0 ROAS with a “good” ACoS benchmark and conclude the wrong thing. Choose one for the report. If you use ROAS as a goal, set it from break-even: break-even ROAS is 1 divided by break-even ACoS, so a 45% break-even ACoS is a break-even ROAS of about 2.2.

CTR taken alone

A high CTR on a broad automatic campaign can be curiosity traffic. Use CTR to compare creatives on the same query, not to rank campaigns.

Recent-days ACoS

Orders credit to a click within the attribution window. Third-party documentation puts the Sponsored Products seller window at 7 days (we could not read Amazon’s own statement; confirm for your account). Sales therefore keep arriving for days after the click, so ACoS for the last few days looks worse than it will end up. Leave the most recent week out of cut decisions, or at least flag it.

Impression share as a goal

Amazon launched top-of-search impression share on September 22, 2021: the share of top-of-search impressions your campaign received out of those it was eligible for, for Sponsored Products and Sponsored Brands, with a 90-day lookback. For Sponsored Products it covers the first page, and Amazon suggests adjusting bids by placement to raise it. Raising it costs money. A campaign at 40% top-of-search share on terms that convert below your break-even is just spending faster on a losing placement. Use it to diagnose whether lost sales are a visibility issue, and only buy more where conversion supports it.

Budget-limited campaigns

A campaign that runs out of budget early in the day shows a flattering ACoS because it only catches the cheapest hours or placements, and its share of the market is smaller than it looks. Amazon’s budget guide says a campaign out of budget stops showing until midnight, and recommends raising the budget of converting campaigns that run out or moving budget from campaigns that are not spending.

New-to-brand and “total sales”

Amazon’s console includes metrics such as new-to-brand orders, and a “total sales” column that can include sales of other products from your brand that the shopper bought after the click. Those can make ad performance look better than direct product sales; check which column your report uses when you compare ACoS between periods or accounts.

Illustrative reading

An illustrative month for one product ($40 price, break-even ACoS 45%, target 30%):

The blended 32.5% looks close to the 30% target. It hides $700 spent above break-even and $1,100 spent over target. With these numbers, the auto campaign is an investment that should be capped and judged by the terms it graduates, not by ACoS.

How an agency should report

A useful report fits on one or two pages:

1. Spend, ad sales, ACoS and TACoS direction for the period and the previous period, on mature dates.
2. ACoS by product against that product’s break-even, with the margin assumption stated.
3. A segment table like the one above.
4. What changed, why, and what the next test is.
5. Anything flagged: stockouts, price changes, lost Buy Box, listing changes.

If the agency’s report shows only blended ACoS or only revenue, ask for the segment view. For questions to put to an agency before you hire one, see how to choose an Amazon PPC agency.

Mistakes to avoid

• Comparing your ACoS to a blog’s “good ACoS”. Amazon itself declines to give one.
• Changing bids, budgets and the listing in the same week, so nothing can be attributed.
• Optimizing a campaign for ACoS when the product’s job is to build rank and reviews for a defined launch period.
• Mixing vendor and seller accounting: first-party vendor margins and cost lines differ from a seller’s fee stack.
• Using a screenshot of one good day as a report.