App Store Reviews Scraper — Ratings, Versions & Countries

What are people saying about this app right now, in this country?

Every review Apple serves for an app, newest first or most helpful first, one row per review: stars, title, text, the version it was written against, author and date. Read the same app from several countries' stores in one run. Or flip the switch and get one summary row per app and country — Apple's own average rating and rating count, the version on sale, how the recent reviews split across one to five stars and how fast they are coming in.

No API key. No login. No proxies.

---

Two modes

Reviews — one row per review. The default. Give it an app link or ID and a list of countries, and get the reviews as a table you can filter by stars, version or date.

Summary — one row per app and country. Give it twenty apps — yours, your competitors' — and get twenty rows you can sort: Apple rating and count, average of the recent reviews, the one-to-five star split, reviews per day, days since the last update.

---

What you get

Reviews mode

The review — rating (1–5), title, body, version (the app version the reviewer had), author, helpfulVotes and helpfulVoteCount, reviewId

When — reviewedAt, daysSinceReview

Which app, which store — appId, appName, developer, country, appUrl

Summary mode

Apple's figures — averageRating and ratingCount as shown on the store page (all ratings, not only written reviews), currentVersion, currentVersionReleaseDate, daysSinceUpdate, firstReleaseDate, price, currency, genre, contentRating, minimumOsVersion, sizeMB, bundleId, releaseNotes

Computed from the newest reviews — reviewsRead, recentAverageRating, stars5 … stars1, reviewsLast30d, reviewsPerDay30d, latestReviewAt, oldestReviewReadAt

How complete — status says whether the figures cover the window you care about (see Honest limits)

---

Typical uses

• Watch your own app after a release. Reviews mode, daysBack: 7, on a schedule. Filter to the new version number and read what changed.
• Compare the competition. Summary mode on the apps in your category: who is rated best, who is slipping in the recent reviews, who ships updates and who has gone quiet.
• See the same app through different markets. countries: ["us", "gb", "de", "jp", "kr"] — the star split and the complaints differ by store more often than teams expect.
• Feed support and product. Sort the low-star rows by helpfulVotes and you have the complaints other users agree with.

---

Input

Field	What it does
apps	App Store links or numeric IDs. https://apps.apple.com/us/app/whatsapp-messenger/id310633997 or 310633997.
countries	Two-letter storefront codes (us, gb, de, jp, kr…). Each app is read from every store listed. Leave empty to use the country in each link, or the US store for bare IDs.
mode	reviews (default) or summary.
sortBy	mostrecent (default) or mosthelpful — the two orders Apple offers.
daysBack	0 for no date filter; 7 or 30 for a tracking window.
maxReviewsPerApp	Cap per app per country, default 100. Apple's ceiling is 500 (see below).

---

What it costs

$1.00 per 1,000 reviews in reviews mode, $10.00 per 1,000 app summaries in summary mode, plus $0.005 per run. Apify's Starter, Scale and Business plans get 10%, 20% and 30% off.

You pay for rows delivered. An app with no reviews in a store, an ID that does not exist, or a date window that matches nothing costs nothing. Reading the same app from five countries is five sets of rows.

For scale: the full 500 reviews Apple serves for one app in one country is $0.50; a summary of 50 competitor apps is $0.50.

---

How it works

Apple publishes a JSON feed of customer reviews for every app and storefront, fifty per page, and a lookup endpoint with the app's rating, version and metadata. This actor reads those two endpoints and nothing else — no browser, no HTML scraping of store pages, no proxies. That is why it is fast (a full 500-review read takes a few seconds), cheap, and does not break when Apple restyles the App Store.

Reviews are walked page by page until the cap, the date window or the end of the feed is reached. Each review's updated timestamp is normalised to UTC.

---

Honest limits

• Apple serves at most 500 reviews per app per country — ten pages of fifty, then nothing. Measured 11 September 2026. The feed also has holes: for big apps a page in the middle comes back empty while later pages are full (WhatsApp US page 3, Instagram US page 2), so this actor reads all ten pages rather than stopping at the first empty one. For a big app that is a few days of reviews; for a small one it is the whole history. status tells you which case you are in, and a scheduled run with daysBack is how you build a longer record.
• The 30-day figures in summary mode are only reported when they are complete. If the reviews read stop short of 30 days back — because of your cap or Apple's ceiling — reviewsLast30d and reviewsPerDay30d are left empty rather than reported as a floor, and status says so.
• averageRating and ratingCount are Apple's figures for all ratings, including the star-only ones with no text. recentAverageRating is computed from the written reviews this actor read. The two measure different things and usually differ.
• Apple no longer publishes a separate rating for the current version. The lookup endpoint returns the same number for "all versions" and "current version", so only one is reported here.
• Developer replies are not in Apple's feed, so they are not here either.
• Review order is Apple's. "Most recent" is by the review's last-updated time, so an edited old review can appear near the top.
• A storefront code that Apple does not recognise returns nothing, reported per app with the reason.

---

Scheduling

Reviews keep coming. Schedule reviews mode with daysBack: 7 on your apps and each run adds the week's reviews; schedule summary mode weekly on your category and you have a dated record of ratings, star splits and update cadence over time.

---

Legal

This actor reads only data that Apple serves publicly for each app, at addresses it serves to anyone. Reviews are public statements their authors chose to publish under a screen name. It requires no login and bypasses no access control. You remain responsible for how you use the exported data.

---

Questions, or an app that behaves oddly? Open an issue on the actor — issues are read and answered.