Lowering ACoS without killing sales: a structured PPC audit

Most sellers try to lower ACoS by cutting bids across the board. It works for about a week, ACoS drops, and then sales drop right along with it, because cutting every bid equally treats a keyword generating $40 in sales the same as one generating $4. ACoS improves on paper while the campaign gets worse at the thing that actually matters, which is profitable revenue.

The fix isn't spending less. It's spending differently, on the keywords that are actually earning their spend and away from the ones that aren't.

What ACoS actually measures, and what it doesn't

ACoS is ad spend divided by ad-attributed sales, and on its own it tells you nothing about whether a keyword is profitable. A keyword with 40% ACoS on a product with 50% margin is still profitable. A keyword with 15% ACoS on a product with 12% margin is losing money on every sale, even though 15% looks like a great number on a dashboard.

Before touching any bids, know your actual break-even ACoS, which is roughly your margin percentage before ad spend. Anything meaningfully below that number has room to scale up. Anything meaningfully above it needs a specific decision, not a blanket cut.

Start with the search term report, not the bid sliders

Pull the search term report for the last 30 to 60 days, not just the campaign summary. The campaign-level ACoS number hides what's actually happening underneath it, where a small number of search terms are usually generating most of the spend and a much smaller number are generating most of the profitable sales.

Sort by spend, and look at the terms consuming the most budget. For each one, check its individual ACoS against your break-even number. Terms spending significant budget with no sales at all are the clearest, lowest-risk cuts, and they're usually where a large chunk of wasted spend is hiding, invisible at the campaign summary level.

Cut waste before you cut bids

Add high-spend, zero-conversion search terms as negative exact match. This is different from lowering a bid, because a bid reduction still lets the ad show for that irrelevant or low-converting search, just less often. A negative match stops it from showing at all, which is what you actually want for a term that's proven it doesn't convert.

Check for search terms that are technically related to your product but represent a different customer intent than what you're selling, like someone searching for a repair part when you sell the full product. These often have reasonable click-through rates and still convert poorly, because the click doesn't match genuine purchase intent for what you're offering.

Then adjust bids on what's left, keyword by keyword

Once true waste is negated, look at what remains. Keywords converting well below your break-even ACoS are candidates to raise, not lower, since they're proven profitable and likely capped by budget or bid rather than demand. Keywords hovering right around break-even are worth testing with a small bid change in either direction and watching for a week before deciding. Keywords consistently well above break-even, after waste has already been cut, are the ones that genuinely need a lower bid or, if they still don't improve, pausing.

This keyword-by-keyword approach takes longer than an across-the-board cut, but it's the difference between actually improving profitability and just making the ACoS number look better while revenue erodes underneath it.

Where TACoS fits into this

ACoS only measures ad-attributed sales, but a meaningful share of your organic sales exist partly because advertising built rank and visibility for those keywords in the first place. TACoS, total ad spend divided by total sales including organic, gives a fuller picture, especially for products where PPC is doing double duty: driving direct sales and also supporting organic ranking.

A campaign with a rising ACoS but a falling or stable TACoS isn't necessarily a problem. It can mean ad spend on a specific keyword is teaching Amazon's algorithm that keyword is relevant, and organic sales on that term are picking up the slack ad spend used to carry alone. Watching ACoS alone in that situation can lead to cutting a keyword that's actually working as intended, just not primarily through the ad click itself anymore.

Structuring campaigns so this is easier to see

Broad, phrase, and exact match keywords performing very differently under the same campaign makes the search term report harder to read, because strong exact-match performance can mask weak broad-match waste in the blended number. Splitting match types into separate campaigns, even for the same core keywords, makes it much clearer which match type is actually earning its spend and which is mostly generating irrelevant clicks.

Similarly, mixing genuinely different products or categories in one campaign blends their performance together, hiding a good keyword's results inside a bad product's overall numbers. Campaigns organized around single products, or at minimum closely related products with similar margins, make the keyword-level data far easier to act on.

What a realistic timeline looks like

Cutting clear waste, negative-matching zero-conversion high-spend terms, usually shows up in the numbers within a week or two, since it removes spend that wasn't generating sales anyway. Bid adjustments on borderline keywords need longer, generally three to four weeks of data, before the change is meaningful rather than noise, since day-to-day and week-to-week conversion rates fluctuate on their own even with no changes made.

Judging a bid change after three days almost always leads to premature conclusions in either direction. Give changes enough time to generate a real sample size before deciding they worked or didn't.