By Ted Gibson — RippleEffect
ACOS is ad spend divided by ad revenue, so there are only three ways to lower it: pay less for a click, convert more of the clicks you buy, or stop buying the clicks that were never going to convert. Everything written under this heading is a version of one of those three. The awkward part is that the fastest way to lower ACOS is to advertise less, which is why the number falls so reliably in accounts that are quietly shrinking. One distinction before the rest, because it is most of the reason the question gets asked at all. ACOS only sees the sales your ads touched, while total advertising cost of sale divides your whole ad spend by your whole revenue, organic included. Move budget onto your own brand name and ACOS improves while TACOS does not, because you have bought sales you already had. What counts as a good number is a different question with its own post. This one is about how you move it.
The published targets disagree, and the only measured one fails its own test
Before optimising towards a number, look at where the number came from. Adbrew puts a good range at 20 to 25%. Trellis says most sellers target 25 to 35%, so its typical band begins exactly where Adbrew's good range stops. Ad Badger says good lands between 15 and 30% for most sellers, and then publishes the only measured figure in the genre: an average of 34.42% per user per day across its own Sponsored Products accounts, and it has been that since the beginning of 2019. The average account on that platform runs above what the platform calls good. None of the three cites a study, because there is not one to cite.
Amazon holds the whole dataset and declines to give you a number. Its own guide says there is no definitive figure for a good ACOS, and offers a rule in place of one: break-even ACOS is your profit margin, so yours has to sit below it for the advertising to make money. That sentence is worth more than the tables. A brand running 45% contribution margin and one running 22% are being handed the same 25% target by pages that have seen neither profit and loss account. Work the ceiling back from your own margin, the same way you would for TACOS, and treat everyone else's range as a sanity check at the extremes.
Two of the three inputs are not in the advertising console
Write ACOS out in full and it stops looking like an advertising metric. Spend is clicks times cost per click. Ad revenue is clicks times conversion rate times selling price. Cancel the clicks from both sides and ACOS is cost per click divided by conversion rate times price. Only the first of those three lives in the campaign manager, which is why bid work on its own tends to buy a few points and then stall. The other two move the number harder. Jarvio puts a 2% conversion improvement at three to five points off ACOS across a portfolio, which is its own unaudited figure, though the arithmetic says it is the right order of magnitude. The point survives without the number. A listing converting at 12% rather than 10% turns the same clicks into a fifth more orders, so the same £10,000 of monthly ad spend buys a fifth more sales, and the same $10,000 does the same, with no bid touched. What actually moves conversion is a separate discipline, and it is usually cheaper than the bidding. Price is the third input and the one nobody writes about, because raising it lowers ACOS mechanically and can cost you the volume that was paying for your rank.
The fast ways down, and what each one costs
Spend less, and ACOS falls while sales fall with it. That is buying growth you cannot hold read backwards, and it is the same mistake. Lean on branded search, and blended ACOS drops inside a week, because clicks on your own name convert at rates non-branded terms never reach. It is the most obvious move on this list and the most expensive one to get wrong, and what your branded spend is actually buying is its own question with its own test. Or pause everything with a high ACOS, which is right for the campaigns that were pure waste and wrong for the non-branded terms holding your organic rank. That version reads as a lower ACOS this month and a smaller business two quarters out.
What to do first, and how to tell whether it worked
Order matters more than the list does. Start with the search term report and negative out the terms that spent with nothing to show for it, since that is the one move with no downside; Jarvio's line of 20 or more clicks and no conversions is as good a starting point as any, and promoting the terms that did convert into campaigns of their own is the other half of the same job. Look at bidding strategy next. On Amazon's up and down strategy, as Adbrew sets out, Amazon can raise your bid by up to 100% for placements at the top of the first page of results and by up to 50% for other placements, so a setting made once and forgotten can be doubling what you pay for a click today. Move bids after that, and only on terms with enough clicks to mean anything. Conversion and price work runs alongside all of it, on a slower clock.
Then the part the guides leave out. Scale Insights ranks at the top of this query and gives no target figure and no test at all, which is typical of the page: a list of tactics with no way to check whether the tactic worked. Compare like periods, same weekdays, no promotion running and no price change. Read ACOS next to total advertising cost of sale and total sales rather than on its own, because two of those three moving the wrong way is how you find out you cut muscle. Change one thing at a time, which is slower and is the only version that teaches you anything.
We do not publish an ACOS figure for any account we run, so we are not going to imply one here. The nearest published number we have is MUUSH, a launch taken from zero to profitable in under six months, where cost per acquisition fell 19% in the first quarter of 2026 against the quarter before. Cost per acquisition is not ACOS, and the two come apart as soon as your average selling price moves. If you asked us to lower yours, the first question back would be what you want to happen to sales, because those two answers point at completely different work. A free account audit separates the spend buying rank you intend to keep from the spend buying sales you already had, in priority order, and what the work costs afterwards is published.