The Amazon search term report shows you the actual phrases shoppers typed before your ad was shown, alongside what each one cost and what it returned. It is the only place Amazon tells you this, and it is the single most useful report in the advertising console.
Download it from the Amazon Ads console under Measurement and Reporting, then Sponsored Ads reports. Choose the Search Term report, set the date range to at least 60 days, and pick Summary rather than Daily unless you need the day-level detail.
Two workflows turn that file into money: cutting the terms that waste spend, and promoting the terms that earn. Everything else people do with this report is decoration.
Search terms are not keywords, and the difference matters
This confusion costs sellers real money. I’d be precise about it.
A keyword is what you told Amazon to bid on. A search term is what the shopper actually typed. On an exact-match campaign these are nearly the same. On phrase and broad match they diverge enormously, and on automatic campaigns you did not choose keywords at all.
The report has a column for each. Reading the keyword column tells you what you asked for. Reading the search term column tells you what you got, and the gap between the two is where the waste lives.
I’d sort by search term rather than keyword every time. The keyword view hides the problem, because a keyword that looks efficient in aggregate can be funding a dozen irrelevant search terms underneath it.
Setting up the file before you start reading
Small thing, large payoff. The report arrives as a wide spreadsheet and most of it is noise for these two workflows.
I’d delete the columns you will not use, keep search term, keyword, match type, impressions, clicks, spend, orders and sales, then add two calculated columns of your own. Cost per order, which is spend divided by orders. And a simple flag for whether spend has exceeded your threshold with nothing to show for it.
That takes five minutes and turns an intimidating file into two sorted lists. In our experience the reason people download this report and never act on it is not that the work is hard. It is that the raw file is unpleasant to look at and the task gets postponed.
One more piece of setup worth doing once. Build a running negative keyword list in a separate sheet, with the date you added each term and the reason. Not because Amazon needs it, but because you will eventually wonder why a term is negated, and reconstructing that decision from memory a year later is impossible.
I’d also record the terms you deliberately chose not to negate, and why. Those are the relevant-but-not-converting ones, and they are a to-do list for the listing rather than for the campaign. Revisiting that list after a listing improvement tells you whether the improvement worked, which is a clean measurement you rarely get in advertising.
None of this is sophisticated. It is just the difference between a report you read and a report you use.
Workflow one: cut the waste
This is the faster of the two, and in my experience usually the more profitable.
Open the report and filter to search terms with spend above your threshold and zero orders. Set the threshold at roughly two to three times your target cost per acquisition, so you are only looking at terms that have had a fair chance and failed.
Read each one. You are sorting into three piles.
Genuinely irrelevant. Someone searching for a different product entirely. Add these as negative exact keywords immediately.
Relevant but not converting. The searcher wants your kind of product but is not buying yours. This is more interesting than it looks, because it usually means a price, image or review problem rather than a targeting problem. I’d note these rather than negating them, then look at why.
Relevant, low volume, no verdict yet. Not enough clicks to judge. Leave alone and check again next month.
The first pile is where the money is. In our experience a report that has never been mined has a meaningful share of spend sitting on terms nobody would deliberately bid on, and negating them is the closest thing to free money in an Amazon account.
Our guide on optimizing negative keywords covers the match-type mechanics of doing this without over-negating.
Workflow two: harvest the winners
The mirror image, and I’ve found it is the one people neglect.
Filter to search terms with orders and an acceptable cost per order. These are phrases that earned money while you were not specifically targeting them, and they usually surface from automatic or broad campaigns.
Move each one into an exact-match campaign or ad group of its own, where you can set the bid deliberately rather than inheriting whatever the broad match decided. Then add it as a negative in the campaign it came from, so the two are not bidding against each other.
That last step is the one people skip, and skipping it means your own campaigns compete for the same placement and push up your own cost.
There is a second benefit to harvesting that is easy to miss. Purchases teach Amazon that your listing is relevant to a phrase, which feeds organic ranking over time. Our guide to Amazon product ranking factors explains that relationship: ads do not directly boost organic rank, but the resulting sales do teach the association.
