Number puzzle books for KDP
The number shelf on Amazon has a default, and the default is sudoku. Empty 9×9 grids on the covers, empty 9×9 grids in the previews, a thousand of them, and a buyer who has solved so many that “another sudoku book” is a decision they make in half a second.
A number book that is not sudoku competes in a different way. Far fewer shoppers are hunting for it by name, so the cover cannot win the click on recognition — and the moment someone does open the preview, the first thing they are checking is whether this is the same grid again. That single preview page is where the whole listing is decided.
This page is the shelf decision for the non-sudoku number interiors: which look to publish, what the preview has to prove for each one, what the answer-key section does to your page count and your royalty, and how one look turns into a series. The sudoku family — classic, Wordoku, jigsaw, killer — has its own shelf and its own hub at KDP sudoku variations.
Sudoku’s advantage, and the part of it you can borrow
Sudoku’s real advantage on Amazon is not demand. It is that the rules cost the publisher nothing. Everyone already knows them, so a sudoku listing spends its preview proving one thing only: that the difficulty is what the buyer wants.
Every other number puzzle costs you exactly one explanation, and the preview has to carry three jobs instead of one:
- Prove it is not sudoku — in the first interior page the shopper reaches.
- Teach the rule in one sentence — because the buyer who arrived from a sudoku search has never heard of your mechanic.
- Prove the difficulty band — the same job the sudoku book has.
Publishers lose this shelf on job two more often than job one. The four looks below are all visibly different from sudoku; what sinks a listing is a preview that shows an unfamiliar grid and never says what to do with it. The review then reads “I couldn’t work out what I was supposed to do,” and it is fair.
The titling side of the same problem — whether to name the mechanic or name the appetite (“Not Another Sudoku”) — is argued out on the number puzzle family page. Do read it before you write the subtitle, because on this shelf the mechanic name in the subtitle is the only thing the small, high-intent audience can search for.
What Look Inside has to prove on a number book
Amazon shows a fixed slice off the front of the file. You do not choose which pages; you choose what sits in front of them.
Land a real board fast. Title page, copyright page, one short how-it-works page, first puzzle. That is the whole front matter budget. A number book whose preview runs out during a foreword has spent its one chance to prove it is not sudoku.
Put the rule in one sentence, not a chapter. One sentence for the rule, one small worked example, done. Anything longer turns the sample into a tutorial and reads as if the puzzle is hard work. The general version of this argument is on what a Look Inside has to prove.
Show it at true print size. Digits shrink faster than corridors do. A 12×12 grid that is comfortable on a monitor can print with cells too small for an adult to write a number in, let alone a large-print buyer. Print one page on the trim you intend to sell and write in it with the pencil your reader will use. Cell size, not page size, is the thing to measure — see trim size and page count and large print puzzle books.
Understand that difficulty is half-invisible here. A maze buyer can see that a maze is dense. A number-grid buyer cannot see that a grid is hard, with one exception worth building the preview around: on the looks that print givens — cages and 0/1 grids — given density is visible, and a board with a comfortable scatter of printed numbers usually is the easier board, because the easier bands leave more of them in. On a Hitori page, where every cell already carries a number, even that signal is gone and grid size is the only difficulty cue a shopper gets. Either way: if the cover promises easy, page one is a small board, not the same big grid with a different word on the cover. Difficulty progression and the ladder inside one volume go further.
Say where the keys are, because the preview will never get there. The answer section sits at the back, past the slice Amazon shows, so the contents page and the description have to state that every puzzle has a solution and where to find it. That is a trust problem on any puzzle book (answer keys, key placement and Look Inside trust) and a sharper one here, because on the shading looks the key page is also where the puzzle’s whole visual signature lives:
That is the same board twice. The left page is what a shopper previews; the right page is what a Hitori actually looks like when it is finished — and what your print bill pays for. Which is why the worked example at the front should show a part-shaded board rather than a clean one: it is the only place in the preview where the buyer sees the puzzle’s real look.
The four looks, and what each page one proves
Four live modules give you interiors that cannot be mistaken for each other, let alone for sudoku. One look per title.
Smart Numbers — a line of values and a blank
Smart Numbers prints hidden-term sequences: a row of numbers, a missing value, a rule the solver has to find. There is no grid at all, which makes it the most visibly not-sudoku page on the shelf and the easiest rule to state in one line.
Its risk is the opposite of the others: sameness. If every line is add-three, the preview has shown the buyer the entire book. The generator draws from a library of formula families — arithmetic steps, doubling and other geometric runs, alternating steps, growing differences, digit rules, and recurrences that build each term from the two before it — and the visible family list is scoped to the difficulty you pick, so an Expert book is a different set of rules rather than the Easy book relabelled. Treat family mix as a publishing decision: no single family over about a quarter of the pages, and at least three of them visible inside the preview slice.
The other thing this look changes is page-count arithmetic. A page holds one to five sequences, so “100 puzzles” is not 100 pages here — it is somewhere between 20 and 100, and you choose. Fewer sequences per page with bigger type is a legitimate large-print product; the same count crammed three to a page and padded with blank sheets is not. The listing-level detail lives in Smart Numbers puzzle book for KDP.
Number Blocks — bold cages, digits 1 to N
Number Blocks is sold elsewhere as Suguru and as Tectonic, which is three keyword routes into one book and worth putting in your backend keywords on day one. A grid is carved into bold-outlined blocks of one to five cells. A block of N cells holds exactly the digits 1 to N, and equal digits never touch — not even diagonally. That diagonal half of the rule is the part listings leave out, and leaving it out teaches your buyer a game they will solve wrongly.
This is the closest of the four to sudoku, so the preview has one extra job: the cage outlines have to be unmistakably heavier than the cell grid at print size. If they print at similar weights, the page reads as a sudoku with a printing fault. Proof that page before anything else.
