Amazon Wasted Ad Spend: How to Find the Search Terms That Never Sell

Wasted ad spend is the money you pay for clicks on search terms that never sell, or that sell at a loss. You find it in your Sponsored Products search term report by judging every search term on its own numbers inside each campaign. Flag three kinds of term: relevant terms with zero orders after three times the clicks your product normally needs to make one sale; irrelevant terms with zero orders after two clicks; and terms with three or more orders whose ACOS sits above 1.3 times your break-even. Negate those where they bleed, and leave everything else alone until it has enough data.

Most sellers know they have waste. The hard part is deciding when a search term has had a fair chance. Cut too early and you kill terms that would have sold. Cut too late and the bleeding becomes a cost you pay every month without noticing.

Below is the process we use in the accounts we manage: where to pull the data, the three rules and the math behind each number, a worked example, and the mistakes we see most often in audits.

What counts as wasted ad spend

Not every click that fails to convert is waste. Some clicks are the price of finding out whether a term works. Waste is the spend that keeps going after the data has already given you the answer. In practice it falls into three groups.

• Relevant terms with zero orders. The shopper searched for something close to your product, clicked, and did not buy, again and again. Usually it is a price, size or feature mismatch your listing cannot fix.
• Irrelevant terms. Searches for a different product that Amazon matched to your broad, phrase or auto targeting. These rarely sell, and you should not pay to prove it.
• Converting terms that lose money. They sell, but each sale costs more in ads than the product earns. They are easy to miss because they show orders.

The first two are pure waste. The third is a judgment call, and in larger accounts it is often where most of the money hides, because a term with orders looks like it is working.

Where to find the search terms that never sell

Amazon gives you two ways to see what shoppers actually typed before they clicked your ad.

1. The search term report. In the advertising console, open Sponsored ads reports, choose Sponsored Products and the Search term report. Each row shows a customer search term with the campaign, ad group and keyword or target that triggered it.
2. A bulk file with search terms. When you download a bulk operations file, tick the option to include search term data. You get the same numbers in a sheet next to your campaigns, bids and existing negatives, which makes it much easier to see what is already negated.

Two things matter before you read a single row. First, the search term report only looks back about 65 days, so pull it regularly or you lose history. Second, for sellers, Sponsored Products credits a sale to an ad click for up to 7 days after the click. Sales from the last week are still arriving, so a term that shows zero orders today may have two by next week. Use 30 to 60 days of data and leave out the most recent 7.

For a quick read on how big the problem is, put your total spend and your zero-order spend into our free wasted spend calculator. It shows the yearly cost and what share of your budget it is. Then use the rules below to decide which terms to cut.

Judge each search term inside each campaign

The most common shortcut is to add up a search term across the whole account and make one decision. That hides what is going on. The same term can lose money in a broad campaign and make money in its own exact campaign, where the bid and placement are set for it.

So judge each term on its own numbers inside each campaign, and negate it only in the campaign where it bleeds. Two filters before you start:

• Leave out paused campaigns, ad groups and targets. Their old spend is still in the report and will make your waste look bigger than it is.
• If the search term is the campaign's own exact keyword, do not negate it. That is a bid or pause decision on the keyword itself.

The three rules for finding waste

Each rule depends on two numbers you work out per product: its break-even ACOS, and how many ad clicks it needs, on average, to make one sale.

Rule 1: Relevant terms with no orders after 3x your clicks per order

Clicks per order is total ad clicks divided by total ad orders for that product. If your product converts 12% of ad clicks, it needs about 8.3 clicks per sale. Three times that is 25 clicks. A relevant search term with 25 clicks and no orders gets negated.

Why three times and not one? Because sales arrive unevenly. A search term that converts at exactly your normal 12% will still go 8 clicks without an order 36% of the time. Negate at 8 clicks and you throw away about one in three good terms. At 25 clicks, the chance that a good term shows zero orders by bad luck falls to about 4%.

This is also why a fixed rule like "negate after 10 clicks" fails. The right number depends on how well the product converts.

Look at the last column. Three times clicks per order gives you about the same small risk of a wrong call whatever your conversion rate. A fixed 10-click rule on a product that converts at 5% would cut a good term about 60% of the time.

Rule 2: Irrelevant terms with no orders after 2 clicks

Some terms have nothing to do with your product. A garlic press showing for "garlic powder" is not a test worth funding. If a term is clearly about a different product and has no orders, negate it at 2 clicks.

The judgment is in the word "clearly". Misspellings of your product, close synonyms and the words a keyword was built around are relevant, even when they look odd. If the search term contains every word of the keyword that triggered it, you were targeting it on purpose, so treat it as relevant and let Rule 1 decide.

Rule 3: Converting terms with ACOS above 1.3x break-even

This rule needs your break-even ACOS, the ACOS at which an ad sale makes zero profit. For the $29.99 kitchen product in our profit calculator, profit before ads is $10.95, so break-even ACOS is 36.5%.

