Group research phrases by the reader problem they express, compare each group with the actual manuscript, and remove terms that misrepresent the book. Keep a short evidence note for each candidate. Treat observed language as evidence of wording, not proof of search volume, competition, or guaranteed visibility.
Preserve where a phrase came from
Record the phrase, source type, date, and context in your research notes. A phrase from a reader question means something different from a phrase copied from a competing title. Keep enough context to understand what the person was trying to solve.
Do not collect private or restricted material for this exercise. Use appropriate public observations or feedback you are permitted to analyze. The goal is to understand reader language, not to reproduce someone else's description or assemble a large list whose origins and relevance can no longer be explained.
Group phrases by intent
Separate phrases about learning a task, choosing an option, solving a problem, or finding a particular kind of book. Terms that share a word may still reflect different needs. A guide to repairing furniture and a photo collection of antique furniture serve different intentions.
Name each group in plain language and identify the corresponding manuscript content. If a group has no meaningful support in the book, set it aside. This step keeps the research grounded in what you can accurately offer rather than encouraging you to chase every adjacent topic that appears popular.
Build a small candidate set
Choose representative phrases that describe the book naturally and specifically. Avoid piling up near-identical variants when they do not clarify a different aspect of the content. Keep a note explaining the reader problem and the relevant chapters for each candidate.
If you use a research tool, distinguish its estimates from directly observed facts. Do not turn a displayed metric into a certainty about future sales or visibility. The usefulness of a phrase still depends on whether it describes the book and complies with the platform's current metadata rules.
Check accuracy and allowed use
Review current KDP guidance before entering the final fields. Remove misleading associations, unsupported claims, and terms that could imply the book is something it is not. Be especially careful with names, brands, and descriptions of other works.
Do not infer that a competitor's visible wording is permitted or effective simply because it appears online. Their listing is an observation, not an authoritative rule.
- Retain the source context for each phrase.
- Group candidates by reader intent.
- Map them to actual manuscript coverage.
- Separate estimates from observed evidence.
- Check the current metadata guidance.
Keep a dated decision record
Save the candidates selected, those rejected, and the reason for each consequential decision. Record the book revision the choices describe. If the manuscript changes, the relevance review should change with it.
Evaluate later adjustments carefully rather than changing several metadata elements at once and attributing every fluctuation to one keyword. Many factors affect discovery and purchasing. A clean research record supports more thoughtful decisions, but it does not turn a small amount of listing observation into a controlled experiment.
Continue with this topic
Frequently asked questions
Clear answers for this publishing decision.
Do reader phrases prove search demand?
No. They show how people express a need in the observed context. Search volume and purchasing demand require different evidence.
Should I copy a successful competitor's keywords?
You usually cannot infer their complete strategy or results from a listing. Use observations to understand readers while choosing accurate terms for your own book.
How often should I revise the candidates?
Review them when the book's scope changes or when you have meaningful new evidence. Avoid constant changes based on unsupported assumptions about ranking.
What should the title, subtitle, description, and keywords look like before I upload?
Check the mechanical limits first, then make sure title, subtitle, categories, and description describe the same book for the same reader; a keyword is useful when it matches how that reader searches, not because it is popular.
Draft the listing fields in a book project