How often, and how far back
Monthly is right for most accounts, and I’d hold to it. Weekly is too frequent to accumulate meaningful data on anything but high-volume products, and quarterly lets waste run too long.
Pull at least 60 days. Thirty is tempting and tends to produce decisions on single-digit click counts, which is how sellers end up negating a term that would have converted on its fourteenth click.
I’d also keep the old reports rather than overwriting them. A term that converted twice last quarter and zero times this quarter tells you something a single snapshot cannot.
What the report cannot tell you
Three limits worth holding in mind, because I’ve watched each one mislead somebody.
It does not show organic search terms. This is an advertising report. Terms your listing ranks for without ads are invisible here, and you need Brand Analytics or Search Console-style data elsewhere for that picture.
It does not explain why a term failed. It tells you a term cost money and produced nothing. Whether that was price, images, reviews or genuine irrelevance is a judgment you have to make by looking at the listing.
It lags. Attribution takes time to settle, so the most recent few days will understate conversions. I’ve found sellers occasionally panic at a recent week that later fills in.
The mistakes I see most
Negating on too little data. One click and no order is not evidence. I’d want a term to have had several times your target cost per acquisition in spend before I killed it.
Negating broad instead of exact. A broad negative can silently block a family of terms including good ones. Exact negatives are precise and reversible.
Harvesting without negating the source. Covered above, and it is the most common structural error in this workflow.
Reading the keyword column and stopping. The whole value of this report is the search term column. Everything else is available elsewhere.
Doing it once. This is maintenance, not a project. A report mined in January describes January.
What good looks like after a few cycles
Expect the first pass to be the dramatic one. In our experience it usually is. There is usually accumulated waste sitting in an account nobody has cleaned, and removing it drops cost visibly within a couple of weeks.
After that the gains get smaller and steadier, which is the normal shape. The account stops leaking, the exact-match campaigns carry more of the converting volume, and the automatic campaigns settle into their proper job, which is discovering new terms rather than delivering sales.
That is the state to aim for: automatic campaigns as a discovery mechanism feeding exact campaigns that do the work. The search term report is the pipe between them, and in my experience an account run this way costs noticeably less per order than one where every campaign does everything.
FAQ
What is the Amazon search term report?
An advertising report showing the actual phrases shoppers typed before your ad appeared, with the spend, clicks, orders and sales each produced. It is the only place Amazon reveals real customer search terms rather than the keywords you chose to bid on.
Where do I download the Amazon search term report?
In the Amazon Ads console under Measurement and Reporting, then Sponsored Ads reports. Select the Search Term report type, set a date range of at least 60 days, and choose Summary unless you specifically need day-level detail.
What is the difference between a keyword and a search term?
A keyword is what you told Amazon to bid on. A search term is what the shopper actually typed. They are nearly identical on exact match and diverge widely on phrase, broad and automatic campaigns, and that gap is where wasted spend accumulates.
How do I use the search term report to lower ACoS?
Two workflows. Filter to terms with spend above two to three times your target cost per acquisition and zero orders, then negate the genuinely irrelevant ones. Then filter to terms with orders at an acceptable cost, move those into exact-match campaigns, and negate them in the campaign they came from.
How often should I review the search term report?
Monthly for most accounts, pulling at least 60 days of data each time. Weekly rarely accumulates enough clicks to judge anything, and quarterly allows wasted spend to run for too long before it is caught.
Why should I negate a harvested keyword in its original campaign?
Because otherwise your exact-match campaign and the broad or automatic campaign it came from bid against each other for the same placement, raising your own cost. Adding the term as a negative in the source campaign keeps them from competing.
Last updated: September 12, 2026. The Amazon Ads console reorganizes its reports periodically and available report types differ by marketplace and campaign type, so confirm current locations in the Amazon Ads console. ZonHack is an Amazon Ads verified partner and an Amazon SPN Verified Partner, and we manage PPC as a paid service, so treat this as informed but interested.
