The way to lower ACOS on Amazon without killing sales is structural: remove the spend that never produced an order, stop overpaying for placements and clicks that do produce, and leave the machinery generating revenue untouched. What it is not is a blanket bid cut. Slashing bids lowers ACOS the way skipping meals lowers a grocery bill: technically true, and the account starves. In the accounts I audit, a high ACOS is almost never one big mistake. It is dozens of small structural leaks: zero-order search terms still taking clicks, top-of-search premiums on terms that convert better elsewhere, proven keywords priced wrong. This guide walks the five levers I actually pull, in order, with the formulas behind them, and ends on a real account that went from 39.66% to 27.02% ACOS with no budget increase.
Why "just lower your bids" backfires
Blanket bid cuts lower ACOS by shrinking the account. Lower bids lose auction position, clicks fall, sales fall, and the organic rank those ad sales were feeding erodes over the following weeks. You end up with a prettier ratio on a smaller business, which is a loss wearing a win's clothing.
The mechanics matter because the advice sounds so reasonable. ACOS is ad spend divided by ad sales, so cutting spend does shrink the numerator. But bids do not just control cost; they control access. Cut them everywhere and the denominator follows the numerator down. The second-order damage is worse, and the ads console never shows it: Amazon treats sales velocity on a search term as evidence you deserve to rank for it. Every ad sale is an audition for an organic spot, and the sales you surrendered were paying for stage time. A month later, total revenue is down and the "efficient" account is quietly more fragile.
None of this means bids never come down. Three of the five levers below touch bids. The difference is they touch specific bids, for stated reasons, with a formula behind each one, not every bid because the topline scared you. And if you watch one number while you work, make it TACOS: as I covered in ACOS vs TACOS, a falling ACOS with a rising TACOS is not efficiency, it is retreat.
Diagnose first: which lever actually moved
ACOS is CPC divided by revenue per click, and revenue per click is conversion rate times average order value. That means a high ACOS has exactly three possible causes: clicks got more expensive, fewer clicks converted, or orders got smaller. Identify which lever moved before you change anything.
Run it on the dataset I grade across these guides: $36,303 in spend, $111,058 in ad sales, 4,140 orders from 45,672 clicks. The toplines give a $0.79 average CPC and $2.43 of revenue per click ($111,058 divided by 45,672 clicks). Divide one by the other and the account's 32.7% ACOS falls back out, allowing for rounding. That is the entire metric: what a click costs over what a click earns.
Each cause has different fingerprints. CPC creep traces to raised bids, "up and down" dynamic bidding, inflated placement multipliers, or hotter competition. A conversion drop traces to the product side: price changes, lost Buy Box, review damage, or irrelevant traffic pouring in through broad match. And when you hunt, weight by dollar impact, never by the ACOS column alone. A terrifying ACOS on pocket-change spend is a rounding error; a merely bad ACOS on serious spend is the actual problem.
The five levers, in the order I pull them
Five levers lower ACOS without surrendering sales: negate the zero-order bleed, pull placement premiums that do not earn their markup, reprice expensive keywords that still convert, harvest proven search terms into exact match, and realign budget toward what converts. Worked in that order, they cut cost first where there are no sales to lose.
Behind all five sits one classification. Every keyword in an account lands in one of four buckets: converting above target ACOS, spending without converting, converting below target, or too little data to judge. Each bucket gets its own move, and not one of those moves is "cut everything 20%." That is the whole case against the blanket bid change: it applies a single verdict to four different crimes.
| Lever | The move | Why sales survive |
|---|---|---|
| 1. Cut zero-order bleed | Negate irrelevant search terms past target CPA with no orders | The blocked terms had no sales to lose |
| 2. Pull placement premiums | Reduce top-of-search multipliers the placement's conversion rate does not earn | The keyword keeps serving everywhere it converts |
| 3. Reprice expensive converters | New bid = revenue per click × target ACOS | The keyword keeps its orders, at a price that lands on target |
| 4. Harvest winners into exact | Move 2+ order search terms into exact match with precise bids | Same shopper, same order, tighter control over the price |
| 5. Realign budgets | Open budgets on capped winners; rebalance overfunded losers | This lever adds sales while ACOS falls |
Lever 1: Cut the zero-order bleed with negatives
Start with spend that produces nothing, because removing it is free. A search term that has taken clicks past your target cost per acquisition with zero orders is pure numerator: it raises ACOS with every click, and removing it cannot cost you sales, because it never made any.
