ACoS answers for the campaign. TACoS answers for the business.

The separate-tool problem

Amazon is where ecommerce budget quietly goes to hide. The ad MCP category grew up around Meta and Google, because those are the APIs everyone integrates first, so Amazon ends up in a different tool with a different login, a different reporting cadence and often a different person looking at it.

That arrangement works right up until you have to decide where the next thousand dollars goes. Then you need Amazon spend, Meta spend, Google spend and store revenue side by side, and you are back in a spreadsheet reconciling exports. Muze covers Amazon in the same connection as the other three. get_blended_performance already includes Amazon in total ad spend when it compares against Shopify revenue, so the cross-channel question is one call rather than a reconciliation project.

Search terms are the weekly job

Most of the recoverable money in a Sponsored Products account sits in the search term report, and most of it stays there because reading it is dull. Muze turns that into a loop with an explicit gap between finding and doing.

refresh_amazon_search_terms recomputes recommendations for a date range and applies nothing. list_amazon_search_terms returns the pending rows, split between terms worth harvesting as keywords and terms worth blocking as negatives, each carrying a row id. apply_amazon_search_term acts on exactly one row and mutates the live account. dismiss_amazon_search_term clears the ones you disagree with so they stop reappearing. You can expand from a seed with get_amazon_keyword_suggestions, which reads Amazon’s own search bar completion API rather than inventing plausible keywords, and create exclusions in bulk with create_amazon_negative_keywords using negativeExact or negativePhrase match.

TACoS needs a number Amazon does not have

ACoS is ad spend divided by ad-attributed sales. It tells you whether a campaign paid for itself and nothing at all about whether advertising is growing the business. TACoS is ad spend against total revenue, and Amazon’s own reporting cannot compute it, because it does not know your total.

Muze handles this the honest way. set_amazon_revenue_estimate stores the monthly total-revenue figure you provide, and TACoS is calculated against that. It is your number, held on the Muze side, and it stays visible as an input rather than being presented as something we discovered. Alongside it, get_amazon_performance returns a daily series of spend, sales, orders, ACoS, ROAS, CTR and CVR for a profile, with the ad-attributed-only caveat stated on the tool itself rather than buried in a footnote.

Bids, with a clip range

run_amazon_bid_autopilot runs the bid adjustment pass once, now, for one profile, and it moves live bids only inside a guardrail clip range of 0.65x to 1.25x. That range is the entire point. An automated bid change that can go anywhere is a way to lose a week of budget to one noisy signal. Bounded moves are boring and survivable.

The deliberate single changes have their own tools: update_amazon_keyword_bid, update_amazon_ad_group_bid, update_amazon_auto_target_bid and update_amazon_product_target_bid. list_amazon_autopilot_actions is the audit log of everything autopilot did on the profile, which is what you read when a number moves and you want to know whether you caused it.

More on Amazon Ads in Muze

What you would actually type

list_amazon_profiles resolves your profile id, so you never paste one. Everything below is a plain question.

• “Which Amazon search terms spent money this month without converting? Add the worst as negatives, and show me the list first.”
• “What is my ACoS by campaign over the last 30 days, and which campaigns are above target?”
• “Build a manual campaign for this ASIN with these keywords and leave it paused.”
• “Compare Amazon spend and sales against what Meta and Google returned last month.”
• “Harvest the converting search terms from the last 14 days into exact match keywords.”

Nothing changes without a confirmation

• Every write previews first. Bid changes, negatives and budget edits describe the change before they execute.
• New campaigns are created paused. Build the structure, review it, then enable it yourself.
• Recommendations are generated, not applied. refresh_amazon_search_terms produces recommendations and changes nothing. Applying one is a separate, explicit call on a single row.
• Bid autopilot is clipped. Live bid moves stay inside a 0.65x to 1.25x guardrail range.
• OAuth, with the API work already done. No Amazon Advertising API application to file and no developer account to register on your side.

What it costs

The free tier is 25 read-only tool calls per month, enough to pull performance, list campaigns and read your pending search term recommendations before deciding anything. Write actions, which on Amazon is most of the surface, require a paid plan.

Muze never takes a percentage of ad spend. Amazon budget is spent through your own Amazon Ads account exactly as it is today.

Frequently asked questions

Is there an MCP server for Amazon Ads?

Yes. Muze covers Amazon Ads with 48 tools, 12 read and 36 write, in the same connection as Meta Ads, Google Ads and Shopify. Most ad MCP servers stop at Meta and Google, so Amazon is one of the harder platforms to reach from an AI assistant.

Do I need my own Amazon Advertising API access?

No. You authorize over OAuth with your normal Amazon Ads login. The API approval work is already done on our side, so there is no developer application to file.

Can it calculate TACoS?

Yes, against a figure you provide. set_amazon_revenue_estimate stores your monthly total-revenue estimate and TACoS is computed from it. Amazon cannot do this on its own because it does not know your total revenue, so the input has to come from you.

Can it manage bids automatically?

It can run a bid adjustment pass on demand, and live bid moves are clipped to a 0.65x to 1.25x range. Individual bid changes at keyword, ad group, auto target and product target level are separate tools. Automatic never means unattended: writes preview and every autopilot action is logged.

Can I see Amazon next to Meta and Google?

That is the main reason to use one connection. get_blended_performance sums Meta, Google and Amazon spend and compares it against Shopify revenue, so you can rank channels rather than reconcile four dashboards by hand.

Which Amazon campaign types are covered?

The tool set covers campaigns, ad groups, product ads, keywords, negative keywords, product targets and negative product targets, plus portfolios, budget schedules and reports. Check the full tool reference for the exact operation you need before assuming a specific campaign type is supported.

What does it cost?

The free tier includes 25 read-only tool calls per month. Write actions require a paid plan, and Muze never takes a percentage of ad spend.

Keep reading

Other segments