What fields does App Store search index?

The title (30 characters), subtitle (30) and the hidden keyword field (100) — per locale, with in-app purchase display names also surfaceable in search. The description, promotional text and screenshot captions are not indexed on the iOS App Store. Apple combines single words across the indexed fields to build ranked phrases.

What are the main App Store ranking factors?

Beyond keyword relevance in the indexed fields: download velocity on the search term, tap-through and conversion behavior of searchers, rating count and average, and retention-shaped engagement signals — the last tier widely observed but never confirmed by Apple. Field placement matters: title outweighs subtitle, subtitle outweighs the keyword field.

Does rating affect App Store search ranking?

Ratings influence ranking both directly (a relevance signal) and indirectly — the star average shows in search results, so it moves tap-through, and user behavior on your result feeds ranking. An app slipping below ~4.0 fights the algorithm and the human simultaneously.

FunnelHound / Guides / How search works

How App Store search actually works: documented, replicated, folklore

By Emin — indie iOS developer building FunnelHound · Published 28 Aug 2026

Nobody outside Apple has seen the algorithm — including everyone selling you a course on it. What exists is three tiers of evidence: what Apple documents, what the community has replicated so often it's effectively fact, and folklore that survives on confidence alone. Knowing which tier a claim lives in is most of ASO literacy.

Tier 1: documented (bet your metadata on it)

Three fields rank you: title (30 chars), subtitle (30), hidden keyword field (100) — comma-separated, per locale. In-app purchase display names can also surface in search.
Description doesn't rank. It's conversion copy. Same for promotional text.
Search is per-storefront: each country indexes its own localized metadata. Blank locale, invisible app.
Ratings and downloads matter. Apple says relevance combines text match with app quality and popularity signals — deliberately vague, officially real.

Tier 2: replicated (safe to act on, watch for drift)

Field weight order: title > subtitle > keyword field. Why renaming shakes everything.
Word combining: Apple builds phrases across fields — title "Focus" + keyword "timer" ranks for "focus timer". Foundation of the no-repeats rule.
Download velocity on a term lifts rank on it — the engine behind the Search Ads halo, and why momentum compounds.
Searcher behavior feeds back. Results that get tapped and kept rise; results that get skipped or deleted in minutes sink. Retention-shaped signals are the strongest "everyone sees it, Apple won't say it" candidate.
Indexing is slow: 2–4 weeks for new metadata to settle — the entire basis of the week-six patience rule.

Tier 3: folklore (stop spending cycles here)

"Screenshot captions are indexed." No. Screenshots move conversion, not rank.
"Repeat the keyword everywhere for emphasis." Repeats add nothing and burn characters.
"Update frequency is a ranking factor." Updates help indirectly (fresh ratings, fixed crashes, re-index on metadata change) — no credible evidence of a freshness boost per se.
"Reviews' text is keyword gold." Recurring claim, never cleanly replicated. Reply to reviews for humans and ratings, not for the index.

What this means for a Tuesday morning

Spend where the evidence is: nail the three indexed fields per locale (tier 1), respect weights and combining (tier 2), then let the behavioral signals work for you by shipping an app people keep — the algorithm is ultimately a proxy for "searchers were happy". And measure like a skeptic: whatever tier a tactic came from, it either moved your search impressions or it didn't. The funnel doesn't care whose course said it would work — which is exactly why FunnelHound watches impressions → installs → purchases instead of theories.

Theories are free. Impressions are evidence.

FunnelHound shows what each ASO change did to your actual funnel, from App Store Connect, milestone-marked.

Get FunnelHound

Share this on X

Data notes: tier assignments reflect Apple documentation and widely replicated community findings as of 2026; Apple revises search behavior without notice, so tier 2 items deserve periodic re-verification. Anyone claiming complete knowledge of the algorithm is selling something.