Amazon Keyword Strategy: How to explore new keyword opportunities

Most Amazon ad accounts are bidding on the keywords the seller thinks describe the product, and leaving money on the terms shoppers actually type. The two lists overlap less than you would expect. Finding the gap between them is what a keyword strategy is for, and the shoppers who search in ways you did not predict are the cheapest customers you will ever acquire, because fewer competitors are bidding on those terms.

The risk runs the other way too. A wide net with no harvesting discipline pays for clicks on terms that never convert, and leaves them running for months. This guide covers how shoppers search at each stage of the funnel, how to cast wide without wasting spend, and how to turn the search term report into a routine that keeps finding new terms after the launch.

Quick answer: cast wide, then harvest

• The opportunity: shoppers describe the same product in many ways, and the terms you did not think of usually cost less per click because fewer sellers are bidding on them.
• Where to look: automatic campaigns, the search term report, and the Search Query Performance report are the three sources of terms you did not predict.
• How to start: run broad and phrase match alongside exact so shoppers can find you at every stage of the funnel, then promote the terms that convert into exact-match campaigns with their own bids.
• How to keep it clean: negate the terms that spend without converting on the same weekly pass that harvests the winners.
• The risk: a wide net with no review schedule, which turns discovery spend into waste that nobody is watching.

Shoppers search differently at each stage of the funnel

At the top of the funnel a shopper is searching for a solution to a problem. Lower down they are comparing options, and at the bottom they know what they want and are looking for the specific item. The same person moves through all three, and the words change at each step.

A worked example: buying a rug for a cold home office

Say a shopper notices the concrete floor in her home office is cold as the weather turns. Her first search is one word, "carpet," because she has a problem and not yet a product. A few searches later she is typing "home office rug" or "area rug," because she has decided what kind of thing solves it. By the time she buys she is typing "beige medallion rug," because she knows the color, the pattern, and the size she wants.

Each of those queries is a different stage of intent, and each converts at a different rate. A seller who bids only on the exact terms that describe the product is present for the last search and absent for the first two. The chart below compares how conversions spread across the funnel for two different product types, and the shape is different in each case, which is the point: you do not know in advance where your buyers convert, so you have to be visible across the funnel and let the data show you.

Where new keyword opportunities come from

Three sources show you the terms shoppers use that you did not write down. Each one works on a different part of the funnel.

Automatic campaigns find the terms you never thought of

An automatic campaign lets Amazon match your listing to queries based on the product content, which surfaces terms outside your own vocabulary: the synonym you do not use, the regional word, the adjacent use case. Our guide to auto campaigns for keyword discovery covers how to set them up as a research tool rather than a budget sink.

The search term report shows what converted

The search term report lists the queries that triggered your ads and what each one did. It is the primary input for keyword harvesting: terms with orders move into manual campaigns, terms with spend and no orders become negative keywords. Read it weekly, and judge each term on enough clicks to be meaningful rather than on the first two.

The Search Query Performance report shows demand you are missing

For brand-registered sellers, the Search Query Performance report shows query volume and your share of impressions, clicks, and purchases for each query, including queries where you are barely present. That is the closest thing Amazon offers to search volume data, and it points at terms with demand where a competitor is taking the sale.

If you want a starting list before any of that data exists, the free Amazon keyword research tool gives you a first pass to seed the campaigns.

How to build the keyword list and keep it earning

1. Seed with every match type. Start with exact match on the terms you are sure of, plus phrase and broad match on the same terms so the campaigns can reach the problem-stage and comparison-stage searches. Broad match now reaches synonyms and related terms, so it does more discovery work than it used to. Our match type guide covers bid posture for each.
2. Run an automatic campaign beside the manual ones. Give it a small, fixed budget. Its job is to find terms, not to carry the account.
3. Harvest weekly. Pull the search term report, promote converting terms into exact-match campaigns with their own bids, and negate the terms that spent past your threshold with no orders. Our post on keyword harvesting strategies covers the thresholds.
4. Negate the harvested term in the campaign that found it. Once a term has its own exact-match home, add it as a negative in the auto and broad campaigns so you are not bidding against yourself.
5. Check Search Query Performance monthly. Look for queries with volume where your purchase share is low, and test them as new targets.
6. Judge on results, not on your own assumptions. The terms you would never have guessed are often the ones with the lowest cost per order. Never assume which keywords will sell; test wide and let buyers tell you how they find you.

Step 3 is where this breaks by hand. One account with a handful of SKUs can run the weekly pass in a spreadsheet. A catalog with hundreds of SKUs and dozens of campaigns produces thousands of search terms a week, and the review either gets skipped or gets rushed, which is how discovery spend turns into waste and converting terms sit in auto campaigns at a bid that was never tuned. That is also where the ACoS damage happens, since the cost of a late negation compounds every day it stays live. Our ACoS guide covers how to read the effect.

Why copying competitor keywords is not a strategy

Competitor titles and listings are a fine source of candidate terms, and a poor source of a plan. The terms that convert for a rival depend on their price, their reviews, and their listing, none of which you share. Use competitor research to widen the seed list, then let your own search term report decide what stays. Optimizing for the volume of impressions is how accounts end up paying for traffic that was never going to buy from them.

Where Trellis fits: Qinetix runs the harvest, Qore runs the review

Qinetix, our ads automation platform, runs keyword harvesting and negation as standing rules rather than a weekly chore. Converting search terms are promoted into manual campaigns at thresholds you set, terms that spend without converting are negated at thresholds you set, and the harvested term is negated in the campaign that found it. Six selectable algorithms then set the bids on the harvested terms per campaign, with an ACoS range as the guardrail and a change log for every move. Qinetix is priced as a flat fee set at onboarding based on scale, not a percentage of ad spend.

The part a rule engine does not do is the judgment pass: which queries in the Search Query Performance report are worth a test, and which harvested terms deserve a bigger bid because the margin supports it. Qore, the Trellis workflow layer, lets you describe that review once, lock it as a skill, and run it on a schedule. A keyword-opportunity skill reads the search term and Search Query Performance reports each week, ranks queries by volume and your purchase share, and stages a list of new targets and negatives for your approval. Actions default to manual approval, so nothing changes in the account that you did not sign off on. Qore is in open beta.

The limit: neither tool knows your product better than you do. A term the data loves can still be the wrong fit for the brand, and that call stays with you.

Conclusion: the best keywords are the ones you did not think of

Cast wide enough to be found at every stage of the funnel, harvest the terms that convert into campaigns of their own, negate the ones that do not, and check the Search Query Performance report for the demand you are still missing. The seller who does this on a schedule finds terms the competition is not bidding on. If the weekly pass is the part that keeps slipping, book a walkthrough to see how Qinetix and Qore run it on your catalog.