The threshold is target CPA: average order value times target ACOS. On the dataset above, a $26.82 order value and a 30% target put the line around $8 of spend. Open the report as covered in my search term report guide, filter orders to zero, sort spend descending, and give each term a fair trial before sentencing: one divided by your conversion rate is what an average order costs in clicks, roughly eleven at this dataset's 9.1%. How much zero-order spend is tolerable depends on phase; my benchmarks guide allows 50–60% of spend at launch and holds a mature account under 25–30%.
The free Wasted Spend Finder ranks your zero-order terms by spend in about a minute, and the Negative Keyword Finder turns the list into a ready-to-upload file, both parsed in your browser. One rule before you upload anything, covered fully in my negative keywords guide: if the term is relevant to the product, fix its bid instead. Negatives are for terms that do not pay at any price.
Lever 2: Pull the placement premiums that don't earn their markup
Placement multipliers silently multiply your bids, which makes them the fastest ACOS fix most sellers never check. Compare conversion rate and ACOS across top-of-search, rest-of-search, and product pages. Anywhere the premium placement converts no better than the cheap one, you are paying a markup for geography.
A 100% top-of-search modifier doubles the effective bid, so a keyword can be priced perfectly at the base and wildly overpriced where it actually serves. Top of search usually earns its premium with the highest conversion rate on the page, but usually is not always, and the exceptions are expensive. My benchmark table holds top-of-search ACOS to roughly 60% at launch, tightening to 30% in the profitability phase. When a placement blows through that, the fix is the multiplier, not the base bid: cut the multiplier and the keyword keeps serving everywhere it converts.
This lever paid for an entire engagement on my results page. A $310K per month home goods and kitchen brand was leaking roughly $11K a month to top-of-search placements on terms that converted better lower on the page. Repricing the placements, alongside the bid rebuild in lever 3, took ACOS from 32.1% to 21.6% in three months and added $28,400 a month to profit.
Lever 3: Reprice expensive keywords that convert. Do not kill them
A keyword converting above your target ACOS is mispriced, not broken, and the fix is a formula, not a funeral: new bid equals revenue per click times target ACOS. That prices every click at exactly what the keyword's own sales history says a click is worth to you at your target.
Revenue per click is the keyword's sales divided by its clicks. Run it account-wide on the dataset: $2.43 of revenue per click at a 30% target ACOS prices the average click at $0.73. The account pays $0.79. The formula's verdict is a six-cent trim, roughly 8%, and that is the shape of this lever at the keyword level too: it usually says trim, sometimes says hold, and occasionally says raise, because the current bid is a status, not a data point.
Compare that to the two popular alternatives. Pausing the keyword surrenders its orders and the organic position those orders were feeding. A blanket 20% cut prices every keyword by your fear instead of its performance. The formula does neither: the keyword keeps selling, at a cost that mathematically lands it on target.
Lever 4: Harvest the proven winners into exact match
Harvesting moves search terms that have proven they convert out of auto and broad campaigns into exact match, where the query gets its own bid. It lowers ACOS from the control side: the same shopper, the same order, but a price you set for that exact search instead of a discovery bid.
The bar matters more than the mechanics. I want at least two orders before a term earns its own keyword: one order can be coincidence, two prove a pattern with search volume behind it. Confirm the advertised product itself got the order rather than a halo sale to a sibling, and search the term on Amazon to check your product belongs in the results. Then add a negative exact in the campaign that found it, so discovery stops re-buying traffic you now own and the data stays clean. Over-harvesting is its own ACOS problem: pull in every one-order wonder and you build a bloated structure of keywords averaging a click a month, each nudging spend upward with no sales attached.
Lever 5: Realign spend with what actually converts
The last lever moves money instead of cutting it: open budgets on campaigns hitting target ACOS that run out of budget early, and pull back the products absorbing spend they never earn back. It is the one ACOS lever that grows sales while it works.
