Amazon PPC Metrics That Matter: Which Numbers Point to a Lever
Your Amazon ads dashboard can look healthy for a quarter while the account underneath it loses margin.
The cause is usually the metric on the screen. ACoS, ROAS, CTR and impressions each answer a narrow question correctly, and none of them says whether the account is making money or why a number moved. The trouble starts when a weekly report puts a dozen figures side by side, every one is green, and nobody can say which lever moved which.
This guide ranks Amazon PPC metrics by one test: does the number point to something you can change? It covers the efficiency metrics, the ACoS and TACoS reading most teams get wrong, the numbers that flatter a dashboard, and what happens to diagnostic visibility once automation takes over the manual work.
Quick answer: the Amazon PPC metrics worth a weekly look
- What matters: the metrics tied to a decision. Break-even ACoS sets the bid ceiling, TACoS reads the business, CVR reads the listing, and search term data reads targeting.
- How to read ACoS: it is ad spend over ad-attributed sales, so it means little until you compare it with your break-even point. Amazon's own ACOS guide says the same.
- How to read TACoS: it divides ad spend by total sales. A falling TACoS can mean the ads are working or that spend was cut, so read it next to total sales.
- The opportunity: pair every metric with the lever it points to and the weekly report becomes a short list of actions.
- The risk: a metric shows what changed, not whether the change was right, and automation can remove the trail that connects a number to its cause.
- Our read: keep a short scorecard, log changes at campaign level, and treat a green dashboard as a reason to check the cause.
The metrics that earn a place on the weekly scorecard
Every metric below has a lever attached. If a number has no lever, it belongs in a monthly summary, not the weekly review.
| Metric | Formula | Lever it points to | Where it misleads |
|---|---|---|---|
| ACoS | Ad spend ÷ ad-attributed sales | Bids, targeting, placements | Ignores margin, organic sales and what the campaign is for. New campaigns run high by design. |
| Break-even ACoS | Margin after COGS, fees and fulfillment, as a % of price | The ceiling for bids and the ACoS range | Moves with price, fees and COGS. A stale number sets a false ceiling. |
| ROAS | Ad-attributed sales ÷ ad spend (the inverse of ACoS) | Same levers as ACoS | Carries every ACoS blind spot. A strong blended figure can hide branded campaigns doing the work. |
| TACoS | Ad spend ÷ total sales | Overall ad dependence and budget level | Moves for reasons unrelated to ads: stockouts, price changes, promotions, a spend cut. |
| CVR | Orders ÷ clicks | Listing, price, reviews, Buy Box | Falls when you bid on weaker traffic, even if the listing has not changed. |
| CTR | Clicks ÷ impressions | Title, main image, targeting relevance | Says nothing about whether the click was worth paying for. |
| CPC | Ad spend ÷ clicks | Bid levels, competition, placement | Rising CPC can be market pressure rather than a bid error. |
ACoS and break-even ACoS set the bid ceiling together
Amazon states that break-even ACOS is tied to profit margin and that ACOS needs to sit below margin to stay profitable. Margin here means what is left of the price after COGS, Amazon fees and fulfillment, before any ad spend.
A worked case: a $40 product leaves $14 before ads. Break-even ACoS is 35% and break-even ROAS is 2.86. A campaign at 30% ACoS earns on every ad-attributed order. A campaign at 40% spends $16 to win a $14 margin and loses $2 per order, unless those orders buy something else, such as a repeat purchase or organic rank. Lifetime value belongs in the ceiling only when you measure repeat purchase and not when you assume it. Our free break-even ACOS calculator does the arithmetic, and the complete guide to ACoS covers the ranges.
ROAS is ACoS upside down, so it shares the blind spots
Amazon defines ROAS as ad revenue divided by ad spend, the inverse of ACOS. Switching to ROAS changes the presentation and nothing else. Both stop at the ad-attributed order, so both ignore organic sales, shipping cost and what the shopper does next.
