Amazon ad keywords, chosen from what shoppers already search

How do I find keywords for Amazon PPC?

Start from the searches that already brought clicks and orders on your own ads, then widen to searches your ads have not reached yet. Your own data is the most certain source; every other source adds reach and loses certainty.

What is the difference between a keyword and a search term?

A keyword is the word or phrase you bid on; a search term is what the shopper actually typed. Each keyword has a match type that sets how closely a search must match it, so one phrase or broad keyword can match hundreds of search terms, and the search term report lists which ones brought a click (how each match type works).

How do I use the search term report to find keywords?

Download it for the last 60 days, sort by orders, and look for three kinds of rows: searches with orders that are not yet their own keyword, relevant searches with many clicks and no orders, and searches that do not fit the product. The first become exact keywords, the second get a lower bid, and the third become negative keywords.

The report lists only searches that brought at least one click, and it looks back 65 days (Amazon Ads). Entries such as “b00ipgvvz4” are product pages where an automatic or product targeting ad showed, not searches. Expect most rows to be small: in our accounts, 96% of the non-branded searches that got a click got fewer than 10 in 90 days.

How do I use Amazon Brand Analytics for keyword research?

Use Search Query Performance to see which searches bring shoppers to your brand and where you lose them between impression and purchase, and Top Search Terms to see the searches across Amazon and who wins them. Both need a Professional selling account and the Brand Representative role for a brand enrolled in Brand Registry (Amazon).

• Search Query Performance lists each search with its volume and your brand’s impressions, clicks, cart adds and purchases. Where your share of clicks trails your share of impressions, look at the listing (image, price, rating) before the bid. Where your share of purchases trails your share of clicks, look at the product page or the price.
• Top Search Terms shows the top three products, categories and brands for each search. If those products are like yours, the search is relevant; if they are a different kind of product, it probably is not, however popular it is.

How do I find the keywords my competitors rank for on Amazon?

Search your main keywords on Amazon and note which competitor products rank and advertise on the first page. Then look for those products in Brand Analytics’ Top Search Terms, and use a third-party tool’s reverse ASIN lookup to list the searches a competitor’s product ranks for. Search Query Performance shows only your own brand’s searches, so it cannot do this.

A competitor’s searches are only candidates: each one still has to fit your product and pass the same profit check as any other.

How do I decide which Amazon keywords to target first?

Rank candidates by relevance first, then by purchase intent, then by the profit a click is likely to leave. Volume and competition decide how much a search can bring, not whether it is worth a click.

1. 01
   RelevanceWould a shopper typing this be satisfied by this product? If not, stop here.
2. 02
   Purchase intentDoes the search name a product, size or use (“insulated bottle with straw lid”), or a question or idea (“how much water per day”)?
3. 03
   DemandHow many searches a month, from Brand Analytics or an estimate, and is it seasonal?
4. 04
   CompetitionHow many advertisers show, how strong the first-page products are, and what CPC the search commands.
5. 05
   Profit per clickExpected CVR × contribution margin per order, against the likely CPC. With no history on the search, borrow a CVR from similar searches for the same product, and see how to work out what a click is worth.

A worked example: choosing among six searches

The product, brands, volumes, CVRs and CPCs below are hypothetical. As in the bidding guide’s example, the product leaves $12.00 contribution per order and the target is $9.00 of ad spend per order, so value per click is CVR × $12.00 and the target bid is CVR × $9.00.

Two readings matter. The largest relevant search is not the top priority, because at its going CPC each click loses money; a lower bid that wins fewer, cheaper clicks still earns. And the 400-search term ranks above it: a small, specific search with a margin left per click adds profit, and volume estimates often miss or understate small searches.

Are low-volume keywords worth targeting on Amazon?

Yes, as a group. Across 20 US brands we run, searches with fewer than 10 clicks in 90 days brought 25% of non-branded orders at the median store, at a cost per order 9% above the store’s larger searches. One small search rarely shows an order on its own, so reach them through phrase, broad and automatic targeting rather than one exact keyword each.

• A quarter of orders came from the smallest searches. At the median store they brought 25% of non-branded orders, between 22% and 36% for the middle half of stores, and at least 15% in 18 of 20 stores.
• At close to the usual cost. Their clicks converted a little less often than the larger searches’ (0.82 times as often) and cost less (0.90 times as much), so an order from them cost 9% more at the median store, and no more than 25% more in 17 of 20 stores.
• Most never sold on their own. Only 13.5% of these small searches had an order, against 73.9% of searches with 10 to 29 clicks. A small search with no order says almost nothing; as a group, they sold.
• Too many to add one by one. The median store had about 13,500 non-branded searches with a click in 90 days. Phrase, broad and automatic targeting reach them; a search earns its own exact keyword once it has enough clicks to judge.

