Your Amazon search term report shows exactly which shopper searches your ads paid for and which of them have never once produced a sale. Here is how to read it, what to cut, what to keep, and the negative keyword system that stops the waste coming back.
Amazon PPC
The Amazon search term report is the only file that shows you what shoppers actually typed before they clicked your ad, as opposed to what you told Amazon to target. Download it, sort by spend, and look for search terms with clicks and no orders. In the accounts we take over, that one column usually accounts for 15% to 35% of the monthly budget.
We manage $112M in ad spend across 200+ brands and 16 marketplaces, and the free audit we run always starts in the same place: this report. Not the campaign view, not the keyword view, the search term view. Campaign level numbers tell you an account is losing money. Search term level numbers tell you exactly which searches are doing the losing, which is the only version of the information you can act on.
This guide covers what the report is, how to pull it, how to read it without drowning in rows, the negative keyword system that keeps waste out permanently, and the harvesting half that most sellers skip entirely.
What is on this page
- What the search term report actually shows
- How to download it in under two minutes
- How to read it without drowning
- The four boxes every search term falls into
- Negative keywords: the system, not the one off
- Harvesting: the profitable half nobody does
- Six mistakes we see in almost every account
- A cadence you can actually keep
What the search term report actually shows
There are two different things in your advertising account that sound identical and are not. A keyword is what you told Amazon to target. A customer search term is what a real person typed into the search box before your ad appeared. With exact match those two stay close together. With phrase and broad match, and with automatic campaigns, they drift, sometimes dramatically.
That drift is the point of the whole system. Amazon uses your broad keywords as a starting suggestion and then serves your ad against anything it judges related. Sometimes that is brilliant: it finds you a phrase you would never have thought of, which converts at twice your average. Sometimes it puts a $60 electric kettle in front of somebody searching for kettle bells. Both outcomes come from the same mechanism, and the search term report is where you separate them.
One more thing worth being clear about, because it costs people money. Exact match is no longer literally exact. Amazon matches close variants, so plurals, small misspellings and reordered words all still trigger an exact match keyword. Your exact campaigns need checking too, just less often.
How to download it in under two minutes
In Seller Central, open the advertising console, then Measurement and Reporting, then Sponsored ads reports. Create a report, set the report category to Sponsored Products, set the report type to Search term, set the time unit to Summary, and set the reporting period to the last 60 days. Choose the spreadsheet format and run it. Vendor Central sits in a slightly different menu but the report itself is identical.
Two settings matter more than people realise. Summary rather than daily, because daily rows split six clicks across six lines and turn a readable file into noise. And 60 days rather than 30, because a term with nine clicks in a month often has twenty five in two, which is the difference between a guess and a decision.
Amazon reports search term data with roughly a 48 hour delay, so the last two days of any window are incomplete. Pull on a Monday and read to the previous Friday.
Those first two figures are the range across accounts we have audited, not a national statistic, and the spread is wide for a reason. Accounts running mostly exact match with a maintained negative list sit near the bottom. Accounts that switched on automatic campaigns eighteen months ago and never looked again sit near the top, occasionally above it.
How to read it without drowning
A 60 day file for a mid sized account can run to eight thousand rows. Nobody reads eight thousand rows, and nobody needs to. Do this instead.
Sort the whole sheet by spend, highest first. Now add a filter on the orders column for zero. What you are left with is every search term that took money and returned nothing, ordered by how much it took. The top thirty rows of that view typically hold half of all the waste in the account. Thirty rows is a job you can finish before your coffee goes cold.
Then flip it. Filter for orders greater than zero and sort by ACOS, lowest first. Those are your harvest candidates, and we come back to them below.
What you are hunting for in the zero orders view falls into recognisable families. Competitor brand names you never meant to bid on. Search terms with the wrong intent entirely, such as replacement parts when you sell the whole unit. Terms describing a size, colour or pack count you do not stock. Research phrases such as how to or vs, where the shopper is reading rather than buying. And the quiet killer: terms that match your product category perfectly but describe a cheaper tier of it.