CVR and CTR diagnose the listing and the traffic
CVR belongs to the listing: price, reviews, images and Buy Box status. When CVR drops after you widen targeting, the listing did not get worse. The traffic did. CTR belongs to relevance and creative. Read it as a diagnostic on the first impression, and never as a score. For category and ad-type reference points, use our Amazon ads benchmarks rather than a single universal target.
ACoS and TACoS answer different questions, and most teams read them as one
ACoS compares ad spend with sales the ads drove. TACoS compares ad spend with all sales, paid and organic. A move in one without the other is the signal. The illustrative months below use the same product and show four different situations.
| Month (illustrative) | Ad spend | Ad sales | Total sales | ACoS | TACoS | What the pair says |
|---|---|---|---|---|---|---|
| A: baseline | $1,000 | $4,000 | $10,000 | 25% | 10% | Reference point. |
| B: organic slips | $1,000 | $4,000 | $8,000 | 25% | 12.5% | ACoS is flat, so the ads are fine. Organic fell by $2,000: check rank, stock, Buy Box and price. |
| C: launch push | $1,600 | $4,000 | $16,000 | 40% | 10% | ACoS is above a 35% break-even, but total sales grew 60% at a flat TACoS. If the organic lift is real, the spend is buying rank. |
| D: spend cut | $500 | $2,000 | $7,000 | 25% | 7.1% | TACoS "improved" because spend was cut while total sales fell 30%. |
Month B is an organic problem that ACoS cannot see. Month C is a bid decision that looks wrong on ACoS and may be right on TACoS. Month D is the one that gets praised in a review and should not be. TACoS cannot separate ad-driven organic sales from baseline organic sales, so deciding whether the ads earned the lift is a judgment call that needs rank and search term data behind it.
Five metrics that flatter the dashboard
A vanity metric moves without pointing at a lever. These five show up in reviews more than they should.
Impressions and CTR without a conversion read
Impressions climb whenever bids or budgets climb. CTR can rise on irrelevant queries. Neither tells you the click paid back. Check them against CVR and orders before reading anything into them.
Blended ROAS across branded and non-branded campaigns
Branded campaigns capture demand that already exists and convert well. Blend them with non-branded campaigns and the average hides which part of the account is finding new customers. Report them on separate lines.
ACoS on campaigns that are still learning
Amazon notes that new campaigns tend to run a higher ACOS simply because they are new. Judge a launch on its stated goal and its time window, not on week-one ACoS.
A TACoS that falls because spend was cut
Month D above is the pattern. TACoS improves on paper while total sales fall. Always read it beside total sales and ad sales growth.
The account-wide average
An average ACoS can sit at target while a handful of campaigns lose money and a few large winners cover for them. Sort by campaign, then use the search terms report to see where the spend inside the weak ones is going.
A metric explains what changed, not whether the change was right
ACoS rising 6 points after a bid increase is correct if the goal was rank, and wrong if the goal was profit. The metric cannot tell you which. Write the intent per campaign before reading the numbers:
- Launch: read total sales, organic rank and TACoS, with an ACoS ceiling set deliberately above break-even for a fixed window.
- Defend: read branded share of ad sales and impression share on your own terms.
- Harvest: read ACoS against break-even and conversion on proven search terms.
- Profit: read ACoS against break-even at the campaign level, with no account-wide average.
The same metric then supports different readings, and the report stops arguing with itself.
When automation takes over the manual work, the diagnostic surface goes with it
When you adjusted bids by hand, you knew what changed on Monday because you changed it. Automated bidding runs through far more changes than anyone reviews. The dashboard keeps the result, and the cause lives somewhere else, if it was recorded at all. A metric tells you ACoS rose 4 points. The question is which campaigns, which targets and which change.
| What you need to see | Manual review | Automated bidding |
|---|---|---|
| Why a bid changed | You made the change and remember the reason, until the third account. | Visible only if every change is logged per campaign with the logic that fired. |
| Coverage | The campaigns you open this week. | Every campaign and target on every run. |
| Time to notice drift | The next weekly review. | The next run, if someone is watching a signal. |
| Explaining a move to a client or boss | Slow, but you can reconstruct it from your own changes. | Only as good as the log. Without one you see the result and guess at the cause. |
If you already run automation without a change trail
You can rebuild the diagnosis with a short routine, and it is worth running before you change vendors or settings:
- Pull week-over-week ACoS, spend and orders by campaign.
