# Ratings and reviews in App Store search: per-storefront and mostly indirect

> Ratings are per-storefront, act on rankings mostly through the conversion loop, and reset only all-at-once. Review text ranks nothing — but it is the best keyword research you did not have to run.

## Per-storefront, not global

The star average and rating count shown on a listing are **scoped to the storefront**: a 4.8 in the US coexists with a 3.9 in Germany, each computed from its own users, each feeding its own market's [ranking signals](https://openaso.ai/reference/search-ranking-factors.md). Any analysis that quotes one global rating is reading a number that does not exist — pull ratings per country, the way the [reviews](https://openaso.ai/tools.md) tool returns them.

## How ratings act on search

- **Through conversion** (the main path): the star shown in search results moves tap-through and install rate, and those behavior signals order the results. A rating repair often lifts rankings without touching metadata.
- **Through rating mass** (correlational): big healthy counts accompany durable top positions — weight and trust accrued per storefront.
- **Not through review text**: there is no reliable evidence the App Store indexes review words into an app's search matching. Reviews persuade humans, not the index.

## The mechanics you control

- **The reset.** On any version release you may reset the summary rating — and it resets everywhere at once, not per storefront. It trades accumulated rating mass for a clean slate: usually right after fixing the thing users hated, wrong as routine hygiene.
- **The prompt.** The in-app rating sheet (`SKStoreReviewController`) is capped by Apple at three displays per user per 365 days, and Apple decides the moment. Spend the three asks on moments of demonstrated success.
- **Responses.** Developer replies are visible on the listing and can convert a 1-star into an edit; they carry no known direct ranking weight.

## Reviews as keyword research

Review text is users describing your app in their own vocabulary — the exact vocabulary they search with. Mining recent reviews for recurring nouns and verbs routinely surfaces [keyword-field](https://openaso.ai/reference/keyword-field.md) candidates no suggestion tool proposes, which is why the openaso metadata playbook feeds reviews into the audit step.

```json
{
  "storefronts": [
    { "country": "DE", "count": 12480, "averageRating": 3.9,
      "distribution": { "1": 1410, "2": 820, "3": 1105, "4": 2530, "5": 6615 },
      "reviews": [
        { "rating": 2, "title": "Widget fehlt",
          "text": "Ohne Homescreen-Widget bringt mir der Tracker nichts…",
          "version": "4.2.0" }
      ] }
  ]
}
```

One excerpt, three decisions: a market-specific rating problem (3.9 in DE), the word users rank you on (*widget*), and where the next localization's keyword budget should go.

## FAQ

### Does review text affect what my app ranks for?

No reliable evidence supports review-text indexing on the App Store. Ratings influence rankings through conversion and rating mass; the text's search value is as research — it tells you what to put in the fields that are indexed.

### Should I reset my rating with the next release?

Only if the release actually fixes what drove the bad ratings, and the current average is costing conversion. The reset applies in every storefront at once and forfeits rating mass — a 4.6 with 80,000 ratings usually beats a 5.0 with 40.

### Why is my average different in every country?

Because it is computed per storefront from that country's users. Different markets rate differently and experience different bugs (payments, localization, network) — which is also why rating repair is a per-market project.

### Do developer responses to reviews improve ASO?

Not directly — no ranking weight is evidenced. They work on the humans: responses visibly handled raise conversion, and a resolved complaint often becomes an edited, better rating, which does feed the loop.

## Terms

- **summary rating** — The star average and count shown on a listing, computed per storefront from that territory's ratings.
- **rating reset** — The optional, all-storefronts-at-once clearing of the summary rating a developer may trigger with a version release.
- **rating mass** — The accumulated volume of healthy ratings in a storefront — the correlational weight behind durable top rankings.