Amazon PPC keyword research is picking the few keywords your budget can test, each with a price you set before you bid.

Amazon PPC keyword research is not finding the most keywords. It is picking the few your budget can test to a verdict, each with a price you worked out before you bid. Most guides that rank for Amazon PPC keyword research end with a list of hundreds of keywords from a tool. That list is where research starts, not where it ends. This post shows where the good candidates come from, how to throw out the wrong ones in a minute, how to price each keyword before it spends anything, how many keywords your budget can actually test, and how to handle holiday searches before Black Friday on 27 November.

Where Amazon PPC keyword research starts

The best keyword data you can get comes from Amazon itself, and most of it is free. Amazon's own keyword targeting guide tells you to "use the automatic targeting available in Sponsored Products to surface keywords," read the search term report after a couple of weeks, and move the terms that perform into campaigns with manual targeting. That is the core loop. Around it, four more sources fill the list:

• The Amazon search bar. Type your product the way a shopper would and read the suggestions. Then add a letter, a use or a size after it and read them again.
• Competing listings. The titles, bullets and reviews of the products shown next to yours. Reviews are where shoppers describe the product in their own words.
• Search Query Performance. If you are brand-registered, this Brand Analytics dashboard shows the search queries shoppers typed with their total search volume (Seller Sonar; Ad Badger, updated May 2026). Our guide to the Search Query Performance report covers how to read it.
• Keyword tools. Useful for breadth, but read their numbers as estimates. Adbrew (October 2026) notes that Amazon shows search volume only in Brand Analytics, and that the Top Search Terms report gives a search frequency rank, a relative measure. A tool's "12,000 searches a month" is a model, not a count. Use it to rank keywords against each other, not to forecast clicks.

If you sell in more than one marketplace, research each one in its own language. A translated keyword list misses the words shoppers there actually type, which is one reason German campaigns often underperform when they are copied from the UK.

Throw out the wrong keywords first

A long list feels like progress. Most of it is noise. Before you price anything, run one test on every candidate: search it on Amazon and look at the first page. If the products there are like yours, in type, size and price range, keep it. If the page is full of something else, the shoppers typing it want something else, and no bid changes that.

Then sort what is left into three groups, because they convert very differently:

• Product terms describe exactly what you sell: "bamboo cutting board set". These convert best.
• Use terms describe a need your product meets: "cutting board for meat". Real buyers, but they compare more.
• Competitor terms are another brand's name. Those shoppers mostly wanted that brand. Our post on Amazon product targeting covers when going after a competitor pays.

Price every keyword before it spends

Every keyword has a highest cost per click it can afford: price × conversion rate × break-even ACoS. Below that, a click makes money. Above it, it loses. The conversion rate is the hard part before you have data, so start from your product's own rate across its existing ads and adjust it for intent.

Take an example $25 product with a 30% break-even ACoS, where clicks in the category cost about $0.80. A product term converting at 12% can afford $25 × 12% × 30% = $0.90 a click. A use term at 8% can afford $0.60. A gift term at 6% can afford $0.45. A competitor term at 4% can afford $0.30. Only the product term clears the going price.

That does not mean you drop the other three. It means you bid them at what they can afford, $0.60, $0.45 and $0.30, and let the auction decide whether they win clicks at that price. A keyword that only gets traffic above its price is not your keyword. The break-even ACoS guide shows where the 30% comes from, and our post on Amazon PPC conversion rate covers which rate to use.

How many keywords your budget can test

This is the step most keyword research skips. A keyword is only researched once you know whether it sells, and that takes clicks. Ad Badger's keyword research guide (2022) puts it at about 30 clicks. Our conversion rate post shows the arithmetic behind a similar line: at 10% conversion, a keyword that works still gets no order in its first 20 clicks 12% of the time, and in 30 clicks only 4% of the time.

So count your clicks before you count keywords. Take an example budget of $20 a day at $0.80 a click. That buys 25 clicks a day, or 750 clicks in 30 days. At 30 clicks each, that is enough to test 25 keywords in a month.

