Back to Resources

Amazon keyword research is the work of finding the search terms real buyers use, confirming which of those terms produce purchases, and placing them where Amazon indexes them. The process itself has not changed much. What changed in 2026 is where the keywords go and how you decide which ones deserve the space.

Two shifts drive everything below.

First, Amazon capped product titles at 75 characters in every category except media as of July 27, 2026, down from a working limit of roughly 200. A title that held twelve or fifteen keywords now holds four or five. Every keyword you cut has to land somewhere else or stop being indexed.

Second, search results are assembled by systems that read intent, not just strings. Amazon’s COSMO knowledge graph connects a query like “keep drinks cold at soccer practice” to an insulated bottle whether or not those words appear in your listing.

The practical response to both is the same. Stop sorting your keyword list by search volume. Sort it by purchase rate, then place the survivors deliberately.

What Amazon Keyword Research Means in 2026

Amazon keyword research is the process of identifying the search terms shoppers use on Amazon, validating them against conversion data, and assigning each one a specific place in your listing. It differs from Google keyword research because Amazon searchers arrive with purchase intent rather than research intent, which makes transactional phrasing more valuable and makes search volume a weaker predictor of results.

The output is not a spreadsheet of terms. The output is a placement map: this phrase goes in the title, these go in Item Highlights, these go in bullets, these go in the backend field.

That distinction matters more under a 75-character cap than it did under 200. When title space was cheap, a sloppy keyword list still worked because you could fit everything. Now the list has to be ranked before it can be placed.

Step 1: Start With Your Own Purchase Data

Most keyword research starts in a third-party tool. It should start in Seller Central.

The Search Query Performance report lives in Brand Analytics and requires Brand Registry. It shows, for every query associated with your ASINs, how many impressions you got, how many clicks, and how many purchases. That last column is the one almost nobody sorts by.

Sort by search query purchase rate rather than impressions. What surfaces is usually uncomfortable: a handful of your highest-volume terms drive the most impressions and very few of the purchases, while narrower terms you never targeted deliberately convert several times better.

We see this pattern often enough across partner accounts that it is close to a rule. A term with 50,000 monthly searches converting at 1% and a term with 5,000 searches converting at 8% produce very different outcomes from the same ad spend, and only one of them looks impressive in a keyword tool.

Three things to pull out of SQP before you go anywhere else:

• Terms where your purchase share beats your impression share. You convert well here and Amazon is under-showing you. These are your priority terms.
• Terms where your impression share beats your purchase share. You are visible and losing the sale. Fix the listing before you spend more on the keyword.
• Terms driving purchases that appear nowhere in your listing copy. Amazon found you through association. Make the connection explicit.

If you do not have Brand Registry, your Sponsored Products search term report is the fallback. It is narrower, because it only covers queries you paid to appear on, but it still reports what people searched before they bought.

Step 2: Build a Seed List From Three Sources

Seed keywords are the broad terms that open the door to more specific ones. Ten to twenty per product is enough.

Pull them from three places rather than one:

• Your own data. The priority terms from Step 1, plus anything in your search term report converting above account average.
• Your product. Write down the jobs the product does in plain language. A joint supplement does “reduce joint pain” and “stay mobile while aging.” Those phrasings become query patterns.
• Your category. Reverse-ASIN lookups on the two or three competitors who consistently outrank you.

Cover three categories with your seeds: what the product is, what it does for the buyer, and which attributes describe it (material, size, use case, audience).

Step 3: Mine Autocomplete and Competitor Listings

Amazon’s autocomplete reflects real search behavior, which makes it a reasonable free signal for relative demand. Type a seed term and read the suggestions on both desktop and mobile, because mobile results lean harder on curated modules and badges that change which terms produce visibility.

These prompt patterns surface intent-rich phrases quickly:

• [seed term] for
• best [seed term]
• [seed term] vs
• is [seed term] good for

On competitor listings, read the reviews and the customer questions before you read the title. Titles tell you what a competitor decided to optimize. Reviews tell you the words buyers use on their own, which is the language COSMO is built to connect to products. Our guide to Amazon competitor analysis covers the reverse-engineering side in more depth.

Step 4: Pick One Research Tool and Learn It Properly

Tool choice matters less than most sellers assume. Every option below returns roughly the same terms. The difference is workflow.

Helium 10 is the default for most sellers. Cerebro handles reverse-ASIN research, Magnet expands from seed terms, and Frankenstein cleans and de-duplicates lists.

DataDive is stronger if you think in clusters rather than individual keywords, with rank tracking and thematic keyword grouping.

Jungle Scout bundles keyword research into a wider seller toolkit, which suits people who want fewer subscriptions.

