Pick the marketplace
Choose the store you publish into. The completion feed and its language follow that storefront.
Type a niche and read back what the store’s search box finishes for it. A phrase that completes is a phrase someone searched often enough for the store to learn it — the only honest signal this feed carries.
Every lookup is a request to the store on your behalf. A free account keeps that traffic attributable and rate-limited, and takes a minute to create.
Choose the store you publish into. The completion feed and its language follow that storefront.
Enter the niche in the words a reader would use, not the title you have in mind. Two characters are enough to start.
Filter the completions down to distinct buyer phrases, then paste them into the seven keyword fields in the backend.
A clever title is what a buyer sees after the search. The keyword field is what decides whether the listing appears in that search at all.
A new listing has one short window to find its phrase. Filling the fields with adjectives instead of searches spends that window on silence.
There are seven fields, not seventy. The job is choosing seven phrases that do not repeat each other and match the book you actually wrote.
Validate that readers search the idea at all. A phrase that never completes is a warning before you write 60,000 words.
Run the search again in the target marketplace. A translation is a new set of buyers typing a new set of phrases.
Re-check the fields before a seasonal campaign. Completions drift with demand, and a stale list quietly stops matching.
Every phrase comes from the store’s own autocomplete. There is no modelled search volume to argue about, only phrases real buyers finish typing.
Results are filtered to the seven-slot limit and deduplicated, so the output is what you paste, not a research document to interpret.
Use this tool for the phrase, then the niche finder for the shelf behind it. Demand and competition read best together.
Yes to run, within a per-account limit. The limit exists because every query is a request to the store’s search box on your behalf, and a reload loop must not become the store’s problem.
The store’s own autocomplete, not a third-party volume database. A phrase that completes is a phrase the store has learned from real searches; a phrase that does not complete is not shown.
No. Amazon does not publish per-keyword volume, and any tool that claims a precise number is estimating. The completion list is the honest signal this feed carries.
Instead of one completion call, the sweep asks the store for every letter after your seed and merges the answers. It sends roughly twenty-six requests, so it is throttled harder than the plain list.
The KDP backend accepts seven keyword fields. More phrases than that do not fit, so the tool filters completions down to distinct, buyer-shaped phrases instead of handing you a list of ninety.
Eight storefronts, including the US, UK, Germany, France, Italy, Spain, Japan and Canada. Pick the one you publish into; the completion language follows the store, not the interface.
Yes. Print and ebook listings share the same backend keyword fields, so the phrases you pick here apply to both editions.
No. A phrase that completes in the US store may return nothing in Germany. Run the search per marketplace and keep the phrases in the language buyers actually type there.

The interesting question is not whether the prose is pleasant. It is whether the prose obeys: a point of view, a tense, a rule, a fact, an ending.

The question is not which one writes better sentences. It is which parts of the job you are actually buying, and which parts stay with you either way.

A generator can produce a competent draft in an afternoon. It cannot produce your cases, your numbers, your taste, or the decision about what the book is willing to say.