Most Amazon keyword lists get sorted by search volume and funded until the budget runs dry. If you are working out how to choose the best keywords for Amazon advertising, run three tests instead.
Does the term describe what you actually sell? Is the shopper close to buying? Can the click pay for itself at your margin?
As an Amazon Ads Partner working across more than 20 categories, we put every new account's keywords through those tests before raising a single bid. Here is the full process, from sourcing to cutting.
TL;DR
- The best keyword is the one that converts at a cost per order your margin can carry, which is rarely the one with the most searches.
- Build the candidate list from Amazon's own data first: your listing, automatic campaigns, the search term report, and Brand Analytics.
- Give every keyword a job (defend, discover, convert, conquest, or season) and judge it against the target that job deserves.
- Before bidding, calculate the highest cost per click (CPC) a product can carry, using its price, its conversion rate, and the advertising cost of sales (ACOS) at which it breaks even.
- Decide in advance how many clicks and how much spend a keyword gets before it is promoted, repaired, or cut.
Want a practitioner to pressure-test your keyword list? Send us the account and the margin you need to protect, and we will show you which terms are paying their way and which are quietly eating profit. Book a Call
Define the Best Keyword by What It Earns
A keyword is worth bidding on when three things line up. Search volume matters, but only after the first two tests pass.
- Relevance. The query describes your product closely enough that a shopper who clicks will recognize what they wanted. A term that pulls the right category but the wrong size, material, or price band will buy clicks that never convert.
- Purchase intent. Specific, descriptive queries usually come from shoppers further along. "Bamboo cutting board" signals browsing; "large bamboo cutting board with juice groove" signals someone comparing finalists.
- Affordability. The click has to cost less than the profit the expected order leaves behind. We cover the math in the scoring section below.
On Amazon, keyword targeting applies to two ad types: Sponsored Products and Sponsored Brands. Sponsored Display works from audiences and product pages, so the selection process here is for those first two.
Build the Candidate List From Amazon's Own Data
Third-party tools can add volume estimates later. The most reliable signals come straight from Amazon, because they reflect how shoppers actually searched and what they bought.
Start With the Words Already on Your Listing
Pull every descriptive term from your product title, bullet points, and backend search terms: product type, material, size, use case, and who it is for. Amazon's own keyword strategy guidance recommends using terms from your product detail page (PDP) because they already carry relevance to the product.
Treat this as a seed list. The words you chose for the listing reflect how your team describes the product, which is a starting point for shopper language rather than a replacement for it.
Run an Automatic Campaign as Paid Research
A Sponsored Products automatic campaign lets Amazon match your ad to queries it considers related. Amazon sorts those matches into targeting groups it calls close match and loose match for shopper searches, plus substitutes and complements for product pages.
Fund it modestly and treat the spend as research. The output you care about is the list of real shopper queries that triggered your ad and turned into orders.
Harvest the Search Term Report
The search term report shows the actual queries behind your ad clicks. Amazon includes only search terms that earned at least one click, so impression-only queries will not appear there.
Sort by orders first, then by conversion rate. Queries with repeat orders at a healthy ACOS (ad spend divided by ad-attributed sales) become candidates for exact match. Automatic campaigns also list 10-character ASIN (Amazon Standard Identification Number) entries; those are product pages where the ad served, and they belong in a product targeting test.
Check Search Query Performance for Brand Share
Brands enrolled in Brand Registry can open the Search Query Performance (SQP) dashboard inside Amazon Brand Analytics. For each query that led shoppers to your products, it reports volume, impressions, clicks, cart adds, and purchases, with your brand's share of each stage.
SQP often surfaces queries that already produce purchases for your brand but sit in none of your campaigns. Those are some of the cheapest keywords to add, since the shopper has already shown they will buy from you on that search.
Scan Top Search Terms for Category Demand
The Top Search Terms dashboard lists popular queries across the Amazon store and, for each one, the three products, categories, and brands that shoppers picked most often after searching. Use it to size category terms and see who currently owns them.
Look hard at the products winning a term before you add it. If the top three sit at half your price or carry a different pack size, that query is unlikely to convert for you, whatever its volume.
Capture Shopper Language From Autocomplete and Reviews
Type your core terms into the Amazon search bar and note how the suggestions extend them. Those suggestions show the modifiers shoppers attach to a product type, such as size, use, and audience.
Read your own reviews and customer questions, then your competitors'. Shoppers describe products in phrases a brand team rarely uses, and those phrases often make strong long-tail keywords.
Add Branded, Competitor, and Seasonal Terms on Purpose
Amazon groups keywords into generic, long-tail, branded, and seasonal types. Add each type deliberately with its own budget, rather than letting it drift into a generic campaign where its results blur together.
