There is a moment in almost every KDP launch that looks a lot like this: it's midnight, the book is finally formatted, the cover is uploaded, and the details page shows seven empty boxes under the heading "Keywords". You type the first things that come to mind ("novel", "mystery", "suspense"), pick two categories that sound about right and hit publish.
That book can be very good and still not sell a single copy outside your circle, for a very mundane reason: nobody who doesn't already know you will ever reach it. Readers on Amazon don't browse shelves. They type something into the search bar or scroll down a category list, and those seven boxes and three categories decide which of those places your book exists in.
This guide is about getting it right without paying for tools: how to find out what people actually type, how to spread it across the boxes and how to pick categories where your book stands a chance. Further down there is a complete worked example, box by box.
Quick summary: Amazon KDP gives you 7 keyword boxes (up to 50 characters each) and 3 categories. Use them for phrases readers really type, taken from Amazon's autocomplete and from the books competing with yours, without repeating words already in your title or subtitle. Choose categories where the top books sell at a pace you can realistically match, not the biggest ones. Keywords get you into search results; sales and clicks keep you near the top, so your cover, title and description matter as much as the boxes.
---
How Amazon decides which books to show (what we know and what we don't)
Amazon doesn't publish how its search engine works, so be wary of anyone who claims to know exactly. But there are two things anyone can verify after publishing a few books.
First: to show up for a search, the words someone types have to appear in one of your book's fields. The ones that definitely count are the title, subtitle, series name, author name and the seven keywords. If someone searches for "balcony vegetable garden" and those words aren't in any of those places, your book usually won't appear, however much it talks about the subject on every page.
Second: once you're in the results, the order is driven mostly by sales and by how people respond. A book that shows up for a search, gets clicked and gets bought, rises. One that shows up and nobody clicks slowly sinks.
That gives you the idea to keep in mind throughout this guide: keywords open the door, but they don't get you a seat at the table. They let you into searches where your book makes sense. What happens next depends on the cover, the title, the price, the reviews and the description. That's why stuffing in words that don't match your book doesn't just fail to help: it puts you in front of people who won't buy, and that eventually counts against you.
---
What KDP lets you enter (and what can get you rejected)
Before you research anything, get the limits clear:
| Field | Limit | Keep in mind |
|---|---|---|
| Title + subtitle | 200 characters combined | The subtitle must describe the book, not be a list of keywords |
| Keywords | 7 boxes, up to 50 characters each | You can change them whenever you like |
| Categories | 3 | Chosen directly on the KDP details page |
| Description | 4,000 characters | Supports bold, italics and simple lists |
And this is what KDP's rules don't allow in keywords, because it can lead to a warning or to Amazon changing your metadata:
- Other authors' names, other books' titles or trademarks ("like Harry Potter", "Stephen King style").
- Names of Amazon programs, such as "Kindle Unlimited" or "KDP Select".
- Subjective or time-sensitive claims: "best", "bestseller", "new", "free".
- Information that is false or unrelated to the book.
One more thing worth knowing before you publish: for print books, the title and subtitle are locked once the book goes live. You can tweak them on the ebook, but on a paperback or hardcover changing them means publishing a new edition. Keywords and categories, on the other hand, can always be adjusted. So think the title through carefully; the boxes you can refine over time.
---
Think like someone who doesn't know you exist
The underlying mistake is almost always the same: the author fills in the boxes by describing the book in their own words. But readers aren't looking for your book. They're looking for a solution, a kind of story or a feeling, and they type it however it comes out.
It helps to split searches into three kinds.
Problem searches. The most valuable ones in non-fiction. Someone typing "how to stop procrastinating" or "anti-inflammatory diet recipes" has a specific need and wants it solved today. If your book solves it, that phrase belongs in your boxes or your subtitle.
Genre and subgenre searches. The most common ones in fiction: "Nordic noir", "office romance", "epic fantasy with dragons". The more specific the subgenre, the fewer people search for it, but the more likely your book is exactly what they wanted.
Tone, trope or reader searches. "Second chance romance", "cozy village mystery", "activity book for 6 year olds". This is where many of the searches that tools miss, and readers actually type, live.
A good keyword set usually mixes all three. And it almost never includes single generic words like "novel" or "book", which compete with millions of titles and say nothing about why anyone should pick yours.
