recipes / app-store-keyword-research

App Store keyword research with Claude

Keyword research by hand is tab-juggling: autocomplete in one window, a popularity tool in another, a spreadsheet holding it together. An agent with openaso connected runs the whole loop in one conversation — and tells you which demand numbers are Apple's own and which are estimates.

the prompt

Do App Store keyword research for a <category, e.g. habit tracking> app
in the US, starting from seeds like <seed 1>, <seed 2>.

Expand the seeds through autocomplete, get demand for every candidate
(tell me which numbers are Apple's real Search Ads popularity and which
are estimates), and check who ranks top today. Finish with a table:
keyword · popularity · source · who owns it · worth targeting?

what your agent does

1. 01

   suggest

   Expands each seed through per-storefront autocomplete — the suggestions users actually see as they type, which is where most real queries come from.

2. 02

   volume

   Scores demand for every candidate: Apple's real Search Ads popularity where Apple carries a signal (source "apple"), a transparent per-signal estimate where it doesn't (source "proxy").

3. 03

   competitors

   Pulls the apps holding the top results per candidate, so "winnable" is judged against who is actually there.

4. 04

   Assembles the verdict table — high demand and weak incumbents float to the top, and flooring or estimate caveats stay attached to their numbers.

what you get back

A scored, source-labeled keyword table. The labeling matters: Apple floors low-demand keywords and carries no signal for some, so a tool that can't say where a number came from will happily point you at noise.

keyword           popularity  source  verdict
habit streaks             38   apple  weak top row — contest it
routine planner           29   proxy  thin demand, cheap to take
habit tracker             62   apple  entrenched leaders — later

make it yours

keep going