Sizes run 5×5 to 11×11, plus 5×10 and 10×5 — the tall and wide formats are the ones to reach for on a narrow trim or when you want fewer, larger cells on the page. Difficulty is Easy, Medium or Hard, and it shows in the givens: Easy keeps a comfort floor of printed digits, while Medium and Hard thin them down to what the deduction strictly needs. That makes an Easy Number Blocks page genuinely look easier in the preview, which is the rare case where the shelf and the honest product agree. The digits are the puzzle here, so the kid-facing version of this listing is a smaller grid, not a symbol swap.
Hitori — the grid arrives full
Hitori inverts the sudoku page. Every cell already holds a number, and the solver shades cells out until no number repeats in any row or column among the cells left white — with two more rules that make it a real deduction: shaded cells never touch edge-to-edge, and every white cell must stay connected to the rest. State all three in your one-sentence rule line, or the book teaches a puzzle it does not contain.
A full grid is the strongest anti-sudoku signal on this shelf, because sudoku’s signature is the empty cell. Nothing else in the number aisle previews like a page with no blanks on it.
Sizes run 5×5 to 12×12, and difficulty works by how much of the grid ends up shaded, so a series climbs by size at least as much as by band: a 6×6 Easy and a 12×12 Hard are different products in the same look. The production fact to plan around is the one in the figure above — key pages print roughly a quarter to a third of their cells as solid dark fill. Proof a key page on the paper you are ordering, watch for show-through onto the reverse, and put the keys in a back section rather than on the back of each puzzle, so the ink-heavy pages are not sitting behind the pages people write on. Hitori puzzle book for KDP covers the listing itself.
Binary — two symbols, balanced
Binary Puzzles is the starkest page of the four: an even-sided grid from 6×6 up to 16×16, a handful of given 0s and 1s, and a lot of white space. Fill the rest so no three identical symbols sit in a line and every row and column splits half and half. Solvers also know it as Takuzu and Binairo — more backend keywords.
Two traps, both listing-side. First: print the rule you actually generated. There is an optional extra Unruly-style constraint — no two rows or columns identical — that is off unless you switch it on, so a description promising it on a file generated without it is simply wrong. Second: size and legibility pull against each other harder here than anywhere else on the shelf. A 16×16 is 256 cells; put that on a 6×9 page and no large-print claim survives it. Big grids want a big trim, and a large-print Binary book wants 6×6 or 8×8.
Binary is also the one look on this shelf with an easy route to a second audience: on Creative and up the two states can print as letters, colour shapes or images instead of 0 and 1, which is a kids or seasonal title with the same engine behind it. That is a separate listing with its own cover promise, not a bonus chapter in the adult book.
Page count, keys and the royalty
Print cost comes out of your royalty, so page count is a decision you make rather than a number the generator hands you.
On the three grid looks, each puzzle prints with a solution, and the key layout is the lever: one solution per page turns a 100-puzzle title into roughly a 200-page book, while ganging several keys to a page keeps the same 100 puzzles nearer 130 pages. Both are honest; they are not the same print bill or the same spine. On the sequences look the arithmetic changes again, because several puzzles share a page.
Two consequences worth acting on. A hundred puzzles is not automatically a better product than sixty — it is a thicker book at a higher print cost that your cover now has to justify. And a “big book of 300 number puzzles” is usually a padded easy band, which the preview shows and the reviews say out loud.
One look per listing, then the next volume
One look per title. A sequences book, a Number Blocks book, a Hitori book, a Binary book: four listings, four page ones that share nothing. A mixed number interior is honest only when the title says collection, and even then four clean single-look titles usually outsell one grab-bag — the argument in full is on when not to mix puzzle types.
Grid size and difficulty are series levers inside a look, not new looks. Three Hitori titles at 6×6, 8×8 and 12×12 are a ladder a buyer can climb and a series page that sells the next one for you; the same three grids shuffled into one volume are a book nobody can be recommended. Volume two is the cheapest book you will make, because every decision was made for volume one — puzzle book series on KDP is the whole pattern.
What the software does after the puzzles exist
Puzzle Book Studio is the assembly step for whichever look you picked: trim, layout, running heads, the key section, the print PDF. Batch generation — a run that walks a whole size and difficulty ladder in one pass, which is how a three-volume series stops being three days of clicking — is the Productivity tier.
Three edition facts, because they decide what you can and cannot promise. Every paid edition produces the static print output a KDP interior needs. The Customer JSON data files, if you ever want the puzzle data rather than the pages, are Productivity. Web play on your own site is the Interactive add-on, bought per puzzle type alongside an edition, and it changes nothing at all for an Amazon-only publisher; if you do want a playable free sample beside the paperback, that funnel is interactive puzzle embed. The free demo is for evaluating the puzzles, not for producing gated output. Which edition should I get lays out the rest.
Commercial rights to publish the generated puzzles are part of the Puzzle Maker Pro license as stated on the homepage; check the edition you buy if you are selling. That is not a substitute for reading Amazon’s own content rules.
The same four looks on other channels
The Amazon interior is one product out of a set. The same boards, cut to a dozen pages with a theme and keys in the ZIP, are an Etsy listing — number puzzle printables for Etsy. One board with its key, sized for a hand-out, is a classroom worksheet: sequences go to TPT math puzzle worksheets, the shading and 0/1 grids to TPT logic puzzle worksheets. Same generator, a different contract each time.
The whole point
The number shelf defaults to sudoku, and that default is beatable in one preview page — but only if the page proves three things at once: this is not that grid, here is the rule in a sentence, and here is the difficulty you wanted. Pick the look whose page one cannot be mistaken for anything else, keep the cells writable at print size, decide the key layout before you count pages, and let the second volume be the easy one.