- Sort by the absolute change in spend and open the top five campaigns.
- For each one, find the bid, budget or targeting change that came before the move.
- Write down whether the change matched the campaign's intent, and what you will do next.
Our PPC account audit workflow extends the same habit to a full review. Step three is the one that breaks in practice, because it only works if the changes were logged with the campaign they touched.
Qinetix keeps the trail and Qore runs the review
Step three is answerable when every change is recorded per campaign. Our platform Qinetix runs ads automation and dynamic pricing, and it keeps a history log per campaign, change annotations and bid history per target, so a move in ACoS can be traced to the change behind it. Operators choose which algorithm runs on which campaign, so the logic behind a change is a choice you made and can inspect.
Qore from Trellis is where you turn a review you run by hand into a workflow that runs on a schedule. You describe the review to Q, the assistant that helps you build the skill. Qore shows the logic as a visible graph, you lock it, and it runs the same way each time. The sort, the top movers and the lookup in the routine above are the kind of review you can build this way. Actions such as bid changes default to manual approval, and autonomy is available once you trust the skill. You can see what it does on the Qore overview, and Qore is in open beta with self-signup.
The limit is the same one named earlier. A locked skill can tell you which campaigns moved and what changed. It cannot tell you whether the change was right, because that depends on the intent you set. Qore's analysis runs on its own, while actions that change bids assume Qinetix is managing those bids.
Build the scorecard around levers and keep the trail
The metrics that matter are the ones that point at something you can change: break-even ACoS for bids, TACoS read beside total sales for the business, CVR for the listing, and search term data for targeting. Everything else is context. Write the intent per campaign, report branded and non-branded separately, and make sure every automated change leaves a record you can read next week. If you want the weekly review to run on its own logic, start with the Qore open beta and build the first skill around the routine above.
Frequently Asked Questions
The ones tied to a decision. Break-even ACoS sets the bid ceiling, TACoS reads the business against total sales, CVR reads the listing, and search term level spend and orders read targeting. CTR, impressions and CPC are diagnostics that support those, not scores.
A good ACoS is one below your break-even ACoS for that product and consistent with the campaign goal. Break-even ACoS equals your margin after COGS, fees and fulfillment, as a percentage of price. Category averages in our Amazon ads benchmarks give context, but the break-even figure is the one that decides.
ACoS divides ad spend by sales attributed to ads. TACoS divides ad spend by total sales, including organic. ACoS judges the efficiency of the ads, and TACoS shows how dependent the business is on them. Read both, and read TACoS beside total sales.
Neither is better. Amazon defines ROAS as the inverse of ACOS, so they carry the same information in different form and share the same blind spots: both stop at ad-attributed sales and ignore organic sales and margin.
Take the price, subtract COGS, Amazon fees and fulfillment, and divide what remains by the price. That percentage is break-even ACoS. Break-even ROAS is 1 divided by that margin. A $40 product with $14 left before ads has a 35% break-even ACoS and a 2.86 break-even ROAS.
Any metric that moves without pointing at a lever: impressions and CTR read alone, blended ROAS across branded and non-branded campaigns, ACoS on campaigns still in their learning window, a TACoS that fell because spend was cut, and an account-wide average ACoS.
Require a change log per campaign that records the change, the time and the logic that fired. Without it, sort campaigns by change in spend each week and trace the top movers back to a bid, budget or targeting change by hand. A platform that records changes at campaign level makes that a lookup instead of an investigation.