Amazon Keyword Research: How to Find the Terms That Sell

Key takeaways: The best Amazon keywords are the ones that already sell, not the ones with the biggest search volume. Start with your own data in Search Query Performance and the ad search term report, add competitor and Brand Analytics terms, then tier the list by intent and attainability. Put primary terms in the title, secondary in bullets, the rest in backend and attributes, and verify indexing 48 hours after upload.

What keyword research on Amazon actually is

Amazon keyword research is the work of finding the searches shoppers type when they want a product like yours, deciding which ones you can realistically rank and convert on, and placing them in the listing fields and ad campaigns where they will do the most good. It is not a one-time list. It is a loop: research, place, verify, read the data, revise.

It matters because Amazon is a search engine with a checkout attached. Every sale starts with a search, a browse or an ad, and the first two depend entirely on being indexed and ranked for the right terms. A listing that is not indexed for a search cannot appear for it at any bid or any rank. A listing that is indexed for the wrong searches gets traffic that does not convert, which lowers its conversion rate and, over time, its rank on the terms that matter.

How the research loop works

Source 1: Your own data

The most valuable keyword data you have is your own, and it is free.

Search Query Performance in Brand Analytics shows, for each of your ASINs or for your brand, the search queries shoppers used, and at each stage how you performed against the whole market: impressions, clicks, cart adds and purchases, with your share of each. A query where you have a high share of impressions but a low share of clicks has a thumbnail or price problem. A query where you have clicks but a low share of purchases has a conversion problem. A query where your share is strong at every step is a term you own and must defend.

Search Catalog Performance, alongside it, gives the funnel per ASIN without the query split, which is useful for spotting which products deserve the research effort.

The advertising search term report shows the exact searches that triggered your ads, with clicks, spend, orders and sales. Sort by orders. Every term with repeated orders at an acceptable ACoS is a proven converter and should be in your indexed copy if it is not already. Automatic campaigns are especially useful here: Amazon matches your product to searches you would not have thought of, and the report tells you which ones sold. The How to Read the Amazon Search Term Report (and Act on It) article covers reading it.

Between these two reports you will usually find that the terms actually driving your sales are a smaller and more specific set than the generic high-volume terms you assumed mattered.

Source 2: Competitors

Open the top ten listings for your main search term and read their titles and bullets. Note the terms they lead with, the attributes they name, and the phrasing they share. Shared phrasing across competitors is a strong hint that the phrasing is what shoppers type.

Third-party tools offer reverse-ASIN lookups that estimate which terms a competitor ranks for. Treat them as leads, not facts: the volumes are estimates and the rankings change daily. Use them to build the candidate list, then validate against your own data and against a live search.

Brand Analytics Top Search Terms is Amazon's own view. It ranks searches by search frequency and shows the three most-clicked ASINs for each with their click and conversion shares. Search for your category's core words and note the terms where the top three ASINs are direct competitors. Those are the terms your product competes for. The search frequency rank is a relative position, not a volume number, but it is real Amazon data rather than an estimate.

Source 3: Customer language

Reviews, questions and answers on your own listing and your competitors' are where the long tail lives. Shoppers describe the problem they were solving and the way they searched for it. A reviewer who writes "I needed something for my camping trip that would fit in a side pocket" has given you a use case and two attributes.

Amazon's shopping assistant Rufus has changed this too. Shoppers now ask questions in natural language, and the systems behind semantic search relate those questions to product attributes. Terms like "quiet enough for a nursery" or "fits a standard drawer" are not classic keywords but they are the queries your listing needs to answer. The Amazon COSMO Explained: How Semantic Search Changes Ranking article covers how this affects ranking.

Product Opportunity Explorer, in Seller Central, groups searches into customer-need niches and shows the search volume trend, the products that serve each niche and their pricing and review levels. It is at its most useful for finding adjacent niches you are not yet targeting.

Step 4: Tier the list

A raw list of two hundred candidate terms is not a plan. Tier it. For each term, judge three things:

• Intent. Does this search describe a product like yours specifically, or a broad category, or something adjacent? "Cast iron skillet 10 inch" is high intent for a 10-inch cast iron skillet. "Cookware" is not.
• Attainability. Look at the top results for the term. Their review counts, prices and ratings tell you whether your listing can compete on page one within a quarter. A term owned by products with ten times your reviews is a long-term goal, not a launch target.
• Evidence. Has this term already produced orders for you, or clicks with a good conversion share in Search Query Performance? Evidence outranks volume every time.

Then assign:

• Primary terms, two to four of them, that describe the product precisely, have real volume and are attainable. These go in the title and get their own exact-match ranking campaigns.
• Secondary terms, a dozen or two, that are more specific, describe attributes or use cases, and have proven or likely conversion. These go in the bullets and in control campaigns.
• Long-tail terms, the rest: synonyms, alternate spellings, regional words, use cases, compatibility. These go in backend search terms and structured attributes, and are left to discovery campaigns to prove.

