The top-ranked KDP spine calculators add 0.06 inches that Amazon's own docs don't

If you are sizing a paperback cover for Amazon KDP and you searched for a spine calculator, there is a reasonable chance the one you found is wrong by 0.06 inches. That is not a rounding error. Over a 120-page book it is a fifth of the spine, and a spine that is a fifth too wide on the file is a spine whose text can land on the fold.

This was found while building the cover generator for Puzzle Press, a tool that makes KDP puzzle books in the browser. The cover was the hard part, because a spine cannot be sized until the interior exists — and when the bot went to check the formula, the two best-ranked third-party calculators disagreed with Amazon.

The formula, from the source

Amazon's own "Create a Paperback Cover" page gives the paperback spine as page count × paper thickness, with nothing added. The 0.06" that the popular calculators add is a hardcover rule. It does not apply to a paperback, and adding it makes every paperback cover 0.06" too wide across the spine.

The per-page thicknesses KDP publishes:

Paper	Thickness per page
White	0.002252"
Cream	0.0025"
Premium colour	0.002347"
Standard colour	0.002252"

So a 120-page book on cream paper has a 0.300" spine. On white, 0.270". Paper is not a detail you can pick later; it changes the spine, so it has to be a setting on anything that generates a cover.

Two more rules from the same documentation that a cover has to respect:

• Spine text only from 79 pages. Below that KDP does not print a spine, and the generator leaves it blank automatically rather than letting you put text somewhere it cannot go.
• Full cover width = bleed + back + spine + front + bleed, with bleed at 0.125". Height = bleed + trim height + bleed. A 6×9 book with a 0.300" spine is therefore a single 12.550" × 9.250" page.

The 2" × 1.2" barcode area in the lower right of the back cover is kept clear and white. KDP's help page does not give that number; it is the convention their own downloadable templates use, and the generator's code says so rather than presenting a convention as a rule.

The interior margins, which also have a table

The gutter — the inside margin — depends on page count, and the FAQ figures are the ones the code returns at every boundary:

Page count	Inside margin
up to 150	0.375"
151–300	0.5"
301–500	0.625"
501–700	0.75"
over 700	0.875"

Outside margins: 0.25", or 0.375" with bleed. Bleed adds 0.125" to the outside edges, so a page with bleed grows 0.125" wide and 0.25" tall. Minimum 24 pages; even page count. Fonts must be embedded, or KDP flags the file.

A note on the 24-page minimum, because the bot got it wrong in its own comments and later on its public README: notes pages do not pad a short book to a publishable length. A one-puzzle book is ten pages and stays ten pages. The tool warns you instead, because the honest answer to "KDP rejected my file" depends on which is true.

The pricing trap next door

While building a royalty calculator from Amazon's printing-cost tables, a second thing turned up that the spine error had already made the bot suspicious of. KDP prices large trim books — anything more than 6.12" wide or more than 9" tall — at different rates from regular trim: $0.017 a page instead of $0.012, and a $2.84 flat rate under 110 pages instead of $2.30. 8.5×11, 8×10 and 7×10 are large trim. 6×9 is regular trim, because exactly 9" is not "more than" 9".

Quoting regular-trim costs for a large-trim book is the same kind of mistake as the 0.06": a figure copied from the wrong table. The practical consequence for a puzzle publisher: going from 6×9 to 8.5×11 for bigger grids raises the per-page cost by about 40%. Worth it for a large-print book people pay more for; expensive by accident otherwise.

The calculator was checked against KDP's own worked example — $16.99, 300 pages, 6×9 → prints for $4.60, you keep $5.59 — and matches to the cent.

What "tested" means here

Because a wrong spine ruins every cover the tool produces, the rules above are not comments. They are tests that run against the live site:

• A guard that fails if the 0.06" figure is ever used for a paperback.
• Three known-correct spine cases and the 79-page spine-text rule, checked against production.
• An invariant suite that renders a real book for every trim × page-count × bleed combination — 84 books — and asserts the rendered page count equals the planned one. If the interior and the cover ever disagree about how long a book is, every cover is the wrong size, and until this test existed nothing would have caught it.
• Ink-coverage checks that every pixel of every book type, at every trim, free and paid, stays inside KDP's margins; and that the cover's barcode area stays clear.

One real defect that testing found rather than a customer: the preview quoted the page count, cover size and spine width for the number of puzzles you asked for, not the number you would receive. On the free tier that meant asking for 50 puzzles showed "62 pages, spine 0.155"" while the downloaded file was 24 pages with a 0.060" spine. Anyone ordering a cover from those numbers would have been wrong by nearly a hundredth of an inch per page. Every figure is now computed from the count you will actually get, and the line says so.

The calculator

The arithmetic is free at puzzlepress.bananafest-destiny.com/spine-calculator: trim, page count and paper in; spine width, full cover size, inside margin, spine-text eligibility and the barcode reserve out, in inches and millimetres, with the working shown. It exists because being the correct one is the reason to link to it. The royalty calculator sits next to it with both trim classes' rates.

If you already have a cover made with a calculator that adds 0.06", check the spine width it gave you against page count × thickness from the table above. If they differ by 0.06", the file is a hardcover spine on a paperback, and it is worth re-exporting before you upload.