The margin arithmetic that decides everything. Before you cut a single term, work out your break even clicks. Take your profit per unit, divide by your average cost per click, and you have the number of clicks you can afford before a sale must land. Sell at a $12 profit with a $0.80 average cost per click and that is 15 clicks. A term with 6 clicks and no sale is not yet a failure. A term with 40 is not a maybe.
Not sure which of your search terms are actually losing money?
Send us the last 60 days and we will mark up your own report, free, with the terms we would cut first.
The four boxes every search term falls into
Once you stop thinking of the report as a spreadsheet and start thinking of it as a sorting exercise, it gets much faster. Every row is one of four things, and each of the four has exactly one sensible response.
The box people get wrong is the bottom left one. A term with three clicks and no sale feels like waste because it produced nothing, and cutting it feels productive. It is not. Three clicks is a coin toss, and if you negate every three click term you will eventually negate a term that would have carried a product. Patience there is not laziness, it is statistics.
The box people underuse is the top left. A term selling at good ACOS inside a broad or automatic campaign is being bid on by accident. You are paying whatever Amazon decides that keyword is worth in a group with twenty other keywords. Pull it out, give it its own exact match keyword and its own bid, and you get control over the thing that is already working.
That curve is the reason this work pays. Ad sales concentrate: a small group of terms carries most of the revenue, and a very long tail of terms carries almost none while still charging you for clicks. Your job is not to optimise all eight thousand rows. It is to protect the head of the curve and stop funding the tail.
Negative keywords: the system, not the one off
Most sellers add negatives the way people clear out a garage: in one furious afternoon, once a year. Then twelve months of new search terms pile up again. The accounts that stay efficient treat negatives as a standing weekly habit, and they use the right type in the right place.
| Type | What it blocks | Use it for | Risk |
|---|---|---|---|
| Negative exact | That one search term, nothing else | Single losing terms you found in the report | Low. Safe default. |
| Negative phrase | Any search containing those words in that order | Whole themes: competitor brands, the word free, wholesale, cheap | Medium. Can block winners hiding inside the phrase. |
| Negative product | A specific ASIN placement | Product pages where you never convert | Low, and badly underused. |
| Campaign level | Everything in that campaign | Terms you never want anywhere in that strategy | Medium. Applies broadly, so check it twice. |
A structure that works: put brand protection negatives at campaign level so they apply everywhere, keep tactical loser negatives at ad group level where you can see them in context, and add negative products for the ASINs that eat display budget without ever converting. Then review the list quarterly and remove anything that no longer makes sense, because a negative list nobody prunes eventually strangles the account it was meant to protect.
There is one thing to be careful of, and it is the reason we never hand a client a bulk cut list without reading their organic data first. Some search terms convert slowly but reliably, and those slow sales feed the sales velocity that drives your organic rank on that term. Cut them and the ACOS chart improves while total sales quietly fall. We wrote about that trade off in detail in how to cut Amazon ACOS without losing your rank, and it is the single most common way a tidy up backfires.
Want the cut list without the risk of cutting the wrong thing?
We read the account properly, separate spend that buys rank from spend that only buys clicks, and write it up.
Harvesting: the profitable half nobody does
Cutting waste protects your margin. Harvesting grows your sales. Most sellers do the first and skip the second, which is like weeding a garden and never planting anything.
Take your report, filter for orders greater than zero, sort by ACOS lowest first, and look at where those winners currently live. If a term is converting well inside an automatic campaign or a broad match ad group, it is being bid on generically. Promote it: create an exact match keyword in a dedicated campaign, set a bid you have chosen deliberately, and add that same term as a negative exact back in the campaign it came from so the two do not compete with each other for the same click.
That last step is the one people forget. Without the negative, your automatic campaign and your new exact campaign both enter the auction for the same search, Amazon picks one, and your data splits across two places. Harvest, then block the source.
Do this consistently and the shape of the account changes over a few months. Discovery campaigns get cheaper because they stop re serving proven terms. Exact campaigns carry more of the sales at bids you actually chose. This is the mechanism underneath the account in our case studies that runs a 16.87% ACOS while total advertising cost of sale sits at 8.68%, and it is worth understanding why those two numbers differ so much if you have not read ACOS vs TACOS yet.
Cutting keeps the money you have. Harvesting is where the growth comes from. The report gives you both, and almost everyone only reads half of it.
