Finding Amazon keywords that actually rank.

KDP gives you seven backend keyword slots, fifty characters each. Most authors fill them with single words. That's wrong — and it's the easiest discoverability mistake to fix on a book that's already published.

Phrases beat single words, every time

Amazon's search ranks books on relevance to the buyer's query, not on how often a word appears in your metadata. The buyer searches for "small town billionaire romance," not "billionaire." A keyword slot containing the single word "billionaire" matches that query weakly. A slot containing "small town billionaire romance" matches it strongly — and also matches "billionaire small town" and "small town romance" and "billionaire romance."

Each 50-character slot is real estate. Spending it on one word is like running a magazine ad with a single word in 72-point type. Spending it on a four-word phrase covers four to six related search queries simultaneously, because Amazon does fuzzy phrase matching across word order.

The 50-character ceiling

Each of your seven slots can hold up to 50 characters including spaces. That's roughly a four to six word phrase. The constraint is design, not failure: it forces you to pick phrases that are specific enough to be searchable but short enough to be typed.

"small town billionaire romance" is 30 characters. "billionaire returns to small town" is 32 characters. "small town second chance billionaire romance" is 44 characters — still legal. "small town second chance billionaire forced proximity" is 53 characters — over the limit. The system silently truncates anything past 50, so a phrase that's 51 characters loses its last word and ranks for nothing useful.

What Amazon flags (and de-ranks)

Some keyword categories will get your book hidden from search results. From Amazon's own KDP help pages:

• Author names — do not put "for fans of Colleen Hoover" or "Stephen King thriller" in your keywords. Amazon's algorithm catches author-name keywords and either ignores them or penalizes the listing. The exception is your own author name; that's already indexed automatically.
• Other book titles — "like Verity" or "The Silent Patient readers" will get flagged the same way. Amazon treats this as trademark misuse.
• Subjective claims — "best mystery 2026," "award-winning," "#1 bestseller." These are filtered as marketing claims and don't index.
• Time-bounded phrases — "new release," "2026," "just published." They expire and signal stale metadata to Amazon's crawler.
• Free or low-price hooks — "free book," "99 cents," "Kindle Unlimited." These are policy violations and can cost you the listing.

Where to find phrases readers actually type

The single best research tool is Amazon's own search bar. Open the Kindle store, type your seed keyword, and watch the autocomplete dropdown. Those suggestions are real searches, ranked by frequency. They're the queries you want to match.

Type "small town billionaire" into Kindle search. The autocomplete will suggest "small town billionaire romance," "small town billionaire forced proximity," "small town billionaire single dad," and so on. Those four suggestions are four backend slot phrases, ready to copy. You can repeat with "grumpy billionaire," "secret billionaire," "country billionaire" and end up with a full pool of validated phrases in five minutes.

The second-best tool is the "Customers who bought this also bought" rail on competitor books. Click into a recently-published book in your subgenre, scan the also-boughts, and look at their titles. The repeated phrases — "second chance," "enemies to lovers," "small town" — are the reader vocabulary in that subgenre. (Or shortcut it: Authorly's keyword finder reads your premise and returns seven slot-ready phrases.)

The seven-slot strategy

Don't put seven variations of the same phrase. Amazon de-duplicates within your own keywords; "small town billionaire romance" and "billionaire small town romance" cover the same query space. Instead, cover seven distinct angles on your book. For a small-town billionaire romance, that might look like:

1. Setting + subgenre: small town billionaire romance
2. Character archetype: grumpy billionaire small town
3. Trope stack: forced proximity billionaire romance
4. Plot hook: billionaire returns to small town
5. Cross-subgenre: cowboy billionaire romance
6. Audience signal: clean small town billionaire
7. Trope combination: second chance billionaire romance

That's seven slots covering roughly twenty-five distinct buyer queries. The next reader who types "second chance billionaire" or "grumpy billionaire small town" or "clean billionaire romance" lands on your book.

Refresh quarterly

Reader vocabulary drifts. The trope phrases that ranked in 2023 (think "morally grey hero," "touch her and die") have already been replaced by 2026 vocabulary. The fix is a fifteen-minute quarterly review: re-run Amazon autocomplete on your seed keyword, compare to your current backend slots, swap in any new phrases the autocomplete is suggesting that you don't already cover.

It's the highest-ROI fifteen minutes in indie publishing. Most authors set their keywords once at launch and never look at them again. The discoverability gap between a book with current keywords and one with three-year-old keywords is real, and it compounds.

§