It is one of the most common patterns in Amazon advertising, and one of the least well diagnosed. Ad spend is up 40% year on year. Revenue is roughly where it was. ACoS has drifted from 22% to 34% over eight months without any single moment where it obviously broke.
The instinct is to attack the bids. Lower everything by 15%, see what happens. That usually produces two weeks of improved-looking ACoS followed by a sales decline, at which point the bids go back up and nothing has been learned.
The reason bid-cutting rarely works is that this pattern almost never has one cause. It has three or four running simultaneously, some of which have nothing to do with your advertising at all. Here is how to find them, in the order they should be checked.
First: Establish Whether It Is Even an Ads Problem
Before touching a campaign, pull two trend lines over the last twelve months: average cost per click and conversion rate.
Four combinations, four completely different problems:
- CPC up, conversion flat. The auction got more expensive. Your campaigns are performing exactly as well as they were; the market simply costs more. No amount of restructuring fixes this. What fixes it is recalculating what you can afford to pay.
- CPC flat, conversion down. This is not an advertising problem. Optimising bids here is treating the symptom.
- Both moving badly. Two problems at once, and they need separate fixes.
- Both stable but spend rising. You have added targets, campaigns, or placements without adding return. This is the bloat case, and it is the most fixable.
Most sellers skip this step and go straight to bids. It costs ten minutes and it determines everything that follows.
Cause 1: Keyword Bloat Nobody Pruned
Campaigns accumulate. An automatic campaign harvests terms, someone promotes the good ones to manual, and the auto campaign keeps running with everything else still live. Repeat that for eighteen months across a growing catalogue and a meaningful share of your budget is going to terms that have never produced an order.
How to check it: pull the search term report for a 60-day window and sum spend against terms with zero attributed orders. On accounts that have not been pruned in a year, 20% to 35% is a normal finding. That number is your ceiling for easy savings.
The trap: zero orders does not mean dead. A term with eight clicks and no sales tells you nothing. A rough threshold — if your conversion rate is around 10%, you want roughly 30 clicks before absence of sales is meaningful. Below that you are pruning on noise, and you will cut terms that would have converted.
Cause 2: Placement Multipliers Running Unchecked
Amazon lets you raise bids by placement, and Top of Search is where the multiplier does the most damage. A 100% adjustment means you are paying double on your most expensive real estate. It is a legitimate lever — Top of Search generally converts best — but it is usually set once during a launch push and never revisited.
How to check it: the placement report shows spend, sales and ACoS split by Top of Search, Rest of Search, and Product Pages. Compare each placement’s ACoS against your target. If Top of Search is running at 60% ACoS on a 25% target while Rest of Search sits at 20%, your multiplier is not paying for itself.
The fix: reset multipliers to zero, let a fortnight of clean data accumulate, then reintroduce one placement at a time. You cannot read the results if you change three things at once.
Cause 3: Duplicate Targeting Across Campaigns
The same keyword sitting in an automatic campaign, a broad match, a phrase match, and an exact match is more common than most sellers realise.
There is a persistent myth that this means bidding against yourself in the auction. It does not quite work that way — when several of your own campaigns are eligible for the same placement, Amazon runs the one with the highest bid, and only that one enters the auction. But the practical consequence is worse than the myth suggests: your most aggressive campaign always wins internally. A forgotten launch campaign with a £2.40 bid set eighteen months ago is quietly setting the effective cost for a keyword you thought you were bidding £0.90 on.
The second cost is data fragmentation. Performance for one search term is split across four campaigns, so none of them accumulates enough conversions to justify a confident decision.
How to check it: export all targets and look for duplicates across campaigns. Sort by bid descending and check whether the highest bid on each duplicated term is one you actually intended to set.
Cause 4: The Listing Degraded, Not the Ads
This is the one that gets misdiagnosed most often, and it produces exactly the symptom in this article’s title.
Nothing about the campaigns changed. The bids are the same, the targets are the same, traffic volume is the same. But the star rating slipped from 4.5 to 4.2 after a bad batch. Or a competitor launched at 20% under your price. Or the main image got flagged and swapped for a weaker one. Or a colour variation went out of stock and took the review count on the parent listing with it.
Clicks cost the same as they always did. Fewer of them convert. Spend holds steady, sales fall, ACoS rises — and the ad account takes the blame for a problem that lives on the product page.
How to check it: Business Reports, Detail Page Sales and Traffic. Look at unit session percentage over the same window as the ACoS drift. If conversion fell while sessions held, stop looking at your campaigns.
Why this matters more than any bid adjustment: optimising advertising on a listing that has lost conversion is spending effort to lose money more efficiently. Fix the page first.