Now launch the 200-keyword list from the tool on the same budget. 750 clicks spread over 200 keywords is 3.75 clicks each. After a month you know nothing about any of them. In practice the spread is worse: a few broad terms take most of the clicks and the rest get none.

The rule: research as many keywords as your budget can give about 30 clicks each this month, and no more. Rank the rest and keep them waiting. A keyword without enough clicks has not been tested. It has only been paid for.

Holiday keywords: research them now, end them on a date

Amazon's keyword targeting guide says to "include relevant seasonal terms before, during, and after" events like Prime Day and Black Friday. Black Friday is 27 November 2026 and Cyber Monday is 30 November. From now until Christmas, a share of shoppers search for who the product is for, not what it is: "gift for home cook", "stocking stuffers for dad".

Find them the same way: type "gift for" plus your buyer into the Amazon search bar and run the first-page test on each suggestion. Then keep them apart from your tests:

• Their own campaign, with its own budget, so gift traffic doesn't eat the 750 clicks your normal tests need.
• Their own price. Gift searches are broader, which is why the example gift term affords $0.45, not $0.90. If conversion rises in deal week, the price rises with it, then falls again. Our conversion rate post walks through that December trap.
• An end date. Set the campaign to stop after the last delivery date that matters to your buyers. A gift term in January is a keyword nobody is looking for.

Our Q4 2026 guide covers budgets and bids around the same dates.

Research never finishes, it moves to the search term report

Once the first 25 keywords are live, research becomes a weekly job on real data. The search term report shows the queries that actually triggered your ads. The ones that sell become exact-match keywords with their own bid, as in our guide to search term harvesting. The ones that pass 30 clicks with nothing to show become negative keywords. Every keyword that reaches a verdict frees clicks for the next one waiting in your list.

What to do this week

1. Build the candidate list from your auto campaign's search term report, the Amazon search bar, competing listings and, if you are brand-registered, Search Query Performance.
2. Search every candidate on Amazon and drop the ones whose first page shows products unlike yours.
3. Price the rest with price × conversion rate × break-even ACoS, and write the number next to each keyword.
4. Count your clicks: monthly budget ÷ cost per click ÷ 30. Launch that many keywords, best-priced first, and park the rest.
5. Put holiday terms in their own campaign with their own budget, their own bids and an end date, before 27 November.

Mirox does this work on the campaigns it builds for you. Its 16 agents each do one job, and keywords and bids are two of them; every bid change is saved with the numbers behind it, as how it works describes. In each of the 12 marketplaces, keywords are matched in the local language, and a safety breaker stops keywords spending far above normal without results. The 30-day free trial runs live on up to 50 of your parent ASINs in 1 marketplace, so your existing campaigns stay untouched.

Common questions

How do I do keyword research for Amazon PPC?

Collect candidates from an automatic campaign’s search term report, the Amazon search bar’s suggestions and the words on competing listings. Keep the ones whose search results show products like yours, price each one with price × conversion rate × break-even ACoS, and launch only as many as your budget can give about 30 clicks each.

How many keywords should I target in an Amazon PPC campaign?

As many as your budget can test, and no more. Divide the clicks your budget buys in a month by the clicks one keyword needs for a verdict. At $20 a day and $0.80 a click, that is 750 clicks, or about 25 keywords at 30 clicks each.

Can I see search volume for Amazon keywords?

Only brand-registered sellers see Amazon’s own count, in the Search Query Performance dashboard in Brand Analytics. The monthly volumes in third-party keyword tools are their estimates, so use them to rank keywords against each other, not to forecast clicks.

Is Amazon PPC keyword research different from Amazon SEO keyword research?

The candidate list is the same. PPC adds a price: every keyword you bid on has a maximum cost per click it can afford and a number of clicks it needs before you know whether it works. SEO keywords cost nothing to keep in a listing.

The one-line version

Find keywords on Amazon, keep the ones whose search results look like your product, price each one, and launch only as many as your budget can give 30 clicks.

Want to see the numbers behind a real bid? Read how transparency works, then compare your click prices with the market in Amazon CPC benchmarks 2026.