Flag a term when it has 3 or more orders in that campaign and its ACOS is above 1.3 times break-even. For this product that threshold is 47.5%. The 30% buffer stops you cutting terms that are only a little over break-even, where a small bid change would fix them.

Terms with one or two orders above the threshold stay for now. One order is too little data to tell you a term's real ACOS.

Before you negate a converting term, work out the bid at which it would sit exactly at break-even:

Ideal CPC = current CPC × break-even ACOS ÷ current ACOS

If the term matters to you, for example a main keyword you are pushing for organic rank, lower its bid toward that ideal CPC or move it into its own exact campaign at that bid instead of cutting it. A ranking push can carry an ACOS of up to about three times break-even on purpose, for a limited time. Waste is spend without a reason. It is not every ACOS above break-even. If you are not sure why a term's ACOS rose, read why ACOS goes up before you touch it.

A worked example

Take the same $29.99 kitchen product, a garlic press, and one of its broad campaigns over 60 days, with the last 7 days left out. Across all its campaigns the product converts 12% of ad clicks, so its no-order limit is 25 clicks. Its break-even ACOS is 36.5% and its Rule 3 threshold is 47.5%. The average CPC in this campaign is $1.10.

At 44.2% the campaign is above break-even, so every sale it makes loses money after ads. The campaign converts 8.3% of clicks, lower than the product's 12%, which is normal for a broad campaign.

Two terms are clear negatives today: "garlic crusher" under Rule 1 and "garlic powder" under Rule 2. Together they cost $37.40 and brought in nothing. Negating them in this campaign brings its ACOS down to 37.3%.

"Ginger grater" is the Rule 3 case. At 50.1% ACOS, its ideal CPC is $1.10 × 36.5% ÷ 50.1% = $0.80. A garlic press is not what a ginger grater shopper is looking for, so there is no ranking reason to keep it here. Negating it as well removes $45.10 of spend and $89.97 of sales, and the campaign lands at 34.7%, now under break-even.

"Garlic mincer tool" and "garlic slicer" stay. One is 7 clicks short of the limit and the other has a single order. Check them again next time.

In total, the flagged terms took 34.6% of this campaign's spend and produced 16.7% of its sales. Zero-order terms alone were 24.0% of spend, which our calculator would call a serious leak. The $82.50 freed up over these 60 days does not have to vanish from the budget. Move it to the terms that are working, like the two keepers above, ideally in their own exact campaigns.

What this looks like across a whole account

One campaign is easy to read. Across 50 or 200 campaigns, owners need the answer in ten seconds. In our audits we sum it up in two lines:

• Spend on targets with zero orders and more than 30 clicks, as a share of total spend.
• Spend on targets with ACOS above 80%, as a share of total spend, next to the share of ad sales they bring in.

When the share of spend is far bigger than the share of sales, you have found budget you can move. Then watch TACOS for the next few weeks: if total sales hold while ad spend drops, the waste was real.

Cutting waste is rarely the whole job, but it is usually the first one. For one EU brand we manage, ad spend fell 37%, from EUR 116,514 to EUR 73,252, while sales rose 4.7% and net profit went from EUR 56,520 to EUR 111,184, up 96.7%. The details are in our case studies.

Mistakes we see

• Negating after a fixed number of clicks. "10 clicks and out" ignores your conversion rate. On a product that converts at 5%, it cuts good terms more often than it keeps them.
• Judging the last 7 days. Sales are still being credited to clicks from last week. A term with zero orders on Monday may have orders by Friday.
• Negating account-wide what only fails in one campaign. A term that wastes money in an auto campaign can be profitable in its own exact campaign. Negate it where it bleeds, not everywhere.
• Using negative phrase too widely. A negative phrase blocks every search that contains those words in that order. Use negative exact by default, and save phrase for words that are never right for your product, like "powder" for a garlic press.
• Counting paused targets. Old spend from paused keywords still sits in the search term report. Include it and your waste looks bigger than it is, and you may negate terms that no longer run.
• Cutting ranking keywords on ACOS alone. A main keyword in a ranking push is supposed to run above break-even for a while. Diagnose before you cut. Our PPC management work starts from ranks and break-even together for this reason.
• Never coming back. New search terms show up every week in broad, phrase and auto campaigns, and the report only keeps about 65 days. Waste you cleaned up once grows back if nobody looks.

Get your waste found for you

If your ad spend keeps climbing while sales sit flat, wasted spend is the first place to look, and usually the cheapest fix. Start with the wasted spend calculator to see the size of the leak.

Or let us do it properly. Our free Profit Audit, normally $500 and free until 31 December 2026, goes through your search terms campaign by campaign, applies these rules with your real break-even, and shows you which spend to cut and where to move it. We manage more than $680K in ad spend across 25 live brands, and this is one of the first things we check in every account. Prefer to talk it through first? Book a free 30-minute call.