Turn the query into an expectation.
Write down what a person is likely trying to accomplish. Then identify the product capability that satisfies it. If the connection requires a long explanation, the phrase may be ambiguous, weakly relevant or simply outside the product’s scope.
Use intent analysis to separate the likely task from the literal words. Preserve uncertainty when several interpretations are plausible.
Check the current product, not the roadmap.
A feature planned for a future release is not the same as a capability available today. Check the version, market and any relevant limitations. A phrase implying offline operation, for example, needs a product that actually supports the expected offline experience.
Bring a product owner into the review when needed. Research cannot resolve a factual feature question by choosing whichever answer makes the keyword list larger.
Ask whether the listing can prove the fit.
If the product genuinely supports the task, identify the screen, workflow or explanation that makes the connection clear. A relevant query paired with vague creative still leaves the visitor guessing.
The matrix below separates supported, uncertain and unsupported ideas. It is an editorial review for a fictional app, not measured keyword performance. Link the conclusion to the visual narrative.
Treat platform accuracy as a requirement.
Apple’s search guidance calls for accurate, relevant keywords and excludes unrelated terms and competing app names. Review the applicable rules for the store you are publishing to rather than assuming every research idea is permissible metadata.
A competitor’s behaviour is not permission to repeat it. Keep policy review separate from demand and difficulty estimates.
Record the decision and revisit when the product changes.
For each important candidate, note the supported task, evidence, constraints and decision. Excluded ideas can remain in a research archive with their reasons. If the product later changes, that record makes a responsible review easier.
Only suitable candidates should move into prioritisation. Pair the decision with demand context rather than letting a number overturn product truth.
