How to Find Negative Keywords for Amazon PPC (Search Term Report Method)
To find negative keywords for Amazon PPC, download the Sponsored Products search term report, filter it to the customer search terms that spent money without producing orders, and add those terms back as negative phrase or negative exact — at the ad group level when you are steering traffic between your own campaigns, at the campaign level when the term is simply wrong for the product. The rest of this guide is about the judgment calls inside that sentence: how much spend is enough evidence, which match type to use, and which rows in the report are not keyword problems at all.
This is a tactical guide for accounts that are already running. If you have not launched a campaign yet, start with the Amazon PPC walkthrough for beginners and come back once you have two to four weeks of click data. For the wider profit framework these decisions sit inside, see the profit-first PPC playbook.
What a Negative Keyword Actually Blocks
A keyword is what you bid on. A search term is what a shopper actually typed. Amazon matches the second to the first, loosely on broad match and tightly on exact, which is why your ads end up on searches you never chose.
A negative keyword is an instruction that prevents your ad from being triggered by shopping queries containing that word or phrase (Amazon Ads: add negative keywords or negative products, checked 2026-09-01). Three things it does not do:
- It does not remove the keyword you are bidding on. Adding
wirelessas a negative while bidding onwireless chargerwill suppress your own traffic. - It does not touch organic ranking. Negatives govern ad eligibility for a query, nothing else.
- It does not recover past spend. A negative only affects impressions from the moment it is live.
That last point is why the cost of a bad negative is asymmetric: a term you should have blocked keeps costing you a little every week, while a term you blocked too early stops producing revenue immediately and silently.
Step 1: Pull the Search Term Report, Not the Keyword Report
In Campaign Manager, go to Measurement and Reporting → Sponsored ads reports → Create report → Sponsored Products → Search term report. Set a 30- to 60-day window and download it. Amazon’s own walkthrough of the report and its columns is on the Sponsored Products search term report help page.
Two facts about this report decide how you read it:
- It only contains search terms that produced at least one click (Amazon Ads: search term report for Sponsored Products, checked 2026-09-01). Impressions without clicks never appear, so this report cannot tell you about wasted impressions — only wasted clicks. That is the right unit anyway, because clicks are what you pay for.
- A Customer Search Term made of letters and numbers is an ASIN, not a search. Amazon’s documentation for this report states that such a value means the shopper came across your ad on another product’s detail page rather than in shopping results (same page, checked 2026-09-01). Those rows are a product-targeting problem and negative keywords will not touch them — see the negative product targeting section below.
One timing caveat before you start cutting: Amazon attributes a sale back to the click that led to it (Amazon Ads: ad campaign attribution, checked 2026-09-01), so the most recent days in any report are still filling in. Judge a search term on a window that ends a week or two before today, not on the last seven days.
Step 2: Four Filters That Turn the Report Into a Negative List
Sort by spend, descending, and run each row through these four filters in order.
| Filter | What you are looking for | Action |
|---|---|---|
| 1. Spend, no orders | Terms with enough clicks that zero orders is unlikely to be luck | Negative phrase or exact |
| 2. Orders, but far above break-even | Converts, but at an ACoS your margin cannot fund | Lower the bid first; negate only if it stays unprofitable |
| 3. Semantically wrong | Wrong size, wrong compatibility, wrong buyer intent — regardless of click count | Negative immediately |
| 4. Duplicate of a harvested keyword | The term already runs as an exact keyword in another campaign | Negative in the source campaign only |
How many clicks is “enough evidence”?
This is where most negative keyword advice hands you a number with no reasoning behind it. Do the arithmetic instead. If your ad traffic converts at rate p, the chance that a genuinely average term produces zero orders in n clicks is (1 − p)^n. At a 10% conversion rate:
| Clicks with zero orders | Chance a normal term looks this bad |
|---|---|
| 10 | 35% |
| 20 | 12% |
| 30 | 4% |
| 40 | 1.5% |
So at a 10% conversion rate, 10 clicks tells you almost nothing and 30 clicks is fairly convincing. At 5% the table shifts hard — 30 clicks still leaves a 22% chance of a false alarm, so you need closer to 60. Take your conversion rate from your own business reports.
The second cut-off is money rather than counts. Your break-even cost per order is your selling price multiplied by your pre-ad margin — the number behind your break-even ACoS in the ACoS guide. On a $25 product at a 30% margin, that is $7.50. A term that has spent $22 with no orders has burned three orders’ worth of profit, and the click count barely matters at that point.
Whichever threshold you use, it is a working rule of thumb derived from your own margin, not an Amazon-published standard. Amazon does not publish a “negate after N clicks” rule.
Negative Exact vs. Negative Phrase
Amazon’s help documentation describes two negative match types for keyword targeting: negative phrase blocks shopping queries containing the complete phrase or close variations of it, and negative exact blocks queries that are the exact phrase or a close variation (Amazon Ads: understand keyword match types, checked 2026-09-01).
| Negative phrase | Negative exact | |
|---|---|---|
| Blocks | Any query containing that phrase | Only that query |
| Blast radius | Wide — one entry kills a whole family | Narrow — one entry, one query |
| Best for | A disqualifying word: a wrong brand, a wrong material, “free”, “used”, “for kids” | A single long-tail loser inside an otherwise healthy family |
| Main risk | Over-blocking terms you never reviewed | Whack-a-mole; the same waste reappears in a new wording |
The practical rule: use negative phrase for the word that makes the search wrong, and negative exact for the search that happens to be wrong. If you sell a stainless steel dog bowl and you are paying for plastic dog bowl, the disqualifier is plastic — one negative phrase entry retires that entire family. If you are paying for dog bowl for great dane and your bowl is small, the problem is that one query; negative exact keeps dog bowl for puppies alive.