Ranking on a short list of attainable, high-intent terms and then expanding is how organic rank gets built. Ranking on everything at once is how budgets disappear. The Amazon Keyword Rank Dropped: What to Check First article covers what to do when a primary term slips.

Step 5: Place terms in the listing

Each field has a job and a limit.

Title. The primary terms, written as a person would say them, within Amazon's 200-character limit and your category's style guide. Front-load the product type and the defining attribute. Never repeat a word to rank for it; Amazon indexes a word once wherever it appears, and repetition lowers click-through. The The Amazon Title Formula That Ranks and Converts article covers structure.

Bullets. Secondary terms worked into sentences that sell. A bullet that reads "Non-stick ceramic coating for easy cleaning" carries "non-stick", "ceramic" and "easy cleaning" without reading as a keyword list.

Backend search terms. Up to 250 bytes. Terms not already in the visible copy: synonyms, misspellings shoppers actually type, regional variants, use cases. No brand names, no competitor names, no ASINs, no repetition, spaces not commas. The Amazon Backend Search Terms: Best Practice for the 250-Byte Field article covers the field in full.

Structured attributes. Material, color, size, item form, special features, target audience and the rest of the category's fields. These drive the filters on the search results page and are increasingly read by Amazon's AI systems. Empty attributes are the most common reason a well-written listing disappears from filtered searches.

A+ Content. Not treated as indexed listing copy. Write it for conversion; its facts help AI-assisted search but it is not where keywords go.

Step 6: Place terms in campaigns

The same tiering drives the ad structure. Primary terms get exact-match campaigns with bids set for ranking. Secondary terms sit in exact and phrase match in control campaigns with bids set for profit. Long-tail candidates live in automatic and broad-match discovery campaigns with capped budgets, and graduate to exact match once they have proven they convert. A search term that has earned its own campaign gets its own budget and its own bid, which is how you control cost per click and rank on the terms that matter. The Amazon Match Types: Broad, Phrase and Exact, Used Properly article covers how match types fit this.

Negatives close the loop. A term that spends and does not sell in discovery is negated. A term that graduates to exact is negated in the campaign it came from so the two never compete.

Step 7: Verify indexing and track rank

An upload that succeeds is not a listing that is indexed. About 48 hours after upload, check each primary and secondary term: search for the term together with your ASIN in the Amazon search bar. If your product appears, the term is indexed. If it does not, something is wrong: the field is over its limit and was truncated, the term is in a field that is not indexed, a category or attribute conflict is blocking it, or the listing is suppressed. The How to Check If Your Amazon Listing Is Indexed for a Keyword article walks through each cause.

Then track organic rank weekly for every primary and secondary term, and read Search Query Performance monthly. Rank moves slowly and mostly follows conversion on the term. If a term is indexed but rank does not move after six to eight weeks with ad support, the term may be beyond reach for now; drop it to a lower tier and put the effort into one you can win.

Marketplace differences

Research does not transfer between marketplaces by translation. Shoppers in Germany, France or Saudi Arabia use different words, different units and different attributes, and the search frequency ranks in Brand Analytics differ by marketplace. Repeat the research in each marketplace's own Brand Analytics and search term report. Backend byte limits bite harder in languages with accented or non-Latin characters, since each takes more than one byte.

Common mistakes

• Chasing volume. The biggest generic term in your category is usually owned by products you cannot outrank this quarter. Attainable and high intent beats big.
• Ignoring your own data. Search Query Performance and the search term report are more reliable than any third-party estimate and cost nothing.
• Repeating words. Amazon indexes a word once. Repetition wastes characters and lowers click-through.
• Skipping attributes. The filters on the results page run on them.
• Never checking indexing. Weeks lost to a term that was never indexed because a field was truncated.
• Treating the list as finished. Search language moves. New competitors appear. Rufus changes how people ask. Revisit quarterly.
• Competitor brand names in your copy. Against policy, and a route to suppression.

How WeSpark does it

Our keyword work starts with your own data, builds a tiered list, places each tier in the right field and campaign, and verifies indexing 48 hours after upload. The processes below are the ones we follow.

B3How we find and index your keywordsa tiered keyword list your listing actually ranks for, verified 48 hours after upload and tracked every week.

B14How a search term earns its own campaigncontrol of cost per click, spend and rank on every keyword that has proven it converts.

C6Ranking on the keywords that matter (keyword ranking)page-1 organic rank on a short list of attainable, high-intent keywords — and a plan to hold it with less ad spend.

Rank on primary and secondary terms is tracked weekly and reported alongside conversion, so you can see which terms are earning organic sales and which are still being carried by ads.