ZonGuru pairs keyword research with AI-assisted listing generation if you want research and copy in one workflow.

Semrush’s Amazon Keyword Magic Tool earns its place only if you run DTC alongside Amazon and want to see where off-Amazon search behavior overlaps.

Google Keyword Planner is worth a look for ideation, but Amazon volume and Google volume correlate poorly. Treat it as a source of hypotheses, not answers.

An AI assistant can produce a first-pass keyword map from seed terms and competitor ASINs in ten or fifteen minutes. That output is a starting point, not a keyword list. Validate everything against Brand Analytics before anything reaches a live listing.

Step 5: Sort by Intent Before You Sort by Volume

Every keyword carries an intent, and intent determines placement more than volume does.

Transactional queries name the thing and the qualifier: “insulated water bottle 32oz.” These earn title and Item Highlights space.

Comparative queries weigh options: “yoga mat vs exercise mat.” These belong in bullets, A+ Content, and Sponsored Brands headlines.

Informational queries describe a problem: “what helps with wrist pain typing.” These belong in Q&A, A+ Content, and your Brand Store, where Amazon’s systems pull answers from.

Then apply a relevance test to every term that survived. Does it describe the product accurately? Is there evidence in your images, bullets, and reviews that you deliver on it? Do your click and conversion numbers support keeping it?

A no on any of the three means the term costs you more than it returns. High impressions with weak conversion actively suppress you on that query over time, which is the mechanism most sellers miss when they stuff a listing with adjacent terms.

Step 6: Give Every Keyword a Home

This is the step the 75-character cap rewrote.

Title (75 characters, all categories except media). Brand, core product noun, and your single highest-purchase-rate qualifier. That is realistically four to six words of keyword content. Front-load the primary term. Titles left over the limit after July 27, 2026 get rewritten by Amazon’s AI, and brand-registered sellers get a 14-day review window on those rewrites. Our breakdown of the 75-character title limit walks through the rewrite mechanics.

Item Highlights (125 characters). New, searchable, and displayed near the title. This field absorbed most of what used to live in long titles: materials, use cases, comparison attributes. Treat it as title overflow that is indexed, not as a nice-to-have.

Bullets. Feature and benefit language, written for a human first. Comparative intent terms belong here.

A+ Content and Brand Store. Problem-state and question phrasing. These surfaces feed the generative answers Amazon assembles, including inside Alexa for Shopping.

Backend search terms. Synonyms, misspellings, regional variants, and anything that would read badly in visible copy. Do not repeat words already in your title or bullets, because Amazon indexes those once and the duplication wastes space you now need more than you used to. Placement only pays off if the field is set up correctly, so it helps to read how the backend search terms field works before you fill it.

Here is the part worth sitting with. The conventional read on the title change was that titles got more important because every character counts. The practical effect ran the other way. Squeezing 200 characters of indexed text into 75 pushed the surplus into Item Highlights and the backend field, which means two fields most sellers treated as afterthoughts now carry keyword weight that used to sit in the title. Our guide to backend search terms covers how to build that field and how to measure it correctly.

Step 7: Review on a Cadence That Matches the Data

Different data moves at different speeds, so a single review cadence wastes effort on one end and misses shifts on the other.

Weekly: search term report. Add converting terms to campaigns, negative-match the waste.

Monthly: Search Query Performance. Watch purchase rate and click share for movement, especially on terms where a competitor is gaining.

Quarterly: full placement review. Re-rank the list, re-check the title against the character limit, refresh Item Highlights and backend terms for seasonality.

Track themes, not individual keywords. A master list organized by cluster (“gift” terms, “beginner” terms, “travel” terms) tells you which arguments are working. A list of four hundred individual rankings tells you almost nothing you can act on.

One check worth running after every backend update: search your ASIN followed by a keyword in the Amazon search bar. If your product appears, the term is indexed. If it does not, something in that field is not registering, and finding out three months later costs real revenue.

Where This Fits in a Wider Ranking Strategy

Keyword research sets the inputs. Whether those keywords produce rankings depends on conversion rate, review velocity, and the relevance signals Amazon reads across your whole listing, which we cover in our breakdown of Amazon’s ranking factors.

Canopy manages keyword and listing work across partner accounts on Amazon, Walmart, and TikTok Shop, and the title change forced a re-placement pass on every catalog we touch. That work is the difference between a listing that survives an AI rewrite intact and one that comes back missing its best-converting phrase. Our partners average an 84% year-over-year profit increase, and listing precision is a meaningful part of that.

If you want a second read on how your keywords are placed, our Amazon SEO team does exactly this.

Frequently Asked Questions