The Short Answer Before the Arguments

KDP still pays. Amazon prints the book, ships it, handles returns and sends you a royalty, and you carry no inventory. That mechanism has not broken. What changed is the crowd around it: more titles compete for every search, Kindle Unlimited pays from a shared fund, and Amazon now asks you to disclose AI-generated content.

"You cannot get rich uploading 50 lazy books" is a fair statement. "Nobody makes money on KDP" is a claim that no source checked, KDP's help pages or the third-party review, backs. The two camps usually describe different strategies, and the gap between them is mostly niche choice, pricing and ad spend.

The verdict in three points: the platform works, broad niches tend to be crowded so check yours, and profit after ads is the number that decides everything. One limit applies to all of it: no public search-volume figure exists for KDP niches, so the "demand has dried up" claim cannot be settled from data, and nothing below pretends to have measured it.

Rates move. Royalty bands, printing fees and AI-disclosure wording sit on Amazon's KDP help pages and get revised, and the Kindle Unlimited page rate is announced monthly through Amazon's KDP Select Global Fund notice rather than a help page. Every figure below is a worked illustration as of September 2026, so redo it with the numbers on those pages on the day you price a book. Re-read the royalty, printing-cost and content-guideline pages before you commit money.

What the "KDP Is Dead" Claims Get Right

These are the claims commonly repeated in seller forums, Reddit included, paraphrased rather than quoted from any single thread. Start with the parts that hold up, because they shape every decision after this.

Broad niches tend to be crowded

Generic coloring books, generic planners and undifferentiated "puzzle book for adults" titles are crowded, by most sellers' accounts. Tools that make production trivial also make it trivial for everyone else. In many broad niches you will find page-one books with reviews, a brand and a long sales history, and a new upload with a stock cover has a hard time displacing them. That is consistent with what sellers report, but no source checked measures it, so run the page-one check below on your own niche.

Kindle Unlimited is a shared pool, not a fixed rate

Kindle Unlimited pays per page read (KENP), and as of September 2026 the per-page figure has landed around $0.004 to $0.005. Amazon sets it monthly by dividing a fund across all pages read, so your pay per page can shift even when your own book does not change. Check the latest monthly Select Global Fund announcement rather than an old blog post.

A 300-KENP book read once in full pays roughly $1.20 to $1.50 at that range. That is roughly what one full read pays, which is why fiction sellers in the program lean on series and long page counts. KENP is a normalized page count, not the physical page count, so your dashboard figure is the one to use.

AI-generated content needs disclosure

KDP's content guidelines separate two categories. "AI-generated" content, meaning text, images or translations created by an AI tool, must be disclosed when you publish or edit a title, even if you edited the output afterward. "AI-assisted" work, where you wrote the content and used a tool to refine or brainstorm, does not need to be reported.

Disclosure is not a ban. It does signal that Amazon is watching the category, and low-effort machine-written books are the ones most exposed to removals and account trouble. Where puzzle grids from a generator fall is my interpretation, not KDP's stated position: the rules talk about text, images and translations, so ask KDP support in writing if in doubt and keep the reply.

Bad listings and thin books get punished

KDP's guidelines say Amazon prohibits content that misleads customers or does not accurately describe the book, and that it enforces with a mix of automation and human review. A puzzle book whose cover promises 200 puzzles and delivers 60 repeated ones is the case those rules exist for.

What the Claims Get Wrong

The wrong claims fall into three groups: outdated numbers, averages that hide the spread, and a confusion between "hard" and "over".

"The royalty is too low to bother"

On paperbacks the royalty is 60% of the list price minus printing cost on Amazon marketplaces, and 40% of list minus printing through Expanded Distribution. Amazon has revised its royalty bands before, and sellers report the rate for cheap paperbacks differently, so read the current rate off KDP's royalty help page on the day you price. The worked example below uses the two documented rates so you can swap in whatever the page says.

Ebooks are a separate case. As of September 2026 the 70% Kindle eBook royalty applies to list prices from $2.99 to $12.99 on Amazon.com, minus KDP's delivery cost (see the eBook royalty page), and 35% applies outside that band. The upper limit was $9.99 until 7 July 2026, and some marketplaces add conditions such as KDP Select, so confirm on KDP's eBook pricing help page.

"Printing eats the profit"

Printing cost is a formula, not a mystery. KDP's printing-cost table combines a fixed fee with a per-page fee that varies by trim, ink and page count, so read the current table. Regular trim and large trim (over 6.12 x 9 in, which includes 8.5 x 11 puzzle books) use different tables, and color ink costs more than black. The free KDP calculators model regular trim, so for large trim you look up the fees on KDP's printing-cost page yourself.

What the complaint usually describes is a thick book at a low price. That combination is a decision, not a law of the platform.

"Everyone loses money"

No audited breakdown of KDP author earnings turned up in the sources checked, so any confident claim about what share of publishers profit is a guess, in either direction. Forum posts are also a lopsided sample of complaints and screenshots, not a census. Treat a thread as a list of ideas to test against the numbers below.

"It is too late to start"

A market that rewards the first uploader in a broad category is gone. A market that rewards the best book for a narrow buyer is not. The test is whether you can find a specific reader with a specific need and serve them better than what exists, which every product business faces.