Competitor brand names are a separate decision with a separate cap. Before targeting another brand's name, review Amazon's branded keyword guidelines in the Amazon Ads support center.
Assign Every Keyword a Job Before You Bid
A single ACOS target across every keyword hides more than it shows. Branded terms flatter the average while category terms look like failures, even when both are doing what they should. Give each keyword one job and judge it on that job.
| Job | Example query | What it is for | How to judge it |
|---|---|---|---|
| Brand defense | "[your brand] cutting board" | Keep your own listings in the top slots on your name | Low ACOS expected; cap spend once share is secure |
| Category reach | "cutting board" | Visibility and rank on a high-volume term | Organic rank movement and total advertising cost of sales (TACOS) |
| Long-tail conversion | "large bamboo cutting board with juice groove" | The profit engine of most accounts | Must stay under target ACOS |
| Competitor conquest | "[rival brand] cutting board" | Win consideration where your offer is stronger | Tight budget cap; cut quickly if it misses |
| Seasonal | "cutting board gift set" | Capture demand during a peak window | Time-boxed budget with a fixed end date |
TACOS is ad spend divided by total sales, ad and organic together. Category terms often pay off in organic sales rather than ad-attributed orders, so TACOS is the fairer yardstick for them.
Branded defense deserves a specific warning. Spend on your own name tends to hit diminishing returns once you hold the top placements, and the budget past that point usually does more work on terms still winning new customers.
Score Each Keyword Against Your Break-Even Math
Profit-first keyword selection starts with a number most accounts never write down: the most you can pay for one click. Work it out per product.
- Find the margin left before ads. Start from the selling price, then take off landed product cost, the referral fee, fulfillment fees, and any other per-unit charge. What remains is your contribution margin before advertising.
- Convert it to break-even ACOS. Pre-ad margin over price gives you that percentage. At this ACOS, ads consume the entire margin, and each order earns zero.
- Set a target ACOS for each job. Conversion keywords sit below break-even, so they leave profit. Category and launch keywords may run above it, but only with a defined end date and a rank goal.
- Calculate your maximum CPC. Multiply price by conversion rate by target ACOS. That figure is the highest bid a keyword can carry and still hit its target.
- Compare it to what the term costs. If the going CPC on a term is far above your maximum, the keyword cannot pay back at your current conversion rate, however relevant it looks.
Here is the arithmetic on an illustrative product. The figures show the method and are no benchmark for your category.
- Price: $30, with $9 of margin left before ads.
- Break-even ACOS: $9 divided by $30, or 30 percent.
- Conversion rate: 12 percent of clicks become orders.
- Maximum CPC at break-even: $30 x 0.12 x 0.30, or $1.08.
- Maximum CPC at a 20 percent target ACOS: $0.72.
Suppose a broad category term costs $2.40 a click. To break even there, the product would need to convert at roughly 27 percent, more than double its actual rate. The term may still earn a small, time-limited place as a rank play, but it should never be counted as a profit keyword.
Rather have a team build this math for every stock-keeping unit (SKU)? We manage advertising alongside SKU-level profit and loss, so bids are set against what each product actually earns after fees. Explore Our Services
Read Search Query Performance to Find Hidden Winners
Return on ad spend (ROAS, ad sales divided by ad spend) tells you how a keyword performed inside your campaigns. SQP shows how your brand performs on that query across the whole search funnel, which is often where the best keyword decisions come from. Compare your share at each stage and look for the point where it drops.
| What SQP shows on a query | What it usually means | What to do with the keyword |
|---|---|---|
| Purchase share at or above click share, but low impression share | A converting term you are under-exposed on | Add or raise it in exact match; consider a Top of Search placement adjustment |
| Healthy impression share, click share lagging | The main image, price, or review count is losing on the results page | Fix the listing before adding spend |
| Healthy click share, cart-add or purchase share falling away | The product page, price, or offer is losing the sale after the click | Hold bids; work on the product page and the Featured Offer |
| Strong share at every stage on a branded query | You already own this search | Test a lower bid and move the savings to growth terms |
| Low share across the funnel on a large category query | The category wants a different product or price point | Keep it as research only, or drop it |
The first row is where we usually find the fastest wins. A keyword that already converts above your average is cheaper to scale than a new term you still have to prove.
Pick Match Types That Fit Each Keyword's Job
Choosing the keyword is half the decision. The match type controls which shopper queries it can actually reach.
- Broad match gives the widest reach. Amazon can match synonyms and related terms, and the shopper's query may not contain your keyword at all. Use it for discovery and category reach, always with negative keywords alongside.