Cause 5: Featured Offer Eligibility
Worth its own check because the symptom is counterintuitive. Sponsored Products ads generally require your offer to be the Featured Offer — the buy box — to be eligible to serve at all.
So when you lose it, ads on that ASIN do not waste money. They stop appearing. If your total budget is fixed, spend redistributes to your other ASINs, often the ones with worse economics. Total spend stays roughly level, total sales drop, and the account-level ACoS deteriorates for reasons that are invisible at campaign level.
How to check it: Featured Offer percentage by ASIN in Business Reports, compared against impression volume in your campaign data. An ASIN where impressions collapsed while nothing changed in the campaign is usually a buy box story.
Cause 6: Measuring ACoS and Nothing Else
ACoS answers one narrow question: what proportion of ad-attributed revenue did the ads cost? It says nothing about whether advertising is growing the business.
Two accounts can both report 28% ACoS. In one, TACoS — total ad spend against total revenue, organic included — is falling quarter on quarter, which means organic sales are growing faster than ad sales and the advertising is buying rank that persists. In the other, TACoS is climbing, which means organic is shrinking and paid is filling the gap. Same ACoS, opposite trajectories.
If you only ever look at ACoS, the second account looks healthy right up until the point where it very obviously is not.
Setting a Target That Actually Means Something
Most sellers are working against an ACoS target they picked up from a blog post or a YouTube video. It has nothing to do with their margins.
Start with break-even instead. Your break-even ACoS equals your gross margin percentage before advertising. Sell at £40, and after COGS, referral fee, FBA fee and freight you keep £14 — your margin is 35%, so break-even ACoS is 35%. At exactly 35% ACoS, that ad-driven sale earns you nothing. Above it, you are paying for volume.
That single number reframes everything. A 30% ACoS is disastrous on a product with a 22% margin and comfortable on one with 45%. Anyone who quotes you a target ACoS without asking what your product costs to make and ship is guessing.
Once you have break-even, set your operating target beneath it by whatever profit you need, and accept that the gap will be smaller during a launch — buying rank is a legitimate use of a temporary loss, provided somebody decided to do it deliberately and wrote down when it ends.
Published benchmark bands are useful as a sanity check on whether your number is wildly off, and agencies that publish theirs typically put ACoS under 15% as the zone for low-margin products, 15% to 30% for higher-margin or growth-phase products, and TACoS under 10% for established products against 10% to 20% during a launch. Treat those as orientation, not as targets. Your margins decide your targets.
The Order of Operations
If you take nothing else from this, take the sequence. Doing these in the wrong order wastes weeks.
- CPC and conversion trends over twelve months, before you touch anything.
- Listing health check — conversion rate, rating, price position, image status, stock. Fix anything broken here before proceeding.
- Featured Offer percentage by ASIN.
- Search term report, 60 days, spend against zero-order terms above the click threshold.
- Placement report — is Top of Search earning its multiplier?
- Duplicate target audit — what is your highest live bid on each keyword?
- Recalculate break-even ACoS from real current costs, not last year’s.
- Only now, bids.
Dayparting belongs at the very end. It is a genuine lever — most categories show real intent variation across the day and week — but it is a refinement on a working account, not a rescue for a broken one. Turning ads off between 2am and 6am does not fix a listing that stopped converting in March.
When This Stops Being a Monday Morning Job
At three products, this whole diagnosis is an afternoon with a spreadsheet. At forty SKUs across three marketplaces, it is not, and the reason most accounts drift is not incompetence — it is that nobody has eight uninterrupted hours a month to do it properly while also running the rest of the business.
That is the point at which brands generally either hire someone whose only responsibility is the ad account, or bring in specialist Amazon ad management with access to the enterprise bid-management tools that make continuous optimisation practical at scale. Both work. The deciding factors are catalogue size and how much of your own week the account currently consumes.
If you go the external route, one filter separates the useful from the rest: ask them to show you the break-even calculation on your actual products before they touch a bid. Anyone who commits to an ACoS target without first asking what your unit economics are is selling you a number, not a strategy.
What to Do This Week
- Pull the twelve-month CPC and conversion trend. Decide which of the four combinations you are in.
- Check unit session percentage before you look at a single campaign.
- Run the 60-day zero-order spend calculation. Whatever the figure, that is your available headroom.
- Work out your genuine break-even ACoS from current costs, including this year’s fee changes.
- Write down what you changed and when. Three months from now, that annotation is the only way to know whether any of it worked.
Rising spend against flat sales is not a mystery. It is a sequence of small unattended things compounding — and it is far more often solved by an audit than by a bid adjustment.