A Worked Royalty Example With the Rate as a Variable

The formula: royalty rate x list price, minus printing cost, equals royalty per copy. The example assumes a 120-page, 8.5 x 11 black-ink puzzle book with a printing cost of $3.00. That $3.00 is an illustrative round number, not a quote from KDP's large-trim fee table, so swap in the real fee for your trim, page count and ink.

Two things stand out. Printing is a fixed subtraction, so a higher list price keeps a larger share of each sale. And a $7.99 price on a book with $3.00 of printing leaves very little, which is why the cheapest price in a category is rarely the right move for thick puzzle books.

The 40% column also shows why Expanded Distribution is a bonus channel rather than a plan. If Amazon's current bands treat cheaper paperbacks differently, redo the rows with the rate on KDP's royalty page.

What Is Left After Ads, Returns and Proofs

Royalty is not profit. In a competitive niche, advertising is often the largest cost, and it is the number the "KDP is dead" claims mostly ignore.

Take the $9.99 book at about $2.99 royalty from the table. Amazon ad cost is usually expressed as ACoS, ad spend divided by ad sales, and break-even ACoS is royalty divided by price: $2.994 / $9.99 = 30%. Ads mostly drive Amazon.com sales, so the 60% row applies; a sale at the 40% rate would break even at $0.996 / $9.99, about 10%.

At 20% ACoS, a $500 monthly goal needs roughly 500 ad-driven copies ($500 / $0.996 = 502), not the 167 that the bare royalty suggests ($500 / $2.994). Organic sales carry no ad cost, which is why books with reviews and a steady ranking behave so differently from new uploads. If your rate turns out lower than 60%, cap your ACoS at your own break-even, never at 30% by habit.

Three more costs sit outside the table. Returns reverse a sale, so your count of paid copies runs below your order count. Proof copies cost printing plus shipping, and a first book usually needs one or two. Cover and interior tooling, if you pay for any, comes off the top before the first sale.

Competition and Why Quality Now Matters More

A common claim is that AI has flooded KDP so completely that a human cannot compete. The more modest reading is that cheap production raised the supply of average books, which raises the value of a book that is clearly better. A shopper facing a page of near-identical covers looks for reasons to distrust most of them.

Where competition bites is in the things you cannot change quickly. In many broad niches you will find a few established books with large review counts near the top for the broadest terms, and their advantage can compound through ranking. Check whether that holds in your niche with the page-one check below. Your route around them is a narrower angle: a word search for a specific hobby, a large-print format for a specific reader, or a school-year workbook that matches a specific curriculum.

Low-content and puzzle formats remain the easiest to enter and the easiest to copy. Books with an editorial spine, such as a workbook that teaches a skill in a sensible order, are slower to make and harder to clone. Print-on-demand is wider than books, and the overview of print on demand as a business shows where books sit beside shirts and stickers.

Claim-by-Claim Check

How to Check Demand in Your Own Niche

No public search-volume figure exists for KDP niches, so the claim that demand is gone cannot be settled from data. Any number you see comes from a third-party estimate. A sturdier check is to read what Amazon itself shows you, and to write down the date.

Type a seed such as "large print word search for" into Amazon's search box and note the autocomplete suggestions, which reflect what shoppers type. The exact strings change by day and account, so record what you see. Then open page one and check three things:

• Do several books have recent reviews, meaning buyers are still arriving?
• Do the covers and titles look identical, meaning differentiation is possible?
• What list prices do the top sellers use, and does your margin work at that price?

You can also read each top book's Best Sellers Rank and turn it into a rough monthly sales estimate with the free BSR-to-sales calculator in the KDP calculator set. Treat the result as an estimate, since Amazon does not publish the rank-to-sales conversion.

A niche where several books with modest review counts sit near the top can suggest real demand without a locked-in winner. A niche with one dominant title is a warning, and a niche with no books at all is usually empty for a reason.

If books feel too crowded, compare other print products first. The guide to print-on-demand niches covers niches on shirts, stickers and other products, where the economics differ from a paperback.

Formats Compared: Income Source, Cost and Risk

The table lists what each format depends on. It carries no earnings figures, because none of these can be audited, so get real quotes for your own trim and page count.

Bulk uploads of near-identical books sit outside the table on purpose. They cost little money but a lot of time, tend to rank poorly and carry the most exposure to duplicate-content and misleading-listing rules. Sellers who last tend to add series structure, so one buyer becomes three purchases.

For distribution beyond Amazon, the comparison of KDP and IngramSpark covers the trade-offs. The notes on KDP interior templates cover the layout side, where many first puzzle books lose reviews.

A Five-Step Test Before You Spend Money

Instead of asking whether the whole platform is alive, test your specific idea. It takes an evening and costs nothing.

If you plan to make puzzle interiors, KDP Creator generates puzzle and solution pages as numbered 300-DPI PNG images with a manifest, and you assemble them into your interior file yourself, so it does not produce a finished interior PDF. Sellers who already use Word or Affinity templates or a free generator may not need a paid tool at all. The trial is 3 days with a card required, and by default a trial account has smaller daily limits than a paid one, so a 200-page book needs a paid plan.

FAQ