Search terms that spend with zero sales are the easy case: you negate them. The harder, more expensive case is search terms that do convert, just at an ACoS that eats your margin. You do not want to kill these terms, because they bring real orders. You want to pay less for the same sales. This SOP shows you how to find them and fix them.
When to use it: monthly, or whenever overall ACoS creeps up without an obvious cause.
Step 1: Know your ceiling before you filter
"High ACoS" only means something against a target. Use your break-even ACoS for the product, ideally the LTV-adjusted one from the Break-Even ACoS page of the LTV & Subscriptions report, since a product with strong repeat purchases can afford a higher ACoS on acquisition. See "Break-Even LTV ACOS. How to Define the Right ACOS Target".
Step 2: Find the terms
Open Advertising Analytics and go to the Keywords and Search Terms page.
Set the date range to the last 30 to 60 days so each term has enough data to judge.
In the Search Term Analyzer, filter by ACoS above your target. Enter the value as a decimal: 0.7 means 70%.
Sort the result by spend, descending. The top of this list is where your margin is going.
Exclude your branded terms from this review. They convert at artificially strong rates and belong in their own playbook (see "How to Optimize Your Branded Ad Spend").
Step 3: Diagnose why each term is expensive
For each big spender, expand the row: click (+) to see the ad type and match type, again for the campaign, and a third time for the individual search terms. Then ask, in order:
Is the CPC the problem? The term converts, but each click costs too much for your price point. This is a bid problem, not a relevance problem.
Is the same term bought twice? Check the Keywords and Search Term Gaps view. A term converting in both an auto or broad campaign and an exact campaign can bid against itself and inflate cost.
Is one placement draining it? On the Placement Insights page, view the keyword by placement. Top of search sometimes costs several times more per order than rest of search for the same term.
Is conversion below market? If the term is relevant but your conversion rate lags, the listing is the real problem, and no bid change fixes that. Diagnose it with the Market Share & Funnel report ("How to Identify Keywords Fully Supported by PPC and Fix Them").
Step 4: Apply the matching fix
Bid problem: lower the bid in steps of 10 to 20% rather than pausing. The term converts, so the goal is a cheaper click, not zero clicks.
Duplication: add the term as exact match in its own campaign at a controlled bid, and negate it in the auto or broad campaign that was buying it expensively.
Placement problem: reduce the top-of-search multiplier and let rest-of-search carry more of the volume.
Conversion problem: fix the listing first. Cutting the bid on a relevant term with a weak listing just hides the symptom.
Step 5: Re-measure in two weeks
Re-run the same filter after two weeks. You are looking for the same orders at a lower ACoS. If orders dropped meaningfully, you cut too deep: raise the bid halfway back and re-check. For bulk changes, export the filtered view via More Options and upload to Seller Central.
Tip: the N-gram Analysis view catches patterns single terms hide. If one word keeps appearing across your expensive converting terms, that word is telling you something about which traffic converts poorly for your product.