KDP Publishing: The Workflow, Start to Live Book

KDP publishing is Amazon's self-serve route to putting a book on the Kindle store and in paperback, free to upload, with Amazon taking its cut on each sale instead of charging you up front. You make an account, you upload a manuscript and a cover, you fill in a pile of metadata, and a day or two later strangers can buy your book. That's the whole shape of it.

The uploading is genuinely easy. The parts that decide whether anyone ever sees the book are the ones buried in forms that look like admin, and I mean that literally, because when I published my first one I spent four months on the manuscript, two weeks on the cover, and something like twenty distracted minutes on the metadata, which turned out to be the only part of that whole sequence that determined whether a stranger ever laid eyes on the thing. That's an uncomfortable thing to write as someone who likes writing. It's still what I saw.

This is the workflow in order, with the places I've watched it go wrong.

Before You Touch The Upload Form

Two things need to exist and one thing needs deciding.

The manuscript needs to be finished and formatted, not almost finished. KDP's previewer will render a mess faithfully. It doesn't fix anything, it just shows you the mess at Kindle dimensions.

The cover needs to work at thumbnail size, because that is the only size a browsing shopper ever sees before deciding. Shrink yours to about a postage stamp. Can you still read the title? Most covers that fail this test fail for the same dull reason, which is that whoever made it, me included, spent the entire design session looking at it full screen on a large monitor where every detail was perfectly legible and the failure was therefore invisible.

The thing to decide is whether you want KDP Select. It's the exclusivity program, and taking it means the ebook can't be sold anywhere else while enrolled. In exchange you get into Kindle Unlimited, where you're paid out of a shared pot per page read rather than per sale. For fiction, especially in a series, Unlimited is usually where the readers already are. For nonfiction with an audience you can reach directly, exclusivity costs more than it returns. You can leave later, so it isn't permanent, but it's the one choice on this list that's a strategy call rather than a form field.

Here's how I'd decide it, based on what the book is rather than what you hope happens.

The pattern underneath the table is simple. Exclusivity buys discovery. If you already have discovery, you're paying for something you own.

Setting Up The Account

The account itself is quick and then stalls on the tax interview, which is the part people don't expect.

You'll give Amazon your bank details for royalty payments and complete a tax questionnaire. If you're outside the US, this is where you either supply a US tax ID or go through the treaty questions to reduce the withholding on your royalties. Get this wrong and Amazon withholds a chunk of every sale you make. It's fixable afterwards, but the withheld money is slow to come back, so it's worth the half hour to get right the first time.

Bank details need to match a bank in a country KDP pays. Check the list before you pick the account.

The Metadata, Where The Sales Actually Live

Here's the form that matters more than it looks like it does.

Title and subtitle. The subtitle is searchable. On nonfiction that's a gift, because it's where the words a buyer would actually type can live without wrecking your title. On fiction, keep it clean, a stuffed fiction subtitle reads as spam to exactly the readers you want.

Series. If it is book one of anything, set the series field. Amazon builds the series page from it, and that page is how a reader who finished book one at midnight finds book two without having to remember your name, search for it, and hope the right thing comes up, which is a chain of small frictions that loses most of the people who would otherwise have bought immediately. I have watched this do more for my sales than any deliberate marketing I have attempted.

Description. This is the sales page. Write it as a hook, not a synopsis. It supports basic HTML, so a bold first line and real paragraphs are worth doing, and a wall of unbroken text is worth avoiding. Nobody buys from a paragraph they didn't read.

Keywords. Seven slots. They're search phrases, not single words, and repeating a word you already used in the title or subtitle wastes the slot because those are already indexed. Think about what a reader types when they don't know your book exists.

Categories. This is the one worth the most and understood the least. Amazon's browse categories are a tree, and the deeper you go the fewer books you're competing with for a bestseller badge. Picking the broadest category available means competing with every book in the store. Picking an accurate narrow one means competing with a few hundred. The badge itself sells books, because it's the only signal a stranger has that anyone else read this.

Pick categories that genuinely describe the book. I want to be clear about why, because the temptation to game this is strong and the downside is delayed enough to feel theoretical. A miscategorized book reaches readers who wanted something else, and those readers return it and leave the kind of review that says the book was not what it claimed, which sits on your listing long after the bestseller badge you chased has gone.

Pricing, And The Cliff

The royalty structure is the thing to design around, not react to.

Amazon pays 70% on ebooks priced within a middle band and 35% outside it, above and below. As of my checking this in September 2026, that band runs from $2.99 to $12.99 in the US store. The practical effect is a cliff. Price below $2.99 and you keep barely a third, so dropping your price to look friendlier can genuinely earn you less per sale than pricing higher would. In the 70% band you're also charged a delivery fee based on file size, which is why an image-heavy book nets less per sale than a plain novel.

