tools / suggest
suggest
read-onlyKeyword suggestions
What does the App Store itself suggest when users start typing?
suggest is an App Store autocomplete API: it returns the suggestions Apple shows in the store's search box for a seed term, in the order Apple ranks them. Those suggestions are the store steering real demand — users tap them instead of finishing their query — which makes autocomplete the highest-signal, lowest-cost keyword discovery source there is.
Suggestions are per storefront and language: habit autocompletes differently in the US and in Japan. Each country in the call is queried separately against Apple's public MZSearchHints endpoint — the same one the store client uses.
The list is discovery input, not a verdict. Feed promising suggestions into volume for a demand number, competitors for who wins them today, and rank for where you stand.
worked example
What does the US App Store suggest for the seed “habit”?
{ "name": "suggest", "arguments": { "keyword": "habit", "countries": [ "US" ] } }
{ "keyword": "habit", "storefronts": [ { "country": "US", "suggestions": [ "habit tracker", "the habit burger grill", "habitica", "habit", "habitshare", "habitify", "plainhabit: habit tracker", "habit tracker free", "habitkit", "habitlink" ] } ] }
real response, captured 2026-07-29 — trimmed only where marked with …
- Order is Apple's ranking of the suggestions — position 1 (
habit tracker) is what the store pushes hardest for this prefix. - Brand terms (
habitica,habitkit) and off-category noise (the habit burger grill) appear exactly as the store serves them; filtering is your agent's job. - Long-tail variants like
habit tracker freeare often the winnable ones for a new app — high intent, weaker incumbents.
when to use it
- Keyword discovery from a seed list: run your category's head terms and harvest what Apple itself surfaces.
- Prefix mining:
hab,habi,habitreturn different lists — how deep your term appears for short prefixes is a demand signal in itself (that is one component ofvolume's estimate). - Checking localization assumptions: the same seed in
DEorJPreveals the vocabulary users actually type there. - For a full scored pipeline in one call — suggestions discovered, then ranked by demand, competition and your current position — use audit instead.