How to do App Store keyword research without guessing search volume

Turn product capabilities into seed terms

List the actions people can complete in your app, not just its category. A reading app might help people record finished books, set reading goals, or save quotations. Those actions suggest different search clusters. Include a term only if a new user can reach the promised function in the shipped product.

Use reviews and support conversations to hear how people describe the problem. Do not treat one expressive review as a representative survey. Group similar phrases, preserve the original wording in your research notes, and record why you think each cluster fits.

Inspect what the search currently means

Search each candidate in your intended country and inspect the apps returned. A phrase can sound right but mostly return a different kind of product. Record whether the visible results share your use case, which promises repeat, and whether specialized apps appear alongside broad platforms.

Back My Side samples Apple’s public search response, up to 200 results. The observed position can differ from a person’s on-device experience. If your app is absent, record not found in the sample rather than assigning a lower position. A response full of relevant apps tells you about competition and intent, not the number of people searching.

Choose a shortlist you can defend

Use a relevance gate before relative popularity, difficulty, or opportunity indicators. Reject a high-scoring term if it promises an unsupported feature. For the remaining candidates, inspect the result set and ask whether your app offers a recognizable reason to choose it.

A useful research sheet contains the term, country, date, intended job, evidence of fit, current sampled position, and decision. Keep untested ideas in a separate queue. A small shortlist with explicit reasoning is easier to learn from than a hundred unexplained scores.

• Keep a primary cluster that clearly describes the core app.
• Add narrower clusters for differentiated features.
• Exclude competitor brand names from your submitted keyword field.
• Mark volume as unknown when you do not have a defensible source.

Turn research into a listing and tracking plan

Allocate selected ideas across the name, subtitle, and keyword field without repeating the same words unnecessarily. Read the visible text aloud: people need to understand it before the app deserves a download. Save the original listing so a change can be reversed if it creates confusion.

Track a stable subset of queries and review it alongside your own App Store Connect metrics. Do not rotate the entire tracking set every week and then compare its average rank as though nothing changed. Keep new experiments separate from the baseline cohort.

Repeat research when product capabilities or the intended market change. A new feature may justify a new cluster; a promising generic phrase does not justify adding a feature claim that is not true.

Frequently asked questions