Amazon categories and keywords

Every book on Amazon sits on shelves, and the only two levers over those shelves are the category picks and the keyword phrases. They are also the least understood part of the metadata form, for a structural reason: Amazon has never documented how the two levers interact. What follows is how they demonstrably behave, plus the parts Amazon does state outright.

What a category is

A category is a browsing shelf — one node in Amazon’s vast browse tree, from Books → Literature & Fiction → Genre Fiction → Historical on down. Each shelf has its own bestseller chart, and each chart ranks by recent sales velocity, not by lifetime sales.

Two consequences make categories worth more effort than a click:

The chart is relative. A book selling modestly can top a small, precise shelf and be invisible on a broad one. Historical → Scottish is a different competitive universe from Genre Fiction, with a different bestseller threshold — and a different orange flag on the cover image.

The shelf is where browsers live. Search captures readers who know what they want; categories capture readers who are looking. A book that is only findable by search is findable only by people who already knew to look for it.

You may pick up to three categories per format. Amazon can and does adjust placement, but the picks are respected far more often than forum lore claims — the lore exists because the browsing tree is deeper than the picker’s list, which brings us to the keywords.

What a keyword is not

A keyword slot is not a tag, a hashtag, or a field the reader sees. It is a search-term field: ten phrases, each up to 50 characters, that describe the book to Amazon’s systems in language the title, subtitle and description do not carry.

Because it is a search field, three behaviours follow:

Repetition is waste. The categories already state the genre. “Fantasy novel”, “epic fantasy book” and “fantasy series” spend three slots saying what the shelf already says. The slots exist to add unclaimed terms.

Reader language beats industry language. Readers search in their own vocabulary — the tropes, the setting, the audience. “Enemies to lovers”, “cozy mystery cats”, “slow burn romantasy”, “regency romance widower” — the phrases readers type, not the words the trade uses to shelve.

Every slot is an independent bet. Ten slots is ten descriptions of the book from ten angles: who it is for, what happens in it, where it is set, what mood it has, which adjacent books it satisfies. A phrase already in the title or the description is a slot doubled up — the field works best when it is additive.

The rules Amazon does state

Breakable keywords are suppressible keywords. Amazon’s metadata rules prohibit, in the keyword field and elsewhere:

• author names and competitor titles;
• trademarks you do not own;
• misleading or subjective claims — “bestseller”, “free”, “sale”, “new”;
• time-sensitive language of any kind;
• repetition of the title, subtitle or categories.

The enforcement is automated and capricious in both directions: most violations are simply ignored until they are not, and a book’s metadata can be suppressed without notice. The safe posture is plain: describe the book, in reader language, with nothing in the slots you would not say to a librarian.

How the two levers connect

The undocumented part, and the source of the category strategy you will find in every self- publishing community: a slice of the browse tree is reachable only through keywords.

Some shelves are not offered in the category picker at all. They are entered when the book’s metadata — most reliably the keyword phrases — carries the terms associated with that shelf. This is the mechanism behind the well-worn practice of spending one or two keyword slots not on search terms but on category signals: phrases that name the precise shelf the book belongs on.

The practical recipe:

1. Pick the three strongest shelves the picker offers, favouring precision over breadth — the shelf where the book can chart beats the shelf where it belongs in the abstract.
2. Spend one or two keyword slots on shelf signals for a more precise category that the picker does not list.
3. Spend the rest on reader language — tropes, settings, audiences, moods — the additive search terms.
4. Revisit after launch. Categories can be changed, and a book whose sales pattern has outgrown its shelves should be moved to shelves where the pattern ranks higher. Metadata is a living surface; updating a published book covers the mechanics.

What this is worth

A sober paragraph, because this topic attracts the most inflated claims in self-publishing.

Categories and keywords do not sell a book that has not earned it: they place the book where its real performance can compound — the chart flags, the also-boughts, the browsing traffic. A strong book in weak metadata sells a fraction of what it should; weak metadata cannot make a weak book sell. The levers are real, and they are multiplier levers, not engine levers.

They are also, uniquely, free to fix. Cover, prose and price all cost something to revisit; the keyword slots are a form field.

---

The category and keyword forms reward one habit: describing the book, everywhere, in the language of its readers, and never the same way twice. It is the cheapest optimisation in publishing and the most reversible.

Nueel writes the metadata files the platforms derive their categories from, and keeps the book’s own description, series and keywords consistent with what the files actually contain — so the shelves the book asks for are the shelves its content supports.

Common questions

How many categories and keywords does KDP give me?

Up to three categories, chosen from Amazon's browse tree, and ten keyword phrases of up to 50 characters each. The category and keyword forms appear for each format — an ebook and its paperback are categorised separately, in trees that are not identical.

What are the rules for Amazon keywords?

Each slot is a phrase of up to 50 characters. They must describe the book's content: no author or competitor names, no trademarks, no misleading claims, no time-limited terms such as 'free' or 'new', and no repetition of the title or the categories. Amazon can suppress metadata that breaks the rules, and it does.

Do keywords put my book in categories?

Some categories are only reachable through them. A slice of Amazon's browse tree is not offered in the category picker and is entered instead when the book's metadata — title, description, or keyword phrases — contains the terms associated with the shelf. This is why keyword slots are spent on reader language rather than on repeated genre words.

Should I repeat the genre in every keyword slot?

No. The categories already carry the genre; the keyword slots should say what the categories cannot — the tropes, the settings, the character types, the audience — in the words readers actually search. Ten copies of 'fantasy' is one category's worth of information spent ten times.