Six mistakes we see in almost every account
Judging on a single day. Search term data at day level is noise. Nothing under a fortnight is a trend.
Negating everything with zero orders. Zero orders on four clicks means nothing yet. Use your break even clicks number, every time.
Ignoring automatic campaigns entirely. Automatics are the best discovery tool Amazon gives you and the fastest way to leak budget. They need reading most often, not least.
Forgetting Sponsored Brands. The report covers Sponsored Brands as well, and brand campaigns can waste money on searches that have nothing to do with your range.
Cutting the term instead of fixing the listing. If a highly relevant term gets clicks and no sales, the problem may be your images, price or reviews rather than the keyword. Check whether the listing converts at all before you blame the traffic, and if you are not sure, read your own listing on a phone the way shoppers do.
Never pruning the negative list. A discontinued competitor from two years ago should not still be blocking searches. Review it quarterly.
A cadence you can actually keep
Ambitious routines fail in week three. This one survives because it is short.
Weekly, about twenty minutes. Pull the last 60 days summary. Filter zero orders, sort by spend, work down the top thirty rows. Apply negative exacts where clicks exceed your break even number. Note anything that surprises you.
Fortnightly, about fifteen minutes. Filter for orders, sort by ACOS. Promote the best three terms into exact match. Add the matching negatives back in the source campaign.
Monthly, about half an hour. Look at what changed. Is total spend down and total sales flat, which is the outcome you want? Or are sales sliding, which means you cut something that was buying rank? Check search term counts too: a sharp rise in the number of unique terms usually means a bid increase pulled you into looser matches.
Quarterly. Prune the negative list. Re read your break even clicks number, because your costs and your average cost per click both move.
Twenty minutes a week is roughly seventeen hours a year. On a $10,000 monthly budget with a quarter of it going nowhere, that is around $30,000 of spend redirected towards search terms that actually sell. There are not many jobs in this business with that return per hour, which is why it is the first thing we do on every account we take on.
Send us the report and we will do the first pass with you
Free, no card, no obligation, and you keep the written answer whether or not you hire us.
Frequently asked questions
Weekly for accounts spending over roughly $5,000 a month, fortnightly below that. The report is delayed by around 48 hours, so pull it on a Monday for the fortnight that ended the previous Friday and you get a clean, settled window. Daily checking is a waste of your time: single day data on a keyword with six clicks tells you nothing you can act on.
Use your break even conversion rate rather than a fixed number. Divide your average cost per click into your profit per unit: that tells you how many clicks you can afford before a sale must happen. For most sellers the answer lands between 10 and 20 clicks. Below 10 clicks with no sale you are usually looking at noise, not evidence.
A keyword is what you told Amazon to target. A customer search term is what the shopper actually typed. Broad and phrase match let those two drift apart, which is the entire reason the report exists. Exact match keeps them close, though even exact match now matches close variants such as plurals and small misspellings.
Negative exact blocks one specific search term and nothing else, so it is the safe default when you are cutting a single loser. Negative phrase blocks any search containing that word order, so it is the right tool for a theme you never want to appear for, such as a competitor brand or the word free. Phrase negatives are powerful and blunt, so read them twice before you apply them.
Not directly. Negatives only control where your ads appear. The indirect risk is real though: if you negate a term that was quietly producing sales at a slow rate, you lose the sales velocity that feeds organic rank on that term. That is why we separate spend that buys rank from spend that only buys clicks before cutting anything.
Taking a search term that has proven itself in an automatic or broad campaign and promoting it into its own exact match keyword, where you can bid on it deliberately. It is the profitable half of the search term report, and the half most sellers skip because cutting feels more urgent than building.
No, and that catches people out. Sponsored Display targets audiences and products rather than searches, so you get placement and targeting reports instead. The search term report covers Sponsored Products and Sponsored Brands only.
In the accounts we audit, somewhere between 15% and 35% of spend sits on search terms that have never produced a sale in the reporting window. It is rarely one catastrophic keyword. It is usually two hundred small ones, none of them big enough to notice on their own.
Send the right reports. Get a straight answer back.
We read the account properly and write back with where the money is going, what it is buying, and the three things we would change first. You keep it either way.
No obligation · no card · a person replies, usually the same working day