Muze AIIn short
Wasted ad spend is money going to ads, keywords or placements that are not producing conversions at or above your break-even return. To find it: set your break-even ROAS, which is 1 divided by your gross margin, pick a single date window and use it everywhere, then list every campaign, ad and keyword with its spend and conversions. Rank by spend that produced zero conversions, then by spend running below break-even. In most accounts the three biggest leaks are ads still running with no conversions, search terms you never intended to buy, and campaigns whose reported ROAS looks fine but sits under the margin they need.
Everyone agrees they have wasted spend and almost nobody has a number for it, because finding it means four tabs, three date ranges and an export. You open Ads Manager, then Google Ads, then Amazon, and the three of them quietly disagree about what last week even was.
This is a mechanical job, which is why it is worth having a fixed procedure rather than a vibe. The pass below takes about half an hour by hand, and it finds the same money whether you run it in dashboards, in a spreadsheet, or by asking an assistant connected to your accounts. Run it weekly and it stops being a project.
The seven-pass sweep
Set the threshold before you look
Calculate your break-even ROAS: 1 divided by your gross margin. At a 40% margin that is 2.5. Anything below it is losing money no matter how healthy the dashboard looks. Write the number down before you open any report, so the data cannot talk you out of it.
Most people skip this and end up judging campaigns against a number they half-remember from a podcast. Your break-even is specific to your margin, your shipping and your returns rate, and it is the only threshold that means anything. Two accounts with identical 2.2x ROAS can be one profitable business and one that is paying for the privilege of shipping boxes.
Pick one window and use it everywhere
Choose 7, 14 or 30 days and apply the identical window to every platform. Most cross-platform confusion is a date range problem, not a data problem. Note whether each platform is reporting on click date or conversion date, because they do not agree by default.
List every line with spend and conversions
Go one level deeper than campaign. Pull ad-level data on Meta, keyword and search term level on Google, and keyword and product target level on Amazon. Campaign-level averages hide the losers inside winners, which is exactly where waste survives.
Rank by zero-conversion spend first
Sort by spend where conversions are zero over the window. This is the cleanest money in the account: no attribution argument, no judgement call, just spend with nothing on the other side of it. Check the line has had enough spend to be judged before you kill it.
The trap is killing things too early. An ad that has spent less than your target cost per acquisition has not failed yet, it has just not had a chance. A reasonable rule is that a line item is only judged once it has spent more than one target CPA, and that low-volume accounts need longer windows before the numbers mean anything.
Read the search terms report
On Google and Amazon, look at the actual queries you paid for rather than the keywords you chose. Sort by spend, scan for intent that is not yours, and add the obvious offenders as negatives. Broad match and auto targeting both drift, so this is a recurring job, not a one-off cleanup.
Check overlap and fatigue on Meta
Look for audiences bidding against each other, near-duplicate ad sets, and rising frequency on the audiences you scaled. Waste on Meta is less often an obvious zero and more often the same person being shown the same ad until they stop caring.
Fix the biggest item, then re-measure
Work in order of money at stake, not in order of how annoying each item is. Pause the largest zero-conversion line, add the negatives, then leave the rest alone for a week so you can tell what your change did. Fixing six things at once means learning nothing from any of them.
Why the dashboards make this hard on purpose
Ad platforms report on themselves. Each one counts conversions it believes it caused, inside its own attribution window, and each one is graded by the same company selling you the inventory. Add the three revenue figures together and you will comfortably exceed what your store actually made.
That is why the ranking above is built on spend and zero conversions rather than on reported ROAS. Spend is the one number every platform agrees on, because it is the number they are charging you. Once you have dealt with the unambiguous waste, move to the judgement calls with a blended view instead of platform-reported revenue.
Making it a weekly habit rather than a quarterly project
The reason wasted spend accumulates is not ignorance. It is that the sweep takes half an hour, produces a list of small dull fixes, and always loses to something more urgent. Anything that only happens when someone feels guilty about it is not a process.
This is the first job worth handing to an assistant that can read your accounts, because it is entirely mechanical and completely repetitive. Connect Meta, Google and Amazon and you can ask for the same ranked list every Monday in one sentence. On the paid plans the fixes can be prepared too, and every one of them previews before it runs.
Ask for the sweep in one sentence
- “Audit Meta, Google and Amazon for the last 14 days. Rank every ad, keyword and target by spend with zero conversions.”
- “Which campaigns are running below a 2.5 ROAS this month, and how much are they spending?”
- “Show me the Google and Amazon search terms that spent money with no conversions last month.”
- “Prepare the negative keyword additions for those terms and show me the list before applying anything.”
Frequently asked questions
- What counts as wasted ad spend?
- Spend that produces no conversions, or conversions at a return below your break-even ROAS. Break-even ROAS is 1 divided by your gross margin, so at a 40% margin you need 2.5x just to cover costs. A campaign at 2x with a 40% margin is losing money even though the dashboard shows a positive return.
- How long should I wait before pausing something?
- Until the line item has spent enough to be judged, which usually means more than one target cost per acquisition. Pausing at half your target CPA is not discipline, it is guessing. Lower-volume accounts need longer windows before the numbers mean anything.
- How often should I run this?
- Weekly for accounts with meaningful daily spend, monthly for smaller ones. Broad match and auto targeting drift continuously, so search terms in particular need repeat attention rather than one cleanup.
- Why do the platforms disagree about my results?
- Because each one self-reports inside its own attribution window and counts conversions it believes it caused. Adding Meta, Google and Amazon revenue together double-counts. Use spend for ranking waste, and use store revenue for judging total performance.
- Should I pause losers or fix them?
- Pause first, then investigate. A zero-conversion ad costs money every day it stays live while you decide, and pausing is reversible. Restructuring on the other hand takes a week and might be the wrong answer.
- Can an AI assistant do this for me?
- The reading and ranking, yes, if it is connected to your accounts through something like the Muze MCP. The fixes require a paid plan, and every change previews first and waits for your explicit confirmation before it runs.
Keep going
Ask for it instead of clicking through it
Connect Meta Ads, Google Ads, Amazon Ads and Shopify to ChatGPT or Claude. The free tier is 25 read-only tool calls a month, every write previews before it runs, and new campaigns are created paused.
Start free