---
How to find what people really search for, without paying a thing
There are paid tools that estimate search volume on Amazon, and some of them are good. But everything you need for your first books can be pulled out of Amazon itself with a bit of patience. This is the process we recommend, in order:
- Use Amazon autocomplete in a private window. Open Amazon in an incognito window so your previous searches don't skew the results, choose the "Kindle Store" or "Books" department and start typing your book's topic slowly. The suggestions that appear are searches people make for real, and often. Write down every one that fits.
- Try the alphabet trick. Type your topic followed by a letter: "balcony garden a", "balcony garden b"… Each letter brings up new suggestions. Ten minutes of this gives you more ideas than an hour of thinking.
- Look at what each search returns. Search every candidate phrase and look at the top ten results. If they're books like yours, the phrase is good. If unrelated things come up, the people using that phrase are looking for something else.
- Check whether those books sell. Open three or four of the top results and find the "Best Sellers Rank" in the product details. That number tells you whether the search moves readers. Authors pass around estimates that are useful for comparison, not as exact figures: on amazon.com, a book ranked better than 10,000 in the Kindle Store sells several copies a day, and one beyond 300,000 barely moves. In smaller stores such as amazon.co.uk, the same sales give you a much better rank.
- Read your competitors' reviews. Especially the three- and four-star ones. That's where readers explain, in their own words, what they were looking for and what they missed: "I expected more beginner recipes", "I wish it had covered watering". Those phrases are gold for your boxes, and sometimes they show you which chapter your own book is missing.
- Put it all in a spreadsheet. One column for the phrase, one for whether the results look like your book and one for the rank of the top books. With twenty or thirty candidates you have plenty to choose from.
If you do this work before you write, rather than on the night you publish, there's a bonus: you'll know which questions your reader is asking and you can turn them into chapters. A book whose table of contents answers what people search for sells better, and not only because of the keywords.
---
✍️ Ready to write the book this article is about?
YourNovel.app writes it with you — a complete, KDP-ready manuscript, with an AI that remembers your entire book. Free to start, no credit card.
A complete example: the 7 boxes of one book
Imagine you've written a practical guide titled Balcony Vegetable Garden, subtitled Grow Your Own Food in a Small Space Even If You've Never Planted Anything. It's sold on amazon.com, as an ebook and a paperback.
This is how a lot of people fill in the boxes the first time:
| Box | First version | Problem |
|---|---|---|
| 1 | garden | A single word, and already in the title |
| 2 | balcony | Already in the title |
| 3 | gardening | Far too generic |
| 4 | gardening book | "Book" adds nothing |
| 5 | vegetables | Already in the title |
| 6 | plants | Generic and loosely related |
| 7 | best container gardening book | Subjective claim, not allowed |
And this is how it looks after going through the process above, using phrases taken from autocomplete and from competitors' reviews. For your own book, check them yourself: autocomplete changes over time and from one marketplace to another.
| Box | Worked version | Why |
|---|---|---|
| 1 | container gardening for beginners | Problem search, no repetition of the title |
| 2 | grow tomatoes lettuce and herbs on a patio | What people want to grow, in their words |
| 3 | when to plant vegetables month by month | A specific question the book answers |
| 4 | watering fertilizer and organic pest control | Comes up again and again in reviews |
| 5 | small space urban gardening for apartments | The same idea in different vocabulary |
| 6 | raised planter boxes for renters | What people living in small flats look for |
| 7 | gift for plant lovers | Reader and occasion search |
Notice three things. No box repeats "balcony", "vegetable" or "garden" on its own, because Amazon already has those words from the title and subtitle. Each box is a phrase of several words, not a single word, so it uses the 50 characters. And every one describes something the book really delivers: if the book didn't cover pests, box 4 would have to go, however good the search.
There's no need for commas or quotation marks. Amazon reads the words; punctuation adds nothing and takes up space.
What about fiction? The process is the same, but the boxes are filled with subgenre, setting, tropes and tone. For a mystery set in a Cornish village with a retired heroine, phrases such as "cozy mystery set in an English village", "retired amateur sleuth" or "clean crime novel without graphic violence" work far better than "mystery", "thriller" or "suspense".
---
The 3 categories: where your book has a chance
Until 2023 there was a widespread trick: email KDP support and ask them to add your book to ten categories. That's over. You now choose three categories directly on the details page, and those are the ones that count.
Choosing them well matters for two reasons. First, many readers buy by scrolling through the best seller lists of a specific category. Second, being number one in a category earns you the orange "#1 Best Seller" badge, which attracts clicks everywhere. And that badge is much easier to get in a small category than in a big one.
