Lowering ACoS is easy. Lowering it without shrinking is the job.

ACoS, and the number it is usually confused with

ACoS is ad spend divided by ad-attributed sales. It is the inverse of ROAS: a 25% ACoS is a 4x ROAS, a 50% ACoS is a 2x. Amazon sellers use ACoS and everyone else uses ROAS, which is why cross-channel conversations tend to go in circles for ten minutes before someone converts the numbers. Both describe the same ratio from opposite ends.

The number that matters more as an account matures is TACoS: total ad spend divided by total revenue, ads and organic together. ACoS tells you whether an ad campaign paid for itself. TACoS tells you whether advertising is buying you a business or subsidising one. A falling TACoS with flat or rising total revenue is the signal that paid is driving organic rank rather than substituting for it. A rising TACoS with flat revenue means you are paying more each month for the same sales, which is the most common quiet failure mode in a mature Amazon account.

Why category benchmarks are worthless here

Two sellers in the same category can have a 60% gross margin and a 22% gross margin. For the first, a 40% ACoS is a healthy launch investment. For the second it is a business that loses money on every order and makes it up on volume. An average across those two describes neither. This is why we will not publish an average ACoS number and why you should distrust anyone who does without asking what your margin is.

Break-even ACoS equals gross margin, and gross margin here has to include everything that scales per unit: cost of goods, Amazon referral fee, fulfilment fee, storage, returns and the packaging. Sellers who compute margin before FBA fees end up with a break-even ACoS that is wrong in the optimistic direction, which is the worst direction for it to be wrong in.

How to actually bring ACoS down

Start with spend that produced nothing. Pull the search term report, find the queries with meaningful clicks and zero orders, and negate them. This is the only ACoS lever that costs you no revenue at all, and in a neglected account it is usually the largest one. Then look at placements: a campaign that is unprofitable overall is often profitable on rest of search and heavily unprofitable at top of search, which is a multiplier problem rather than a keyword problem.

Bids come next, and they should move in small increments against sufficient data. Cutting a keyword's bid by half because it had eight clicks and no order is not optimisation, it is reacting to noise. Structure comes last: separating high-intent exact match from discovery, splitting ASINs with different margins into different campaigns, and giving each one its own target. Restructuring first is tempting because it feels productive, and it wipes out the performance history you needed to make the earlier decisions.

How to set an ACoS target you can defend

1. Calculate break-even ACoS from gross margin. Break-even ACoS equals your gross margin percentage. At a 35% margin, a 35% ACoS is exactly break-even on the incremental sale.
2. Subtract the profit you actually want. If you want 15 points of contribution after ad cost, your target ACoS is margin minus 15, not whatever the category is doing.
3. Set different targets per product lifecycle stage. A launch ASIN buying reviews and rank can justify running above break-even for a defined period. A mature ASIN cannot.
4. Track TACoS alongside it. Total ad cost of sale measures ad spend against total revenue, including organic. It is the number that tells you whether ads are growing the business or just renting it sales.
5. Fix the account in that order: waste, bids, structure. Cut the converting-nothing spend first, correct bids second, restructure last. Restructuring first destroys the data you needed.

The mistakes that cost the most

• Optimising to a benchmark instead of a margin. Your break-even ACoS is your gross margin. Nobody else's average is relevant to it.
• Cutting ACoS by cutting the account. Pause everything except branded exact and your ACoS looks superb. So does your revenue chart, downward.
• Ignoring TACoS entirely. ACoS can improve while total revenue falls. TACoS is what catches that.
• Computing margin before FBA and referral fees. It produces a break-even target that is too generous, so you approve losses as wins for months.
• Reacting to keyword-level data that is too thin to read. Sponsored Products keyword data is sparse. Set a minimum click threshold before you allow yourself to act on a keyword.

What you can ask Muze to do

Muze connects Amazon Ads to ChatGPT, Claude or any MCP client with 48 tools, 12 read and 36 write. These are the real tool names behind each request.

• “Which Amazon campaigns are above my target ACoS this month?”
  Rolling spend, sales, orders, ACoS, ROAS, CTR and CVR per campaign, plus the daily account-level series behind it. Tools: list_amazon_campaigns, get_amazon_performance.
• “Set my monthly revenue estimate so TACoS is meaningful.”
  Stores the total-revenue figure Muze uses for TACoS. Muze holds the advertising campaign management scope, not SP-API, so it cannot see your marketplace revenue on its own. This is the honest workaround and we say so. Tools: set_amazon_revenue_estimate.
• “Cut the spend that produced no orders this month.”
  Recomputes search-term recommendations for a date range, lists them with row ids, and applies the ones you approve as harvested keywords or negatives. Tools: refresh_amazon_search_terms, list_amazon_search_terms, apply_amazon_search_term.
• “Compare Amazon ACoS against what Meta and Google returned.”
  Total ad spend across Meta, Google and Amazon against Shopify revenue, with blended MER, ROAS and ACoS. The tool reports its own caveats, including that Amazon marketplace sales are not in the revenue side. Tools: get_blended_performance.

Nothing spends without you

• Every write previews first and waits for explicit confirmation.
• New campaigns, ad groups and product ads are created paused. Negative keywords are the exception: they are created enabled, because an exclusion that is switched off is not an exclusion.
• Pause and archive are separate tools. Archiving is irreversible on Amazon, so it is never what happens when you asked to stop spend.
• OAuth only. No Amazon Advertising API application to file, no keys to manage, and access is revocable.
• Never a percentage of ad spend. You pay for the software. Your budget stays yours.

Frequently asked questions

What is a good ACoS on Amazon?

The only honest answer is: below your break-even ACoS, which equals your gross margin after all per-unit costs including referral and fulfilment fees. At a 35% margin, anything under 35% ACoS is contributing. Published category averages average across margins that have nothing to do with yours.

What is the difference between ACoS and TACoS?

ACoS is ad spend divided by ad-attributed sales. TACoS is ad spend divided by total revenue, organic included. ACoS judges a campaign. TACoS judges whether advertising is growing the business or subsidising flat sales.

Can AI optimise my Amazon ACoS?

It can do the parts that are analysis and mechanical execution. Muze reads ACoS per campaign, surfaces the search terms spending without converting, and can apply negatives, change keyword bids and adjust budgets over MCP. It previews every write and waits for your confirmation, so the judgement call stays with you.

How fast should I expect ACoS to move after a change?

Slower than you want. Amazon attribution runs on a delayed window, so sales land against clicks that happened days earlier. Judging a bid change after 48 hours mostly measures reporting lag.

Does Muze know my total Amazon revenue for TACoS?

No, and it says so in the tool description. Muze holds the advertising campaign management scope, not SP-API, so marketplace organic sales are invisible to it. You set the monthly total revenue figure yourself with set_amazon_revenue_estimate and TACoS is computed against that.

Should launch products run at a higher ACoS?

Often yes, as a deliberate and time-boxed decision to buy rank and reviews. The failure is not having a date on which the target tightens, so the launch ACoS quietly becomes the permanent ACoS.

Keep reading

The rest of the Amazon cluster