Amazon Keyword Research: Find, Rank and Place the Right Terms

A keyword list is only useful if it reflects how real shoppers search and if every term ends up in the right place. Many sellers start with a tool, export a long list, and paste the highest-volume words into the title. That produces a listing that reads badly, attracts the wrong clicks and still misses the terms that matter. This article builds the list in a better order: seed terms, customer language, competitor listings and your own data first, then grouping by intent, then placing each term in the title, bullets, backend fields or ads.

Start with seeds, then widen

Write down ten to twenty plain words and phrases before you open any tool: what the product is called, what it does, who uses it, where it is used and which attributes shoppers care about, such as material, size, color or count. Do this from memory first. It keeps you anchored in what the product is rather than in whatever a tool happens to surface.

Then widen the list from several independent sources, because each one catches terms the others miss:

• The Amazon search bar. Type each seed and note the suggestions, which reflect real searches.
• Competitor listings. Search your main terms and read the titles, bullets and category placement of the top organic results. Look for words they share.
• Customer language. Reviews, questions and return reasons show how buyers describe the product, including words you would never have chosen.
• Amazon's own and third-party tools. Brand Analytics (if you have Brand Registry) and the keyword suggestions in the ad console come from Amazon. Third-party volume figures are estimates.

Use your own data first

If you already run ads, your Sponsored Products search term report lists the actual queries that produced clicks and orders. A term with orders has proven that it converts for your product, which no estimate can match. Auto campaigns are useful here as a research tool, since they show queries you did not think of.

If you have Brand Registry, Brand Analytics reports on search terms and on your brand's own search performance can show how your listing compares with the overall market for a query, including impressions, clicks and purchases. Your business reports add traffic and conversion by product.

Know the limits. A search term report only covers queries where you showed an ad and got a click, so it is evidence of what works, not a map of the whole market. Use it to confirm and rank terms from the wider research, and fill the gaps with the other sources.

Capture everything in one sheet with these columns: the term, where you found it, how relevant it is (high, medium or low), a rough demand signal, a rough competition signal and any proof from your own data. Anything you cannot fill in is itself a signal.

Group by shopper intent, then rank

Terms that look alike can come from shoppers in very different states of mind, so sort the list into groups before you rank it:

• Category terms: the broad name of the product. High traffic, wide intent, usually harder to win.
• Attribute terms: size, material, color, quantity or compatibility. Narrower and closer to a decision.
• Use and problem terms: what the buyer is trying to do, such as a task, a room or a problem to solve.
• Audience terms: who it is for, such as a gift recipient, a profession or an age group.
• Brand and competitor terms: useful in ads, never in your listing text.
• Ambiguous or wrong terms: words that also describe a different product. Keep these for your negative keyword list.

Rank in this order. Relevance is a filter, not a score: if a term would disappoint a buyer who landed on your product, drop it. Among relevant terms, weigh demand and winnability, which means how hard it is to get visible, using ad suggestions and the strength of the listings you would have to beat. Proof from your own orders counts for more than any estimate.

A fast way to test a doubtful term is to search it yourself and look at the results page. If the top results are a different kind of product, mostly accessories, or a mix of items that have little in common with yours, shoppers using that term want something else, however much volume it shows. Searching the term is a free check on intent that no keyword export can give you.

Aim for a short top tier, three to five core terms, then a supporting tier and a long tail of specific phrases. This gives every term a job instead of treating all of them as equal.

Where each keyword belongs

Notice that the highest-demand term does not always belong in the title. If a term is only loosely relevant, put it in a lower-stakes place such as ad discovery, and let the title speak to the buyer you want.

Titles are also shortened on smaller screens, so the words that matter most should sit near the start, and the whole title should still read as something a person would write. A useful test is to read it aloud. If it sounds like a list of search phrases, it will read like one to a shopper too.

Why keyword stuffing hurts

Stuffing means cramming extra terms into a field because it feels free. It is not. A title that reads like a list of search phrases lowers click-through, and shoppers who do click are less likely to trust the listing. A weaker click rate and conversion rate count against you, because Amazon's results lean on how shoppers respond as well as on relevance.

There are direct costs too. Titles and fields that break category rules can lead to a suppressed listing. Competitor brand names and misleading terms break policy. Repeating a word you already used elsewhere adds nothing, and Amazon's own guidance on search terms advises against it. And irrelevant terms in ads simply pay for clicks that do not convert.

Finally, treat the list as a living document. Search behavior moves with seasons, trends and the products that enter your category. Add a recurring review of your search term data, and refresh the list when a month of new queries comes in or when you make a significant change to the product or the listing. Terms that once looked out of reach may become winnable as your listing gathers proof.

Where to start

• Write ten to twenty seed words from memory, then expand them from the search bar, competitor listings and customer language.
• Pull your own search term data and rank terms by proof first, estimates second.
• Group terms by intent and drop anything that would disappoint a buyer.
• Choose three to five core terms for the title, and give the rest to bullets, attributes, backend fields and ads.
• Check each field against the current category rules, and revisit the list when your data changes.