Exact, phrase or broad match on Amazon: which should I use?

Use each for its own job: exact match to control the bid on searches you have proven, phrase and broad match to find new searches, and automatic targeting to find searches and product pages you would not have listed (what automatic targeting covers). Give each its own campaign, so the bid on a proven search is not set by the broader ones.

Because the highest bid wins, add your exact keywords as negative exact keywords in your phrase and broad campaigns, so each proven search is bought only at its own bid.

How do I move search terms from auto campaigns into manual campaigns on Amazon?

Move a search once it has enough clicks to judge and its orders put its CVR above break-even: add it as an exact keyword in a manual campaign, then as a negative exact keyword in the automatic campaign so the two do not bid against each other. Amazon lets you add searches from an ad group’s Search terms section to other ad groups as keywords or negatives (Amazon Ads).

Product pages (the ASIN rows) move the same way, into a product targeting campaign. Negative keywords take up to 72 hours to take effect (Amazon Ads), so expect a few days of overlap. Keep the automatic campaign running: it keeps finding the small searches that bring a quarter of orders.

How many clicks before pausing an Amazon keyword?

Decide by clicks, not days, and set the number before the test from your break-even CVR: at a $0.90 CPC and $12.00 contribution per order, about 39 clicks without an order shows a keyword converts below break-even. Even then, lower the bid to what its CVR supports before you pause it.

Zero orders from a handful of clicks is weak evidence. A keyword that truly converts at 8% still goes 15 clicks without an order 29% of the time, and 40 clicks 3.6% of the time. Our accounts show the same at scale: 86% of searches with fewer than 10 clicks had no order, yet as a group they converted at 10.5%.

The stopping point comes from the break-even CVR, which is CPC ÷ contribution margin: $0.90 ÷ $12.00 = 7.5%. A keyword converting at exactly 7.5% goes 39 clicks without an order less than 5% of the time (4.8%). So about 39 clicks, roughly $35 at $0.90 each, decides whether a keyword with no orders is below break-even.

For where to start the bid while the test runs, see what starting bid to use for a new keyword.

When should I add negative keywords on Amazon?

Add one when a search does not fit your product, which needs no data, or when a relevant search still loses money at the lowest bid that wins clicks. A relevant search that converts below target usually deserves a lower bid first, because a lower price can make it profitable (what a negative keyword does).

Negative exact keywords also keep a search you bid on in exact match out of your phrase, broad and automatic campaigns.

Should I bid on competitor brand names on Amazon?

Only with a capped bid and a profit test. A shopper who names another brand asked for that brand, so fewer of them buy yours: bid from what their clicks are worth, not from how big the search is. Run competitor searches in their own campaign, so their results do not lower the bids on your category searches.

Should I run ads on my own brand name on Amazon?

Usually yes, in their own campaigns with their own target, while checking how many of those shoppers would buy through your organic listing anyway. Across 15 US stores we run (Sponsored Products, May 28 to August 25, 2026), the median store’s branded clicks were 29% cheaper and converted 1.9 times as often as its non-branded clicks (branded vs non-branded on Amazon). One ACOS target across both moves budget onto branded searches, which may need it least.

How Laurence finds and chooses keywords

Laurence maps the searches in your markets before it spends, groups them by what the shopper wants, and builds campaigns around those groups. More on how Laurence maps your markets.

• A map before spending: Laurence sizes the searches in each of your markets from Amazon’s Brand Analytics (search frequency rank and Search Query Performance), your account’s own search term history, and a keyword tool’s volumes and organic ranks.
• A campaign per group: searches that share one shopper intent form a group, and each non-branded group gets its own exact match campaign, beside a phrase campaign, an automatic campaign and product targeting. Broad match is off unless a strategist turns it on.
• Branded, competitor and category apart: each runs in its own campaigns. Negative keywords come from one rule set at launch: phrase campaigns exclude the searches the exact campaigns bid on, and a brand fence keeps your branded searches out of the non-branded campaigns.
• Careful with thin data: a keyword with little history borrows from similar searches for the same product and bids conservatively until its own clicks build up.
• No fixed click count to pause: each click without an order lowers the CVR Laurence expects from that keyword, and its bid follows every hour.
• People choose what launches: a Laurence strategist reviews every keyword set before it goes live.

How we measured this. Sponsored Products search term reports (Amazon Ads API) for 20 US stores on the Laurence platform, August 6 to November 3, 2026, at Amazon’s default attribution window. Searches for a store’s own brand or its own products are branded and left out; every other search, competitors’ brands included, is non-branded. Rows whose search term is a product page (an ASIN) are left out, and searches are grouped by the clicks each got over the whole window. Each figure is computed per store first, then summarized as the median across stores; no store is named.