Metadata before marketing

There is a version of book marketing that begins with a launch plan, a graphic, and a budget. It is the version most authors run, and it skips the step that everything else depends on.

Before a reader forms an opinion about your cover, a retailer has already decided whether to show it to them. That decision is made from your metadata: the keywords you chose, the categories you sit in, and the books the store thinks yours resembles. Get it wrong and paid traffic becomes the only way anyone finds you. Get it right and the store starts doing unpaid work on your behalf, indefinitely.

Keywords are queries, not adjectives

The most common mistake is filling keyword slots with words that describe the book rather than words a reader would type. "Poignant" describes your novel. Nobody searches for it.

Write down ten phrases a reader would actually enter if they wanted your book but did not know it existed. Not your themes — their situation. "Books like Educated." "Small town second chance romance." "Grief memoir mother." Then check each one in the store's own search bar and watch what autocompletes. Autocomplete is a record of real demand, free of charge.

Keep the phrases that autocomplete. Discard the ones that return nothing, however well they describe your book. You are not trying to be accurate here. You are trying to be findable.

Categories reward the specific

Authors chase broad categories because the numbers look bigger. Broad categories are where you are invisible. A narrow category with genuine readers is worth more than a wide one where you sit at position four hundred, because visibility inside a category compounds — rank brings readers, readers bring rank.

Look for the deepest category that still honestly describes your book. If you have to squint to justify it, go back up one level. Miscategorising gets you returns and one-star reviews from readers who wanted something else, and both are more expensive than the visibility was worth.

Comp titles are a promise

Your comparable titles tell the store which readers to test you on. Choose books that share your reader, not your quality. A debut that comps itself to a decade-old bestseller is teaching the algorithm nothing, because that book's readers are too broad a group to be useful.

The better comp is a book from the last two or three years, selling steadily rather than spectacularly, with a reader you could describe in one sentence. Those readers are reachable. A million-seller's readers are not a segment, they are the public.

A ninety-minute version

• Write ten reader-phrased search queries. Test each in the store's search bar. Keep the ones that autocomplete.
• Find the narrowest category that describes your book honestly. Note its top twenty and read what those covers and blurbs have in common.
• Pick three comps from the last three years with a clearly describable reader.
• Rewrite your first blurb line so it contains your strongest keyword phrase in natural language.
• Set a reminder for ninety days. Metadata is not permanent — it is a hypothesis you revise.

None of this costs anything. All of it decides what your money buys later.