Before you add a phrase negative, scan the report for every term containing that word and check none of them convert. One negative phrase entry can quietly remove a profitable long-tail you had not noticed.
Campaign Level or Ad Group Level?
Negative keywords and negative products can be applied at either the campaign or the ad group level (Amazon Ads: add negative keywords or negative products, checked 2026-09-01). The choice is not cosmetic:
- Ad group level — use when the term is fine for the product but belongs somewhere else. This is the standard move for the harvesting workflow: once a discovery campaign finds a converting term and you promote it to its own exact-match campaign, negate it in the discovery ad group so the two campaigns stop bidding against each other.
- Campaign level — use when the term is wrong for this product no matter which ad group serves it: the wrong compatibility, the wrong size class, a competitor’s brand name you do not want to pay for.
A workable default: irrelevance goes to the campaign, traffic routing goes to the ad group. Applying routing negatives at campaign level is how sellers accidentally block a term across ad groups that were supposed to keep it.
The Half Most Sellers Skip: Negative Product Targeting
Those alphanumeric rows in the search term report are ad placements on other products’ detail pages. No keyword negative reaches them. Sponsored Products supports negative targeting for products and brands as a separate mechanism (Amazon Ads API: negative product and brand targeting, checked 2026-09-01), and it is the only tool for this class of waste.
Read those rows the same way as keyword rows: an ASIN that has taken 30 clicks and returned nothing is a detail page whose shoppers do not want your product — often a much cheaper or much more premium item where your price is the mismatch. Add it as a negative product target. If an entire brand’s catalog behaves that way, negate at the brand level rather than adding ASINs one at a time.
Manual, Bulksheets, or Software
For a handful of negatives, the console is fine. Past roughly a dozen at a time, the file is faster: bulk operations lets you add campaign-level negatives one row each, which is the single most defensible use of a bulksheet.
Automation tools add scheduled search-term harvesting on top of that: they scan the report on a cadence and propose or auto-apply negatives against thresholds you set. That is genuine time saved, but the thresholds are still yours to choose, and the subscription has to earn itself back against ad spend you actually have — the honest breakdown of when it does and does not is in the PPC software comparison. Below a few thousand dollars of monthly ad spend, a recurring 20-minute manual pass usually beats paying for the same logic.
Five Mistakes That Make Negatives Expensive
- Negating on impressions instead of clicks. The search term report only reports terms that were clicked; a term with a poor click-through rate is a relevance signal, not a spend problem.
- Negating on last week’s data. Attribution keeps filling in, so a fresh window makes healthy terms look dead.
- Phrase-negating a word that also appears in a converting term. Always scan the whole word family first.
- Negating instead of bidding down. A term at 60% ACoS against a 30% break-even is often a bid problem — half the bid may put it in the black. A term with zero orders is a relevance problem. Only the second one is a negative.
- Never revisiting the list. Listings change, prices change, and a term that was wrong for last year’s variation may be right for this year’s. Export your negative lists once a quarter and read them.
Generic, high-volume head terms deserve special care here — the trade-offs of bidding on them at all are covered in Amazon generic keywords.
A Weekly Routine You Can Keep
- Weekly (20 minutes): pull the last 30 days, sort by spend, apply filters 1 and 3 to the top 50 rows, add negatives, log what you added and why.
- Monthly: review harvested terms and add the ad-group-level routing negatives that go with them; check the ASIN rows for negative product targets.
- Quarterly: export the full negative list and hunt for over-blocking — any negative phrase entry that predates a listing change is a candidate for removal.
- Always: record the date next to every negative you add. Without a date you cannot tell later whether a drop in impressions was your doing.
Frequently Asked Questions
How many clicks should a term get before I add it as a negative?
Enough that zero orders is statistically unlikely at your conversion rate. At 10% conversion, 30 clicks with no orders has roughly a 4% chance of being bad luck; at 5% conversion you need closer to 60 clicks for the same confidence. A spend-based cut-off works too: once a zero-order term has spent two to three times your break-even cost per order, the click count no longer matters.
Do negative keywords hurt my organic ranking?
Negative keywords govern which shopping queries can trigger your ad. They do not change your listing, your indexed keywords, or your organic position. What can move indirectly is sales velocity: if you negate a term that was quietly producing orders, you lose those sales and the ranking they supported.
Should I add negative keywords to automatic campaigns?
Automatic campaigns are where the widest matching happens, so they generate the most negative candidates. Negative keyword targeting “can be used at either the campaign level or ad group level” (Amazon Ads API: negative keyword targeting, checked 2026-09-01), so an automatic campaign is covered either way. Keep the discovery function intact: negate the terms that are clearly wrong, not every term that has not converted yet.
Can a negative keyword stop my ad from showing on a competitor’s product page?
No. Detail-page placements are product targeting, and they show in the search term report as ASINs rather than words. Use negative product or brand targeting for those.
What is the difference between a keyword and a search term?
The keyword is what you bid on; the search term is what the shopper typed. You add negatives against search terms, and the terms you find are also the raw material for new keywords — the harvesting half of the same workflow, covered in keyword research.
The Bottom Line
Finding negative keywords is not a creative exercise. It is a report, two thresholds, and a match-type decision: pull the search term report on a settled date window, cut the terms whose zero-order record is statistically real or whose spend has passed your break-even cost per order, use negative phrase for a disqualifying word and negative exact for a single bad query, and put routing negatives at ad group level and relevance negatives at campaign level. Twenty minutes a week, logged with dates, does more for ACoS than any bidding tweak.
Data checked 2026-09-01 against Amazon Ads help documentation. The click and spend thresholds in this guide are arithmetic derived from your own conversion rate and margin, not Amazon-published standards.