- Phrase match requires your words in sequence, with close variants, and allows other words before or after, which suits long-tail families such as "cutting board with juice groove."
- Exact match is the tightest option and the home for proven converters. Close variants such as plurals and misspellings can still match, so watch the search term report even here.
- Several match types on one keyword are allowed. When more than one could serve on the same query, Amazon uses the match type with the highest bid, so set bids with that in mind.
- Plan match types before launch. Amazon does not let you change a keyword's match type after the campaign is created; you add the keyword again with the new type instead.
Set Clear Rules for When a Keyword Earns a Verdict
Most wasted keyword spend comes from deciding too early or too late. Amazon suggests letting a campaign run for about two weeks before making optimization decisions, and we pair that with a click floor and a spend cap tied to your own numbers.
| What the keyword shows | Verdict | Next step |
|---|---|---|
| No orders after spend passes one allowable cost per order (price x target ACOS, $6 in the example above) | Cut | Pause it and add it as a negative keyword |
| No orders after about three times the clicks you would expect per order (25 clicks at a 12 percent conversion rate) | Cut | Pause it; check relevance before trying a related term |
| Orders at or under target ACOS across two or more weeks | Promote | Move it to exact match in its own ad group and block it in the source campaign |
| Orders, but ACOS above target | Repair | Lower the bid toward your maximum CPC, then review placement and the product page |
| Relevant term with few impressions | Feed | Raise the bid in steps and confirm the term appears in the listing |
The click floor has simple math behind it. If a term truly converted at your 12 percent average, 25 clicks with no order would happen less than 5 percent of the time, so silence that long is a real signal.
Negative targeting on Amazon comes in two forms: negative phrase, capped at 4 words, and negative exact, capped at 10 words, each with an 80-character limit. There is no negative broad option, so use negative phrase on the root of an irrelevant theme and negative exact for single queries.
Avoid the Keyword Choices That Drain Margin
Ranking keywords by volume alone. High-volume head terms tend to be the most expensive and the least specific. Volume is a tiebreaker between relevant, affordable terms and should never be the first filter.
Reading a blended ACOS as account health. Branded keywords usually post the lowest ACOS in an account, which can make a weak non-branded program look fine. Report branded and non-branded performance separately.
Running broad match with no negatives. Broad match is a discovery tool, and discovery without pruning turns into steady leakage. Review the search term report on a fixed schedule and add negatives every time.
Promoting a winner without isolating it. A keyword moved to exact match keeps competing with the broad campaign it came from unless you block it there. Add the promoted term as a negative exact in the source campaign.
Bidding on terms the listing cannot win. If the product page does not mention the attribute a shopper searched for, the click rarely converts. Fix the listing first, or leave the term out.
Frequently Asked Questions
Quick answers to what operators ask us most often about picking ad keywords.
How Many Keywords Should a New Campaign Start With?
Amazon recommends a minimum of 25 keywords for manual keyword targeting. Group them by job so each ad group shares one theme and one target, rather than piling every term into one list.
Are Third-Party Keyword Tools Worth Using Alongside Amazon's Data?
They help with volume estimates and competitor research, especially before you have sales history. Treat their numbers as estimates and let your search term report and SQP data make the final call, since those reflect real shoppers on your products.
Should I Bid on My Own Brand Name?
Usually yes, at a controlled level, because competitors can place ads on your brand searches. Once your listings hold the top placements and SQP shows strong share, test lower bids and track whether branded orders hold.
Is It Smart to Target Competitor Brand Names?
Conquesting can work where your offer is genuinely stronger on price, reviews, or features. Give it a separate campaign with a hard budget cap, and cut it fast if conversion lags, since shoppers searching a brand name often want that brand.
How Often Should I Revisit My Keyword List?
Review search terms and negatives weekly on active campaigns, and rebuild the full candidate list each quarter or before a major season. SQP and Top Search Terms shift with demand, so last quarter's best keyword can fade.
Do Ad Keywords Need to Appear in My Listing?
They do not have to, but it helps. Amazon points to the words on your product page as a source of relevant keywords, and a listing that names the searched attribute gives the shopper a reason to buy after the click.
Choose Keywords That Pay for Themselves
Picking keywords for Amazon ads is a margin decision before it is a volume decision. Source candidates from Amazon's own reports, give each one a job, and bid only up to the CPC your break-even math allows.
Then hold every term to a verdict rule so winners get scaled and losers get cut on schedule. A list run this way tends to get more efficient over time, since spend keeps drifting toward the queries that already sell.
Not sure which of your keywords are carrying the account? Tell us where your advertising feels stuck, and we will tell you what we would change first. Book a Call