Keyword Research for Amazon Book Ads: How to Choose Keywords That Sell

Where keywords that sell come from, how they differ from KDP’s seven keyword boxes, which match type to use, and how many clicks to wait before you judge one.

Daniel 25 September 2026 6 min read
ACOS · one search term, 30 days
$18 spend on the term $63 sales from it
= 29%
0%break-even 70%
On this page

“KDP keywords” means two different things, and mixing them up costs authors money. There are the seven keyword boxes you fill in when you publish, which help Amazon decide where your book appears unpaid. And there are the keywords you bid on in Amazon Ads, where every click costs money. This guide is mostly about the second kind.

Two kinds of keyword

KDP keyword boxesAmazon Ads keywords
WhereBook details, when you publishYour ad campaigns
How manySeven, each a word or short phraseAs many as you like, in ad groups
What they doHelp Amazon match your book to searches, unpaidDecide which searches your ad bids on
CostFreePaid per click
Other authors’ namesNot allowedCommon practice, with care

They reinforce each other: a book that Amazon already understands well, from its categories, description and keyword boxes, tends to do better in automatic campaigns. But they are separate systems, and a keyword in one does nothing directly for the other.

Where keywords that sell come from

Listed from the best source to the weakest:

  1. Your own search term report. The exact searches real shoppers typed before buying your book. Nothing else comes close. Run an automatic campaign for two to three weeks to get one. How to read it.
  2. Amazon’s search box. Type the start of a phrase readers might use and read the suggestions. They are real searches in real use, though they come without volumes. Try adding one letter at a time: “cozy mystery a…”, “cozy mystery b…”.
  3. Comparable books and authors. The books a reader of yours also buys. Their titles and author names can become keywords, and their product pages can be targeted directly with product targeting, which is often cheaper and better matched than keywords.
  4. Genre language and tropes. Readers search by what they want to feel: “enemies to lovers”, “small town romance”, “locked room mystery”, “stoicism for beginners”.
  5. The reader’s problem. For non-fiction, the search is usually the problem, not the topic: “how to stop overthinking” rather than “cognitive behavioural therapy”.
  6. Gift and occasion searches. For books people buy for someone else, such as puzzle books, cookbooks, journals and children’s books, especially in Q4. Q4 plan.

Match types: broad, phrase and exact

Every Amazon Ads keyword has a match type, which decides how closely a shopper’s search has to match it:

Match typeYour keywordCan show forUse it for
Broadcozy mysterycozy mysteries with cats, mystery books cozyFinding new searches, at a lower bid
Phrasecozy mysterycozy mystery series, best cozy mystery booksA phrase that must appear in order
Exactcozy mysterycozy mystery, cozy mysteriesSearches you have proven, at your highest bid

The usual pattern is broad and phrase to discover, exact to exploit. When a broad or phrase keyword turns up a search that sells, give that search its own exact-match keyword with its own bid.

How many keywords to start with

For a manual campaign on a new book, 20 to 50 carefully chosen keywords is a practical start. That is enough to find winners, and few enough that each gets clicks. Hundreds of keywords on a small budget mostly means hundreds of keywords with three clicks each, and nothing you can conclude from any of them.

Bid each keyword from your royalty (break-even bid = royalty × conversion rate), a little higher on exact match and lower on broad. Setting a first bid.

Judging a keyword: how many clicks is enough?

This is where most keyword budgets are wasted, in both directions. Authors cut keywords that were just unlucky, and keep keywords that were never going to sell. The maths is simple. If your book converts at 5%, the chance that a keyword gets no sales at all by bad luck alone is:

Clicks so farChance of zero sales by luck (5% conversion)What you can conclude
2036%Nothing yet
4013%A warning sign, not a verdict
605%Probably not converting — lower the bid or pause
1000.6%Not converting
Rule of thumb

judge after about 3 ÷ conversion rate clicks

At 5% conversion that is 60 clicks; at 10% it is 30. Before that, a keyword without sales is not yet evidence.

Apply the same logic the other way: a keyword with two sales from ten clicks has not proved it is brilliant. Raise its bid gradually, not all at once.

Filling KDP’s seven keyword boxes

For the unpaid side, use each box for a phrase rather than a single word, and use them for things readers search that are not already in your title, subtitle or categories. KDP’s guidelines rule out other authors’ names, the word “free”, claims about quality such as “best”, temporary words such as “new”, and Amazon programme names such as Kindle Unlimited or KDP Select. Breaking them can get your book’s details rejected.

Common questions

How many keywords can I add on KDP?

Seven keyword boxes per book, and each can hold a short phrase, not just one word. These are separate from the keywords you bid on in Amazon Ads, where you can add far more.

Do KDP keywords affect my Amazon ads?

Indirectly. Automatic campaigns use your book’s details — title, description, categories and keywords — to decide where to show your ad, so clear metadata helps them. Manual campaigns bid only on the keywords you add there.

What is the best free tool for Amazon keyword research?

Your own search term report, because it shows what real buyers of your book typed. Amazon’s search-box suggestions come second. Paid tools add estimated volumes and competition, which help most when you are choosing a niche or a new book.

Can I use other authors’ names as ad keywords?

In Amazon Ads it is common practice to bid on comparable authors and titles, and it often converts well when the books genuinely suit the same reader. In the KDP keyword boxes it is not allowed.

What are Amazon platinum keywords?

A field from Amazon Seller Central’s old Platinum Merchant programme for large sellers. It does not exist in KDP and has nothing to do with book ads, so KDP authors can ignore it.

How long should I wait before pausing a keyword?

Until it has had roughly three divided by your conversion rate in clicks — about 60 clicks at 5%. Before that, a keyword without sales is more likely unlucky than bad.

Key takeaways
  • KDP’s seven keyword boxes (free) and Amazon Ads keywords (paid per click) are separate systems.
  • The best keywords come from your own search term report; Amazon’s search box is the next-best free source.
  • Broad and phrase to discover, exact to exploit — promote proven searches to exact match with their own bid.
  • Start a manual campaign with 20–50 keywords so each gets enough clicks to judge.
  • Judge a keyword after about 3 ÷ conversion rate clicks — about 60 at 5% conversion.
See your own ACOS, every period, side by side. Sign in with Amazon and up to 24 months of history arrive within hours. Start free, no card needed.
Start free

Related reading

All articles →