How readers actually choose a book
A reader almost never starts with your book.
They start somewhere else — a category page, a friend, a post they scrolled past, a search for something next to what you wrote — and your book arrives late, as one option among many. What happens after that happens in a fixed order. Every cover and blurb decision you will make is downstream of that order, so it is worth knowing exactly.
How they arrive
Four routes account for nearly all of it.
Browsing a category. A store list, a bestseller grid, a table in a shop. Your book is seen beside thirty others, at the same size, in the same moment.
A recommendation. Someone they trust said the book was good — a friend, a reading group, a reviewer they follow, an author they already read.
Seeing it mentioned. A newsletter, a podcast, a review, a post. Weaker than a recommendation: the book gets remembered rather than bought, and they look it up later, which drops them into a search result.
Searching for something adjacent. Not your title, which they do not know. The subject, or a book like yours. What comes back looks like a category grid.
Three of those four end in a grid of covers. That is why the first link in the chain is what it is.
The chain, in order
The cover thumbnail decides whether they stop at all. It is not being judged; it is being sorted. The eye moves across the grid and either catches on your book or does not, in a fraction of a second, at the size a phone shows it. Your cover is not competing with its own full-size version — it is competing with the covers on either side of it. If this step fails, nothing after it happens, no matter how good it is.
The title and author confirm what kind of book it is. The cover made a promise about genre; the title either agrees with it or muddies it. A reader who has to hold two contradictory signals moves on rather than resolve them. If they already know your name, the chain short-circuits here and most of the rest is a formality. See Choosing a title that can be found and remembered.
The blurb's first line decides whether they keep reading. They are not reading the description, they are sampling it. The opening sentence is the whole test; everything below it only gets read if that sentence earns the scroll. This is why a blurb that opens with setup — a world, a date, a paragraph of context — loses to one that opens with the situation. See How to write a blurb that sells the book.
Price and reviews resolve the remaining doubt. Both answer the same question: how much am I risking? Price is read against what the reader expects to pay for this kind of book, so a price far off the category's norm reads as a signal about the book rather than a bargain. Reviews are skimmed for a pattern, not counted — a reader looks for whether complaints describe something they would mind. See Pricing a self-published book.
The sample closes it, where one is offered. The first pages settle whether the writing is for them, and a sample that opens with a wall of front matter or uneven spacing spends its first impression on production rather than prose. See The sample, and what it is selling.
In the tools. The Cover Maker checks your design at thumbnail size, which is the only step in that chain where a tool can tell you something you cannot see yourself; the Interior Formatter produces the sample the last step reads. Start a book → · about the tools
What this means for the cover
At step one, the cover has exactly one job: survive a grid at a small size and say what kind of book this is. Not summarize the plot. Not reward close attention. Anything that only works at full size is being spent on readers who have already stopped.
That is why the thumbnail test exists, and why detailed illustration loses to a strong shape. Detail dissolves at small sizes into texture; a shape with real contrast survives. See The thumbnail test and Composition and negative space.
Recommendation is the strongest force in the chain
It is also the one you can least control. A recommendation skips the grid entirely — the reader arrives already sold and uses the cover only to confirm they have the right book. No design decision manufactures this. The thing that earns it is the book.
What you can do is make a recommended book easy to find and easy to pass on: a title someone can repeat from memory, a cover recognizable from a description, and a series that visibly belongs together so the next one is obvious. See Series covers: the same shelf, three books along.
The two levers you still have
After publication, the text is fixed in practice and the recommendations are out of your hands. The cover and the description are not. Both can be changed in an afternoon, on the same listing, with the reviews intact — which makes them the only parts of the chain you can test, improve and test again.
Diagnose in chain order. If nobody stops at the thumbnail, the blurb is irrelevant; if plenty click and few buy, the cover is working and the description is not. Fixing the wrong link is the most common wasted month in self-publishing. See When to redo a cover, and when to leave it alone.
Make yours
Everything here is built into the tools, and they are free to use: name your book, bring your manuscript or your artwork, and work as long as you like. You only pay when you want the files without the preview watermark.
Next: Covers that age badly · More: What makes a book cover sell · Comp titles: reading the shelf you want to be on · What a cover cannot fix