Sort campaigns by spend and look for two mismatches. The first is a winner on a leash: a campaign at or under target that exhausts its daily budget by afternoon is capping your cheapest sales, and raising that budget lowers account ACOS by adding efficient revenue to the denominator. The second is the imbalanced catalog: one ASIN absorbing most of the budget while products that convert sit starved. The healthy state is spend share roughly tracking each product's sales contribution, and in the accounts I audit it rarely does. Nothing in this lever touches a bid, which is the point: allocation errors are invisible to bid optimization, and no amount of bid tuning fixes money parked in the wrong place.
What the full rebuild looks like on a real account
The health and personal care account on my results page is the cleanest proof that ACOS is structural: ACOS fell from 39.66% to 27.02% in 90 days, worth $4,735 a month in added profit, with no budget increase. Nothing was slashed. The structure stopped leaking.
On the surface this seller was doing everything right: campaigns built, bids adjusted, budgets allocated. Underneath, the account had search term bleed across campaigns splitting its conversion data, wasted spend on non-converting keywords that looked active, and a bid architecture that rewarded impressions over conversions. The rebuild was levers 1 through 5, not a bid massacre, and the tell is in the numbers. Spend stayed level while ACOS fell by nearly a third, which is only possible if the same budget moved from clicks that did not convert to clicks that did. TACOS fell from 9.25% to 4.96% over the same window, so the business got less ad-dependent while the ads got cheaper, and profit margin climbed from 34.11% to 41.58%.
A bid-slashing "fix" produces the opposite signature: ACOS down, sales down, TACOS flat or worse, and an organic-rank erosion that surfaces a month later. If you want to see the diagnosis mechanically, row by row, the audit dashboard demo runs the full read on a sample account.
Frequently asked questions
What causes a high ACOS on Amazon?
Mathematically, only three things: your cost per click rose, your conversion rate fell, or your average order value fell, because ACOS is CPC divided by conversion rate times order value. Structurally, the usual culprits behind those moves are zero-order search terms taking clicks, inflated placement multipliers, and proven keywords priced above what their revenue per click justifies.
Does lowering bids lower ACOS?
On a converting keyword, yes: a lower CPC against the same revenue per click drops that keyword's ACOS. Across the board, it is a trap. Blanket cuts lose auction position, sales fall with spend, and the organic rank those sales fed erodes. Lower specific bids with a formula (revenue per click times target ACOS), never every bid by a percentage.
Should I pause keywords with a high ACOS?
Not if they convert. A high-ACOS keyword with orders is mispriced, and repricing it (revenue per click times target ACOS) keeps the sales while pulling its cost to target. Pausing surrenders both the orders and the organic position they fed. Pause or negate only terms that are irrelevant to the product or have failed a fair trial with zero orders.
How long does it take to lower ACOS on Amazon?
The structural fixes act at different speeds. Negatives stop bleed the day they upload, placement and bid repricing show inside a week or two, and harvests need time to accumulate data. The rebuilds on my results page ran about 90 days start to finish. Judge changes on 7+ day windows, because Amazon's attribution can lag as much as two weeks.
Can you lower ACOS without losing sales?
Yes, if the reduction is structural. A health and personal care account I rebuilt went from 39.66% to 27.02% ACOS in 90 days with no budget increase, worth $4,735 a month in added profit. That only works when the spend you remove was producing nothing: zero-order terms, unearned placement premiums, and overpriced clicks on keywords that kept converting.
Why did my sales drop when I lowered my ACOS?
Almost always because the ACOS drop came from blanket bid cuts. Lower bids lose auctions, clicks fall, ad sales fall, and within weeks the organic rank your sales velocity was feeding slips too. Check TACOS: if it held or rose while ACOS fell, you retreated rather than optimized. The fix is restoring bids on converters and cutting only unproductive spend.
Find the leak before you touch a bid
The free Account Health Snapshot grades nine metrics from your bulk file against the phase benchmarks in this guide, wasted spend and top-of-search ACOS included: about 60 seconds, parsed in your browser, no email, no account. If the scorecard confirms the leak but not the fix, the free 30-minute diagnosis call is where I walk your account's own levers with you, in order, and you leave with the list either way.