App Keyword Research: From Competitor Phrases to a Scored Shortlist
A practical app keyword research process: mine competitors, expand with store autocomplete, filter junk, score relevance and difficulty, then choose what to write.
**App keyword research** is the work of finding the phrases people type into a store, then deciding which ones your app can realistically satisfy and rank for. The hard part is not collecting phrases. It is throwing most of them away.
This guide shows a repeatable process that starts with competitors, expands through real store suggestions, filters for intent, and ends with a scored shortlist you can actually write into a listing.
Key takeaways
- Competitor listings are the cheapest source of category language.
- Store autocomplete confirms phrasing; it does not give you exact volume.
- A quality gate matters more than list size: two- and three-word phrases beat one-word noise.
- Score relevance before difficulty. A hard phrase for the wrong user is still wrong.
- End with five to ten phrases that belong in high-weight fields.
Start with competitor language, not a blank page
Your competitors have already spent time learning what the category calls itself. Read three to five apps that rank for the obvious phrases. Capture their titles, subtitles, and the phrases they repeat across descriptions. Look for two-to-four-word sequences that appear in more than one listing. Those repetitions are your roots.
Do not stop at the category leaders. Mid-sized apps often reveal the longer, more specific phrases that a big brand ignores. NewASO calls these newcomer keywords and separates them from phrases giants own, because the two require different strategies.
Expand with live store autocomplete
Take each root and type it into Google Play and the App Store search bars. Record the suggestions. Then try a few variants: the root plus a benefit, a use case, a device, or a user type. Autocomplete is direct evidence that someone types the phrase in that store, even though it says nothing about how many people do.
- Expand in the store and country you are targeting; suggestions differ by market.
- Add plain-language synonyms, not internal jargon.
- Keep the source of each phrase so you can audit the list later.
- Reject phrases that describe a different job to be done.
- Deduplicate aggressively; the same phrase arrives from several roots.
Run a quality gate before scoring
Raw autocomplete is noisy. One-word suggestions are often broad navigational queries or unrelated brands. A useful gate keeps phrases of two or three words, removes duplicates, drops terms your app cannot honestly satisfy, and caps generic one-word roots unless they are proven category terms.
| Phrase | Keep? | Reason |
|---|---|---|
| blood pressure log | Yes | Specific job, matches a real app function |
| blood pressure | Maybe | Broad, high competition, weak intent |
| blood pressure app free | Yes | Clear intent, but confirm the app is free |
| health | No | Too broad to convert or compete |
| blood pressure monitor | Careful | Implies hardware measurement the app may not do |
That last row matters. A phrase can be popular and still be dishonest for your product. Relevance is not just topical; it is a promise you can keep after install.
Score the list on five signals
A consistent score forces comparison. NewASO uses volume index, competition, difficulty, relevance, and opportunity, each on a 0–100 scale with a confidence label. Volume is modelled from autocomplete position, source overlap, and phrase length. Competition and difficulty come from live search results. Relevance is judged against the app's actual job.
Opportunity is the sorting key: relevance multiplied by volume and reduced by difficulty. That means a moderately searched phrase with high relevance can outrank a huge phrase with weak fit. The score is an estimate, not a measurement, and the confidence label tells you how much to trust it.
Turn scores into a writing plan
- Pick one Focus phrase for the title or subtitle.
- Choose two or three Primary phrases for the second field and keyword field.
- Move Secondary phrases into the description and supporting screenshots.
- Park Maybe phrases for another market or a future update.
- Write the Skip list down so you do not re-research it next month.
How NewASO handles the boring parts
NewASO starts from competitor roots, adds LLM synonyms, expands through Google Play and App Store autocomplete, and deduplicates the result. A real project can produce over a thousand phrases. The value is the next step: scoring and sorting them into Focus, Primary, Secondary, Maybe, and Skip so the list becomes a decision.
Everything stays in a local SQLite project. If you want the exact scoring inputs, read the features page. If you are still choosing where to start, the ASO checklist puts research in order.
Frequently asked questions
How many keywords should app keyword research produce?
A first pass might collect forty to a few hundred phrases depending on the niche. The useful output is much smaller: five to ten phrases for high-weight fields and a ranked list for descriptions and future updates.
Is Google autocomplete the same as App Store autocomplete?
No. Google web suggestions reflect web search behaviour. App Store and Google Play suggestions reflect store queries. Use the store search bar for ASO, and use web suggestions only as a secondary signal.
Can I do app keyword research for free?
Yes. Competitor listings, store autocomplete, and manual rank checks are free. Paid tools add convenience and broader data, but the core research loop does not require them.
Should I target a keyword my app only partly solves?
Usually no. Partial relevance may produce installs, but it also produces disappointed users, poor retention, and negative reviews. Choose phrases where the app delivers the promise.