Articles › Keywords & Metadata
Amazon Autocomplete — The Free Keyword Data Most Authors Ignore
Published 12 July 2026 · KDP Metric team
Type "cozy mystery" into Amazon's search bar and stop. Before you press enter, Amazon shows you a dropdown of suggestions: "cozy mystery books", "cozy mystery with cats", "cozy mystery series book 1". That dropdown is keyword research. Real searches, made by real Amazon shoppers with their wallets already out, served to you free and current as of today.
Paid tools dress this up, and some add genuinely useful layers on top. But the raw material they're refining is sitting in the search bar of the biggest bookshop on earth, and most authors never harvest it deliberately.
Why autocomplete is worth taking seriously
Amazon's suggestions aren't guesses or dictionary entries. They're drawn from what shoppers actually type, which gives them three properties nothing else matches for our purposes.
They're commercial. A Google search for "cozy mystery" might be someone looking for a definition. An Amazon search is someone shopping.
They're current. The dropdown reflects what readers are typing now, not a database snapshot from months ago. When a new trope takes off, autocomplete knows before the round-up blog posts do.
And they're ordered. Suggestions appear roughly by popularity, so the position of a phrase in the list is itself a rough demand signal. Rough is the operative word (Amazon publishes no numbers), but for ranking phrase A against phrase B it's honest data with no model in between.
How to harvest it properly
Random typing gets you random scraps. Do it systematically:
- Set the search dropdown to Books or Kindle Store first. Suggestions are department-specific, and you want book shoppers' searches, not everyone's.
- Type your seed and record the base suggestions. "Cozy mystery" gives you the head terms.
- Walk the alphabet. "Cozy mystery a", "cozy mystery b", and so on through the letters. Each letter surfaces a different slice of the long tail: "cozy mystery animals", "cozy mystery bakery", "cozy mystery Christmas". This is the single highest-yield trick in free keyword research, and it takes about ten minutes per seed.
- Flip the word order. "Mystery cozy", "cats cozy mystery". Different orderings surface different suggestions.
- Seed with reader vocabulary, not yours. Try tropes ("enemies to lovers"), settings ("small town"), and audience words ("clean", "for adults") as their own seeds. The dropdown will finish shoppers' sentences for you.
Record everything in a list as you go. Twenty minutes of this on a niche typically yields 40 to 80 real phrases.
Add to Chrome — it's free
Qualifying what you found
A phrase being typed doesn't make it a phrase you can win, so the second pass is qualification. For each promising phrase, actually run the search and look at two things.
First, demand behind it: do the top results have healthy Best Sellers Ranks, or is this a search that leads nowhere commercially? (How to read the ranks.) Second, competition on it: is page one a wall of thousand-review incumbents, or is there room for a newcomer?
The phrases that survive both checks (typed by readers, backed by sales, not sewn up) are the ones worth spending metadata on. And spending them is a solved problem: your strongest phrase belongs in your subtitle, the next seven different searches go in your backend keyword slots, one intent per slot.
A ten-minute harvest, start to finish
Here's what the process actually looks like for a cozy mystery author. Set the search dropdown to Kindle Store and type "cozy mystery". The base suggestions might include "cozy mystery series", "cozy mystery books", "cozy mystery box set". Write them down.
Now walk the alphabet. "Cozy mystery a" surfaces "cozy mystery animals". "Cozy mystery b" gives "cozy mystery bakery" and maybe "cozy mystery beach". "Cozy mystery c" brings "cozy mystery cats", "cozy mystery Christmas", "cozy mystery coffee shop". Keep going. By the time you reach z you'll have thirty or forty phrases you'd never have brainstormed, each one a search a real reader made.
Then flip the order and reseed. Try "cats cozy" and "Christmas cozy". Try adjacent seeds entirely: "amateur sleuth", "small town mystery", "clean mystery". Each new seed spawns its own alphabet of suggestions. Fifteen minutes of this and you have a working vocabulary for the entire niche, in readers' own words, which is far richer than the handful of terms most authors think to use.
The output is just a raw list at this stage. That's fine. Harvesting and qualifying are separate jobs, and trying to judge each phrase as you type slows the harvest down. Collect first, filter second.
Fiction and non-fiction seed differently
What you seed with should match how your readers think. Fiction readers search by the experience they want, so seed with genres, tropes, settings and moods: "enemies to lovers", "cozy fantasy", "dark academia", "beach read". The dropdown will complete those into the specific flavours readers are chasing.
Non-fiction readers search by the problem they have or the result they want, so seed with those instead: "how to", "budgeting for", "learn", "beginners guide". Type "how to stop" into a self-help niche and Amazon will finish the sentence with the exact anxieties people are trying to fix. That completion is priceless, because it's your audience telling you, unprompted, what they're actually looking for.
What autocomplete can't tell you
Fair is fair, so the limits. It gives you no absolute volume numbers; position in the dropdown is ordinal, not a count. It won't show searches too rare to suggest, some of which are still worth owning if they describe your book exactly. It varies by marketplace, so .com and .co.uk dropdowns differ, and if you sell mainly in one, harvest there. And it can be nudged by your own history, so a private/incognito window gives cleaner results.
None of that undermines the core use. For deciding which real searches exist and which matter more, the dropdown is as close to ground truth as anyone gets without Amazon's internal data, including the paid tools.
The lazy version
Everything above is free and manual, and honestly worth doing by hand at least once, because you develop a feel for your niche's language. After that, it's clicking and typing that a tool does faster: KDP Metric's free Keyword Explorer pulls the autocomplete suggestions for any seed straight from Amazon, tiers them by their dropdown position, and lets you analyse the competition behind any phrase in one click, without leaving the page you're browsing. Same data, same honesty about what it means, minus the typing.
Add to Chrome — it's free