Paperback pricing works differently, since printing cost comes off the top. KDP shows you the print cost as you set the price, and the minimum viable price is higher than most first-timers guess. Check it before you promise anyone a cheap paperback.

Work it through once and the cliff stops being abstract. Take a book at $4.99, inside the band. Seventy percent is $3.49, and the delivery fee comes off that, so you land a little under three and a half dollars depending on how big your file is. Now drop the same book to $2.49 to look friendlier. You've fallen under $2.99, so you're on 35%, which is about $0.87, with no delivery fee to worry about because there's so little left to take it from.

You cut the price by half. You cut the royalty by three quarters.

To match the $4.99 book's earnings you'd now need roughly four times the sales. Price cuts in this store are not a volume lever, they're a cliff with a friendly-looking sign on it.

The reverse trap is real too. Pricing above $12.99 also drops you to 35%, so a $14.99 ebook earns about $5.25 while a $9.99 one earns roughly $7 before delivery. You charged more and took home less.

Amazon changes these bands, and the current one took effect in July 2026, so check the KDP help pages yourself before you set a price. A blog post is a bad place to learn what the numbers are today.

Where AI Helps, And Where It Ruins The Book

I use AI in this workflow every week, so this isn't a warning from the sidelines.

It's genuinely good at the metadata layer. Description drafts, keyword brainstorming, category shortlists, back matter, the blurb you rewrite eleven times. That's structured work with a known shape, which is what these models handle well, and it's also the work most writers hate enough to rush.

It is bad at being the book. A manuscript generated end to end reads like one. Reviewers say so, in public, on your listing, in the words a future buyer reads first. Amazon also asks you to disclose AI-generated content at upload, so the real question was never whether you could slip it past the system, it was whether you want your name on a book that people put down at thirty percent and then tell other people they put down.

The honest split I've landed on is that AI belongs everywhere around the manuscript and mostly not inside it.

What Happens After You Hit Publish

Review takes somewhere in the region of a day or two, sometimes longer for paperbacks or a first book on a new account. The listing then goes live and looks completely dead, because it is. No reviews, no ranking, no browse placement.

The first job after publishing isn't marketing, it's checking the listing renders right. Look at the description on mobile, where most of your buyers are. Check the Look Inside. Check the series page if you set one. Then link the paperback and ebook editions if they didn't link themselves, because a split listing divides your reviews across two pages and neither one ever looks credible.

Then it is slow. My honest experience is that Kindle income is a lifetime number rather than a monthly one. Books that sell nothing for months keep selling in small amounts for years, and the single most effective thing I have ever done for a book's sales was not a promotion or a price drop or a burst of posting about it, it was publishing the next one, which quietly pulled readers backwards into the first.

The Five Places I've Watched This Go Wrong

Ordered by how much they cost, not how often they happen.

Four of those five are fixable on a live book in about ten minutes. That's the useful thing to know. The one that isn't cheap to fix is the cover, and it's also the one most people spend the least time on.

Straight Answers To The Questions People Actually Search

Does KDP cost anything to publish? No. Uploading is free for ebook and paperback. Amazon takes its cut per sale, and print cost comes off paperback royalties. Anyone charging you a fee to "get on KDP" is selling you a form you can fill in yourself.

How long does approval take? Usually a day or two. First-time accounts and paperbacks skew longer.

Do I need an ISBN? Not for the ebook. KDP gives paperbacks a free ISBN if you want one, though it lists KDP as publisher. Buy your own if that matters to you, and it mostly matters if you plan to distribute outside Amazon.

Can I publish under a pen name? Yes. The author name on the listing is separate from the account and the bank details.

Can I change things after publishing? Yes, and this is the part people don't realize. Description, keywords, categories, price, even the manuscript and cover can be updated on a live book. A weak listing is not permanent.

So, Is It Worth Doing

If you have a finished book, yes, because the alternative is that nobody reads it. The barrier is real work rather than money or gatekeeping, and the work is mostly the manuscript.

What I'd push back on is the version of this sold as passive income. I've published on KDP, and the money arrives in small amounts over long periods, not in a launch spike. Treat it as building a back catalog that compounds and the maths eventually works. Treat it as a payday and you'll quit at book one, which is the exact point where it hasn't started working yet.

If you want the version of this with my actual files in it, the covers, the metadata I used, and the series structure, that's what I put in the KDP Publishing Playbook.