If you publish on Amazon KDP, the keyword boxes are one of the few places where a well-chosen phrase can put your book in front of a reader who was never going to find it through the browse menus. KDP gives you seven boxes for keywords, and the rule that trips up most authors is not how many you can enter but how much each box holds. KDP counts bytes, not characters, and that difference is why a title that looks short on screen can overflow its box.
The seven keyword boxes
On the paperback and ebook setup pages, KDP asks for up to seven keyword entries. Each one sits in its own box. You do not have to fill all seven, and you should not treat them as seven slots for seven synonyms of the same idea. Each box is read as a separate search signal, so the most useful habit is to give each one a different angle: a genre phrase in one, a tone or trope in another, a comparison or reader intent in a third, and so on.
Keep each entry to the phrase itself. Repeating words from your title or subtitle wastes space, because KDP already indexes those fields. Multi-word phrases are fine. A keyword like "cozy mystery with a cat sleuth" is one entry, and it uses one box, which makes it worth more than three single words that each carry less meaning.
How the byte limit works
KDP's help pages have long described each keyword box as limited to 50 bytes. A byte is the unit a computer uses to store text. For plain English letters, digits and standard punctuation, one character is one byte, so a 50-byte box holds roughly 50 ordinary characters. Spaces and commas count too, so every separator you type costs you one byte.
The limit is per box, not across all seven. That means you can use the full allowance in each box, but it also means one long entry cannot borrow room from another. KDP's rules have changed in the past, so check the current wording on the KDP help pages before you publish. The counting method, though, is the part that matters for your planning.
Why characters are not enough
Plenty of authors paste a keyword into a word processor, see that it is 48 characters, and assume it will fit. Most of the time that is correct. The trouble starts when the phrase contains anything outside basic English. A word processor's character count tells you how many symbols are on screen. It does not tell you how many bytes the text takes up in the encoding KDP uses, and the two numbers diverge as soon as you use accented letters.
That is why the same title can be fine in one box and cut off in another. Ordinary letters are cheap. Certain letters are expensive.
Why accents cost double
Most modern systems store text in UTF-8, a encoding that uses one byte for the basic Latin alphabet and more bytes for everything else. Letters with diacritics, such as é, ö, ñ, ü and ç, normally take two bytes each. Some other scripts take three or four. Emoji, for example, usually take four.
Here is what that looks like in practice:
"cafe" is four bytes. "café" is five, because the é takes two.
"Levy" is four bytes. "Lévy" is five.
"Mörder" is seven bytes, where "Morder" would be six.
None of these examples looks dramatic, but the effect adds up quickly in a 50-byte box. A phrase like "Nordic noir with a melancholy détective" can end up several bytes over the limit even though it reads as roughly 40 characters. If your genre, character names or setting use accents, expect to lose a little room in every box.
This is not a reason to strip accents from names or drop them from the book. Your title page and description should keep the spelling you intend. The point is to recognise where the bytes go when you are squeezing keywords into the fields. If a reader searches for "café" with the accent, the accented form may still be what they type, so it can be worth keeping the accented version in one box and the plain version in another if you have room. Decide that on purpose, rather than discovering it when KDP truncates your entry.
Checking your boxes before you publish
The simplest way to avoid surprises is to count each box before you paste it into KDP. Pendrilo's free KDP Keyword Byte Counter does this without sign-in. Paste an entry and it shows the byte count, so you can see at a glance whether an accented name or a stray symbol has pushed a phrase over the line.
If you already use Pendrilo's Publishing Desk, the same byte count appears alongside your other publishing details. The Publishing Desk also keeps a readiness list, a 4,000-character description and keyword research from data you paste or import. It does not look anything up online, so the research is only as good as the data you bring to it. Pendrilo is free during its beta, and you can create a free account if you want to try it.
Whichever tool you use, count the final text you will submit. Do not count a draft in a different document and assume it carries over unchanged. Smart quotes, non-breaking spaces and copied hyphens can all behave differently from what you typed.
Practical ways to fit good keywords
Start by sorting your candidate phrases into groups: genre, subgenre, tropes, setting, audience and comparison titles. Then assign one strong phrase per box rather than trying to cram several into each. A box with one specific, accurate phrase usually beats a box with four vague ones.
Where a phrase is too long, look for the word that adds the least. "Enemies to lovers small town romance" might lose "small" in one box and appear as "small town romance" in another. You keep the meaning and reclaim the bytes.
Be honest with your keywords. KDP expects entries that describe the book, and a keyword that promises a feature the book lacks will not help you once a reader arrives and leaves a poor review. Pick phrases a reader would type and recognise as a fit.
Finally, remember that keywords are one part of discovery. Your categories, title, subtitle and description all shape how the book is found and understood. Treat the seven boxes as a tidy, carefully counted list, and spend your effort on the phrases that genuinely describe what you wrote.
