Most brand owners open the search term report, scroll for thirty seconds, feel overwhelmed, and close it. That report is the single most honest document Amazon gives you. It shows the exact phrases real shoppers typed before they clicked your ad, and what happened next. Every wasted dollar and every undiscovered keyword is sitting in those rows. The problem is not access. The problem is that almost nobody reads it like a strategist, with a decision waiting at the end of every line.
A strategist does not "look at" the report. They run every term through a sorting machine. Each search term gets pushed into one of four buckets: harvest, negate, promote, or watch. Once you have that machine in your head, the report stops being a wall of numbers and becomes a to-do list. Here is exactly how to build it.
Pull the right report and set it up to read
Inside Campaign Manager, go to the sponsored ads reports section and run the Search Term Report for Sponsored Products. Pull at least 60 days of data. Anything shorter and your low-volume terms will not have enough clicks to judge fairly.
The columns that matter are customer search term, impressions, clicks, spend, orders, sales, ACoS, and click-through rate. Sort by spend, highest first. This is the most important move in the whole process. Sorting by spend puts your money where your eyes are. The terms burning the most budget rise to the top, and that is where the fastest wins and the fastest leaks live.
Before you judge anything, decide on your break-even ACoS. That is the ACoS at which an ad-driven sale earns zero profit. If your break-even is 35 percent, then a term at 50 percent ACoS is losing money and a term at 20 percent is printing it. Without that number, every decision below is a guess.
The four buckets every search term falls into
Read down the list and assign each term to one bucket. Do not overthink it. The rules are mechanical on purpose.
Harvest: winners hiding inside broad and auto campaigns
A harvest term is a customer search term that is converting well inside a broad, phrase, or auto campaign, where you are not bidding on it directly. It has orders, its ACoS is at or below your target, and it has enough clicks to trust (10 to 15 clicks with at least a couple of orders is a reasonable floor for most brands).
You harvest it by adding it as an exact match keyword in a dedicated campaign, where you control the bid and the budget. This is how you stop renting a winning keyword through a loose match type and start owning it. The discovery campaign keeps finding new terms; the exact campaign scales the proven ones.
Negate: spend with nothing to show for it
A negate term has real spend, real clicks, and zero or near-zero orders. It is the leak. A term with 40 clicks, 30 dollars spent, and no sales is telling you plainly that this shopper is not your shopper. Add it as a negative exact keyword in the campaign where it is bleeding.
Be specific here. Negate the exact term, not a whole theme, unless the theme is clearly wrong (a competitor's brand name, the wrong product type, the wrong size). Aggressive broad negatives can choke off relevant traffic you have not discovered yet.
Promote: proven terms that deserve more room
A promote term is one you already target that is converting below your ACoS target but is capped by budget or a low bid. It is winning on the margin and asking for more. Raise the bid in steps, or move it to its own campaign with a clean budget so it is not competing with weaker keywords for the same dollars. Promotion is how good terms become great ones.
Watch: not enough data to act
A watch term has some signal but not enough volume to commit. Five clicks and one order is interesting, not conclusive. Leave it running and check it next cycle. The discipline is resisting the urge to kill or scale a term before it has earned a verdict. Watching is a decision, not indecision.
A search term you cannot assign to a bucket is not confusing. It just does not have enough clicks yet, and your job is to wait, not to guess.
Read the numbers in context, not in isolation
ACoS alone will lie to you. A 60 percent ACoS on a brand-new product you are launching can be the correct cost of buying ranking. The same 60 percent on a mature, profitable hero product is a leak. Always read a term against what you are trying to do with that product.
Click-through rate and conversion rate tell you where the problem lives. High clicks and no sales point at the listing, not the keyword. If shoppers are clicking a relevant term and not buying, the term is fine and the page is failing. That is a listing problem, and our guide on why your Amazon listings need optimization walks through how to spot and close those conversion gaps before they drain your ad budget.
Low click-through on a term you think is perfect is a relevance or creative signal. The shopper saw your ad and scrolled past. That points at the image, the title, or the price, not the bid.
Turn the buckets into a weekly routine
The strategists who win with this report are not smarter. They are more consistent. Reading the report once does nothing. Running the same loop every week compounds.
Here is the rhythm. Each week, pull the trailing 60 days, sort by spend, and work the top terms first since that is where the dollars are. Harvest the proven winners into exact campaigns. Negate the confirmed leaks. Promote the capped winners. Move everything ambiguous to watch and revisit it next week. The discipline of cutting waste while protecting your best keywords is the same one we lay out in our framework for lowering ACoS without killing your sales, and the search term report is where that framework gets real.
This loop also feeds your broader strategy. The terms shoppers actually use reveal demand shifts, new competitors, and emerging product angles long before they show up in your sales. Pattern-watching across these reports is one reason the brands adapting fastest in 2026 keep pulling ahead, because they treat search data as a forward signal, not a rearview report.
What to do this week
Open Campaign Manager and run a 60-day Sponsored Products Search Term Report today. Sort by spend, highest first. Take only the top 20 terms and assign each one a single label: harvest, negate, promote, or watch. Then make those 20 moves. Add the negatives, build one exact-match harvest campaign, and bump the bids on your capped winners.
Twenty terms is enough to find real money and few enough that you will actually finish. Next week, do the next 20. That is the whole job. The report was always telling you what to do. Now you know how to read it.