How to self-publish an ebook in 2026: KDP, Apple Books and Kobo without the mythology

Self-publishing advice splits into two genres: get-rich screenshots and 40-step checklists that make a Tuesday upload feel like a moon landing. This guide is the third genre: what actually happens between a finished manuscript and a published ebook, with the numbers that drive the decisions.

Before anything: the file question

Every major store wants a reflowable EPUB. Amazon KDP ingests EPUB and converts it internally; Apple Books and Kobo use it natively. A PDF is not a substitute. It technically uploads in places, and reads terribly on every phone and e-reader that touches it. If your toolchain produces only PDFs, fix that before opening any store account. (The full format logic is in our PDF vs EPUB guide; InkMagnet ships both files from one source precisely because of this split.)

A print edition needs the opposite: a fixed, print-grade PDF with embedded fonts. Same book, two artifacts, two jobs.

The royalty math that decides your price

Amazon KDP pays two ebook royalty rates: 70% for list prices between $2.99 and $9.99 (minus a small per-megabyte delivery fee), and 35% outside that band. The consequences are blunt: a $12.99 ebook earns about $4.55, while a $9.99 ebook earns roughly $6.90. The band explains why so many ebooks cost exactly $9.99: it is the ceiling of the 70% zone, not a coincidence. Heavy image books pay attention to the delivery fee: it is deducted per download at the 70% rate, so a 50 MB photo-rich file quietly eats margin.

Apple Books pays a flat 70% at any price, no delivery fees. Kobo pays 70% on prices above ~$2.99 in most markets. Neither requires exclusivity.

The exclusivity question (KDP Select)

Enrolling in KDP Select puts your ebook in Kindle Unlimited and unlocks promo tools, in exchange for ebook exclusivity to Amazon for rolling 90-day periods. The trade is real on both sides: KU page reads can outearn sales in fiction and high-volume niches, while nonfiction and books that double as lead magnets usually lose more by vanishing from Apple, Kobo and your own site than KU pays back. A workable default: launch wide (KDP + Apple + Kobo), and consider Select later for a specific book with evidence, not as a reflex at upload time.

What the stores actually require

• Manuscript: EPUB, validated (stores run their own checks; a clean source-built EPUB passes, a converted PDF rarely does).
• Cover: KDP’s recommended size is 2,560 × 1,600 px at a 1.6:1 ratio, JPEG. It will be judged at thumbnail size in search results, so title legibility at 120 px matters more than detail. A cover that fails the thumbnail test fails the store.
• ISBN: not required for KDP, Apple or Kobo ebooks, since they assign internal identifiers. Buy your own ISBN only if you care about appearing in industry catalogs under your own publisher name. Print editions are a different story; KDP offers a free one.
• Metadata: title, subtitle, description, 7 keywords (KDP), up to three categories. The description is sales copy, not a synopsis; its first two lines show before the “read more” fold.

The launch sequence that matters

1. Pre-upload: read the book once on an actual phone and an actual e-reader. Most formatting embarrassments are visible in ten minutes of real reading.
2. Upload week: publish on all three stores; KDP review typically clears within 72 hours, often faster.
3. Price the launch: a $0.99 or free launch window builds early reviews and rank; the 70% band can wait two weeks.
4. The list does the launching. This is the part the screenshots skip: books launched to an email list sell; books launched to the void rank nowhere. If you do not have a list yet, your first ebook should arguably be a lead magnet, not a product. The email ROI math ($36 per $1, per Litmus) outearns most royalty checks.
5. Post-launch: the book is now an asset. Excerpt chapters as articles, gate a companion workbook, bundle it with a course. One manuscript, several products.

Honest expectations

The median self-published ebook sells double-digit copies, and no tooling changes that. Distribution does. What changed in 2026 is the cost side: producing a credible, typeset, illustrated book dropped from four figures (ghostwriting alone runs $1,000–10,000 at Reedsy rates) to under twenty dollars. When production is nearly free, the rational strategy flips from one perfect book to a portfolio: more niches, more keywords, more shots on goal, each with a real cover and real typesetting, because the store thumbnail and the “Look Inside” preview are still where buyers decide.

The bottleneck moved. It is no longer writing the book; it is choosing topics worth a book and getting eyes on them. Plan for that, and the upload itself is a quiet afternoon.