Home › Resources › Insights › Marketplaces › Cutting Amazon ACoS
Most ACoS reductions are volume reductions in disguise. Lowering bids across the board improves the ratio and shrinks the business. The fix is architectural — separating campaigns by the job they do, so you can cut waste in one place without switching off demand in another.
The Dazzle Commerce Team
eCommerce Management & Marketing
Updated
5 September 2026
Reading time
8 min read
ACoS is a ratio between two advertising numbers. Managing it well means looking at a third.
There are two ways to bring an Amazon ACoS figure down. The first is to spend less: cut bids, pause the expensive keywords, watch the ratio improve and the revenue line fall with it. The second is to stop paying full price for demand you already own, while continuing to pay properly for demand you do not. Only the second one is a strategy.
The difference between them is almost entirely structural. If branded, competitor, category and discovery traffic all sit inside the same campaigns, you cannot lower the cost of one without lowering the volume of another. Architecture is what makes those decisions separable.
KEY TAKEAWAY
If your ACoS improved and your ordered revenue fell by a similar proportion, you did not lower your cost of sale. You bought less of it.
ACoS is advertising spend divided by advertising-attributed sales. It is a ratio between two numbers that both belong to the ad account, which makes it a useful efficiency signal and a poor business one.
It says nothing about your margin, nothing about the organic sales the advertising helped rank, and nothing about whether the customer would have found you anyway. Three metrics between them fix most of that, and you need all three on the same screen.
Metric
What it tells you
What it leaves out
ACoS
Ad spend as a share of ad-attributed sales
Margin, organic lift, incrementality
TACoS
Ad spend as a share of total sales
Which campaigns earned the organic lift
Break-even ACoS
The ratio at which a sale stops making money
Rank ambition, launch strategy, inventory position
Break-even ACoS is simply your contribution margin. If a unit contributes 34% after product cost, referral and fulfilment fees, returns and carriage, advertising can absorb up to 34% of revenue before the sale loses money.
Set a break-even ACoS per product group before you touch a single campaign, then set a target beneath it. The gap between target and break-even is your profit. The gap between target and actual is your opportunity.
is your break-even ACoS
A 34% margin makes 34% a ceiling, never a target
break-even, on harvest
Where you defend proven, converting search terms
break-even, on discovery
Rank and review velocity are bought here, deliberately
These are structural rules rather than benchmarks. The right numbers depend on your margin, your rank ambition and how much inventory you can afford to move.
That last line matters more than it sounds. A discovery campaign running above break-even is not a failure if its job is to buy rank and review velocity on a new ASIN. It becomes a failure the moment nobody has written down when it is supposed to stop.
We use five tiers. Each has one job, one target and one set of rules, and no keyword is allowed to do two jobs at once.
1
Brand defence — exact match on your own brand and ASIN terms. Low bids, high conversion, tight budget. Its job is to hold the top of your own search result, not to grow the account.
2
Harvest — exact match on search terms that have already converted, promoted out of discovery. This is where a target below break-even belongs, and where most of the profit sits.
3
Discovery — broad and phrase match plus auto campaigns, deliberately run above break-even behind a hard budget cap. Its output is search terms, not profit.
4
Competitor and category — ASIN and category targeting. Expensive, strategically useful and the easiest tier to overspend. Cap it separately or it will quietly eat the account.
5
Defensive placements — sponsored display and product targeting on your own listings, so competitors are not buying the space underneath your detail page.
Campaign architecture only pays off if the catalogue underneath it is clean: listing quality sets the conversion rate every bid is priced against.
1
Change bids on data, not on dates. A keyword needs enough clicks to say something before a confident cut — for most catalogues that means several hundred clicks or roughly ten conversions.
2
Move in steps of ten to fifteen per cent. Larger jumps change placement, which changes conversion rate, which invalidates the comparison you were trying to make.
3
Cap discovery by budget, not by bid. Lowering discovery bids buys worse placements and slower learning; capping the budget buys the same learning at a known cost.
4
Review harvest weekly, discovery fortnightly, competitor monthly. Each tier accumulates signal at a different speed, so a single review cadence is wrong for at least two of them.
EXPERT VIEW
“Almost every account we inherit has one campaign doing four jobs. Splitting it out usually finds ten to twenty per cent of spend that was never buying anything, before we have changed a single bid.”
Dazzle Commerce · Marketplace team
If rebuilding the structure yourself is not the best use of your month, this is the work our Amazon team does in the first four weeks of an engagement.
Restructuring an account that is currently generating revenue is a nerve-racking exercise. Sequencing it properly removes most of the risk, because each week only changes one variable.
Week
What you do
What you should see
Week 1
Build a break-even ACoS per product group; change nothing in the account
A ranked list of which products can afford which spend
Week 2
Split brand defence and harvest out of the existing campaigns
Flat sales with lower spend on branded terms
Week 3
Cap and isolate discovery; negate every promoted search term
Higher discovery ACoS, lower blended ACoS
Week 4
Introduce competitor and defensive tiers with their own budgets
TACoS starting to move independently of ACoS
Volume usually wobbles in week two or three as placements settle. Judge the restructure on week five, not week three.
ONE RULE FOR THE RESTRUCTURE
Restructures fail when they are judged before the data has caught up. Agree the review date before you start, and then actually hold it.
Three services that pick up the structure this article describes.
Catalogue, content, advertising and account health managed as one Amazon programme.
The same segmentation logic applied across Google, Meta and marketplace advertising.
Multi-marketplace strategy for brands selling across Amazon, Walmart, eBay and regional platforms.
WRITTEN BY
eCommerce Management & Marketing
Dazzle Commerce manages Amazon, Walmart, eBay, Etsy and regional marketplace accounts alongside owned storefronts. This guide reflects the campaign structure we rebuild inherited accounts into during the first month of an engagement.
More from the marketplace and paid media side of the practice.
What has to be true before you open a second marketplace, from catalogue readiness to fulfilment.
Product feed quality decides more Shopping performance than bidding does. What to fix, in what order.
ROAS ignores returns, carriage and platform fees. How to rebuild the report around contribution margin.
We will map your current structure against the five tiers and show you where spend is doing two jobs at once. No obligation to work with us afterwards.
