DEV Community

CleanFeed Data
CleanFeed Data

Posted on

How to export App Store reviews by country from Apple's official feed

Apple publishes written App Store reviews as a public customer-review feed, one storefront at a time. You can read it with curl. There is no official bulk "all countries" API, so if you want the same app in the US, UK, Germany and Japan, you request each country separately and merge the results.

This is a short guide to that feed, and to a one-run export. At the end I also show an Apify actor I built that loops the storefronts, deduplicates, and can tag sentiment and topics. Disclosure: I work on CleanFeed Data, and that actor is ours. You can follow everything here with curl alone.

What you'll need

  • curl, or any HTTP client.
  • An app's numeric Apple ID (the number after /id in the App Store URL). Duolingo is 570060128; Spotify is 324684580.
  • The two-letter storefront codes you care about, such as us, gb, de, jp.

No Apple developer account and no login. The feed is public.

Your first request

This is Apple's public JSON feed for Duolingo's most recent reviews in the US storefront:

curl -sS "https://itunes.apple.com/us/rss/customerreviews/id=570060128/sortBy=mostRecent/json"
Enter fullscreen mode Exit fullscreen mode

You should see a feed whose entry list includes the app itself and then reviews, each with a title, the review text, a 1–5 star rating, the app version, a date, and the reviewer's nickname.

Swap the country and the id for other storefronts or apps:

# Most recent Duolingo reviews in the UK
curl -sS "https://itunes.apple.com/gb/rss/customerreviews/id=570060128/sortBy=mostRecent/json"

# Most helpful Spotify reviews in the US
curl -sS "https://itunes.apple.com/us/rss/customerreviews/id=324684580/sortBy=mostHelpful/json"
Enter fullscreen mode Exit fullscreen mode

Useful facts about this feed:

  • It is paged, 50 reviews per page, up to 10 pages — so at most 500 written reviews per app, per country, per sort order. Older reviews are not in the public feed.
  • sortBy=mostRecent and sortBy=mostHelpful are different lists. They overlap, so if you fetch both, deduplicate on the review id.
  • It contains written reviews only, not star-only ratings. The app's overall average and rating count live on Apple's iTunes Lookup API instead: https://itunes.apple.com/lookup?id=570060128&country=us.
  • Reviews are per storefront. A US review and a UK review of the same app are different records, even if the text looks similar.

A straightforward loop is: for each app and country, walk pages 1–10 (or stop early when a page comes back short), then optionally do the same for mostHelpful. Space the requests out; Apple's search/lookup guidance is about 20 calls per minute, and the same politeness is reasonable for the review feed.

A real export, with numbers

On 13 Nov 2026 at 9:03 AM ET I ran a capped export of Duolingo and Spotify in the US and UK storefronts, 100 reviews in total, split evenly (25 reviews per app per country, most recent first). The run finished in 16.9 seconds. It made 42 HTTP requests, with no retries and no errors, and wrote 100 review rows.

Apple's feed actually had a full page window in every cell of that grid (500 unique written reviews fetched per app per country). The 25-row cap is what I set, not a shortage of reviews.

These figures are for those 25 most recent written reviews in each storefront, not for the app's overall App Store rating:

App Storefront Reviews Average stars 5★ 4★ 3★ 2★ 1★ Sentiment (pos / neu / neg)
Duolingo: Learn Languages US 25 4.16 17 2 2 1 3 19 / 2 / 4
Duolingo: Learn Languages UK 25 3.56 12 3 2 3 5 15 / 3 / 7
Spotify: Music and Podcasts US 25 3.96 14 4 3 0 4 18 / 3 / 4
Spotify: Music and Podcasts UK 25 4.04 15 2 4 2 2 17 / 4 / 4
All four cells 100 3.93 58 11 11 6 14 69 / 12 / 19

Dates of those 100 most-recent reviews ran from 10 Nov 2026, 2:41 PM ET (oldest Spotify UK row) to 11 Nov 2026, 6:09 AM ET (newest Duolingo US row). Every Duolingo review in the sample was for app version 7.148.0; every Spotify review was for 9.1.98.

That is why country matters: the same two apps, same cap, same sort, produced a 4.16 average in US Duolingo reviews and a 3.56 average in UK Duolingo reviews in this window. A single-storefront export would have missed that.

Exporting a tidy table (optional)

If you would rather not write the paging, country loop and deduplication yourself, the App Store Reviews by Country actor on the Apify Store calls only those official Apple feeds, merges storefronts into one dataset, and exports JSON, CSV or Excel. You can filter by stars, dates or keywords before you are charged, and optionally add sentiment and topic tags (bug-crash, pricing-subscription, ads, and so on). Those tags are a simple heuristic on the review text and star rating, not an official Apple field.

A typical input looks like this (the 13 Nov 2026 run above used the first two apps, us and gb, and maxResults: 100):

{
  "apps": [
    "https://apps.apple.com/us/app/duolingo-language-lessons/id570060128",
    "com.spotify.client"
  ],
  "countries": ["us", "gb"],
  "sortBy": "mostrecent",
  "maxResults": 100,
  "enrich": true
}
Enter fullscreen mode Exit fullscreen mode

apps also accepts a numeric id, a bundle id, or an app name. countries accepts storefront codes, or major / all if you want a wide sweep.

Pricing is pay-per-result, currently $0.005 to start a run and $0.10 per 1,000 reviews ($0.0001 each). The 100-review run above would bill $0.015. The Apify free plan includes $5 of usage each month, which is enough to try this. I built this actor; it is not affiliated with or endorsed by Apple.

Using the data legally

The actor (and the curl commands above) only read data Apple publishes openly: the public customer-review feed (itunes.apple.com/{country}/rss/customerreviews/…) and the documented iTunes Search/Lookup API. They do not log in, scrape apps.apple.com pages, or get around any protection.

Reviews include public nicknames written by users. If you store or process them, you are the data controller. Follow GDPR/CCPA and Apple's terms if you republish any of the text, and drop reviewer identities when you do not need them.

Further reading

Top comments (0)