How to choose them:
See where the books competing with yours sit. On any book's page, under "Best Sellers Rank", you'll see the categories it's in and its position in each one. Check ten books similar to yours and note the categories that keep coming up.
Measure each category by its number 20 book. Open the category's best seller list and look at the overall rank of the book in position 20. If it's 2,000, the category is brutal: you'd need to sell dozens of copies a day just to appear. If it's 80,000, a few sales a day during launch week are enough to get noticed.
Mix them. A good combination is usually one broad category, where the real audience is, and two narrower ones, where your book can stand out early.
Don't cheat. Putting a gardening book into a small category it doesn't belong in might get you the orange badge for a few days, but it puts you in front of readers who didn't want it, and Amazon may move it. On top of that, reviews from people who expected something else do more damage than the badge is worth.
---
The mistakes we see most often
Some have come up already, but they're worth having in one place:
- Repeating in the boxes what's already in the title or subtitle. It's wasted space.
- One word per box. You waste almost all the space and compete for impossibly generic searches.
- Stuffing the subtitle with keywords. "Urban Garden Guide, Balcony, Patio, Containers, Organic, Beginners, 2026" looks desperate on the cover, can get the book rejected and, in print, can no longer be changed.
- Writing keywords in another language. If the book is in English, its readers search in English. Keywords in Spanish only make sense for a book in Spanish.
- Filling them in once and forgetting about them. The boxes aren't final. It's normal to review them a month after launch, once you have data.
- Expecting the boxes to fix a weak book or a weak cover. They can put you in front of a thousand people. If the cover doesn't catch the eye and the description doesn't convince, none of them buy.
---
After you publish: refine with data, not gut feeling
Keyword changes on a published book take up to 72 hours to apply, so there's no point touching them every day. A good rhythm looks like this.
Three or four days after publishing, check that the book shows up when you search for its exact title and for a couple of your keyword phrases. If it doesn't even appear for the title, give it a little longer: full indexing sometimes takes time.
After a month, if you've tried Amazon Ads, look at the search term report. It's the only source that tells you exactly what the people who ended up buying your book typed. Phrases that appear there and aren't in your boxes are the first candidates to replace the ones that aren't working.
When you publish the second book in the series, put what you learned to use. The series name counts in search too, and a catalogue of related books with consistent metadata reinforces itself. If you see KDP as a business, this is what makes each new book cheaper to launch than the last, as we explain in how many books you need on KDP to make a living.
---
Where YourNovel fits into all this
Let's be clear: YourNovel.app doesn't fill in your keyword boxes, and you should be wary of any tool that promises to do it without looking at what your reader searches for. That research is yours, and it's the part that adds the most value.
Where we do help is with what comes before and after:
- Before you write, the book title generator gives you title ideas by genre, so you have options before you lock in your title and subtitle. And the KDP profit calculator tells you how much each sale earns you by format and price, which also affects whether someone buys once they finally find you.
- While you write, the questions you found in autocomplete and reviews can become the book's outline. In YourNovel you plan chapter by chapter and the AI keeps the whole book in mind, so what you promise in the subtitle ends up delivered in the text.
- Before you publish, the rest of our KDP tools help you check the manuscript and run the numbers.
---
Frequently asked questions
Can I repeat words from my title in the keywords? You can, but it's wasted space. Amazon already takes the title and subtitle into account, so the boxes work harder with new words.
Do I have to fill in all 7 boxes? It isn't mandatory, but there's no reason to leave any empty. Every box is one more chance to appear in a search your book fits.
How long does a keyword change take to show? KDP takes up to 72 hours to apply changes. The effect on sales usually needs several weeks to stand out from the noise, especially if you sell few copies.
Is it worth paying for tools like Publisher Rocket? They can save you time, especially if you publish a lot of books, because they estimate search volume and competition. They aren't essential to start with: autocomplete, rankings and reviews give you almost all the information.
Can I change my categories after publishing? Yes, just like keywords. You can change them whenever you want from the book's details page in KDP.
Should keywords be in the store's language or the book's language? The book's language, because that's the language its readers search in. A book in English is searched for in English, on amazon.de too.
---
Nobody finds your book by chance
In a store with millions of titles, nobody stumbles on your book by luck. They find it because they typed something into a search bar and your book was there, with a cover that caught their eye and a description that convinced them. The seven boxes and three categories are the part of that path that depends only on you, and getting them right takes an afternoon.
If you're still writing, start there: research what your reader is searching for and let it shape the book. And if you want to write it with an AI that remembers the whole book as you go, try YourNovel.app.