App Review Radar — Apple App Store & Podcast Review Monitor

Watch your app (and your competitors) on the Apple App Store and Apple Podcasts — and know exactly what changed since last time.

App Review Radar is an unattended review-intelligence monitor, not a one-off review dumper. Point it at one or many App Store app IDs (and/or Apple Podcasts show IDs), add a few competitor IDs, and each run returns the newest customer reviews enriched with version correlation, keyword-based complaint themes, rating deltas, and a "what changed since last run" digest.

It runs entirely on Apple's official, zero-auth public APIs (iTunes RSS Customer Reviews, iTunes Lookup, iTunes Search). No headless browser, no residential proxies, no login walls, no Cloudflare/reCAPTCHA. That means it keeps working unattended — and because it's pure HTTP, runs are cheap.

---

What it does

For every app/show × country you monitor, each run gives you:

1. Latest customer reviews — author, star rating, title, content, the exact app version each review was left on, vote count/sum, and timestamp.
2. Per-version breakdown — average rating + review count + top complaint themes grouped by the version reviews were left on, so you can pin complaints to a release.
3. Theme breakdown — crash, login, price, ads, bug, performance, feature requests (fully customizable), counted from real review text with sample snippets.
4. Rating deltas vs last run — movement in Apple's official averageUserRating and userRatingCount, plus version-change detection.
5. Change digest + negative-spike flag — new reviews, spiking complaint themes, and an isNegativeSpike alert when 1-2 star share jumps past your threshold.

State is persisted in the Apify key-value store, so the sinceLastRun delta works automatically across scheduled runs.

---

Inputs

Field	Type	Description
appIds	array	App Store numeric IDs or app URLs to monitor
podcastIds	array	(optional) Apple Podcasts show IDs/URLs — same pipeline
searchTerms	array	(optional) names resolved to IDs via iTunes Search
competitorIds	array	(optional) competitors included for side-by-side comparison
countries	array	2-letter storefront codes (default ["us"])
maxPages	int	1–10 pages per app per country (≈50 reviews/page)
minRating / maxRating	int	(optional) star filter, e.g. only 1–2★ for support triage
sinceLastRun	bool	only emit new reviews/changes since last run (default true)
themeKeywords	object	(optional) custom theme → keyword groups
includeAggregates	bool	also pull rating/version snapshot via iTunes Lookup (default true)
negativeSpikeThreshold	number	fraction jump in 1–2★ share that flags a spike (default 0.15)

---

Outputs

Each dataset item is one app/show × country, with:

Field	Description
appId, trackName, country, kind, role	identity + primary/competitor role
sellerName, genres	publisher + categories
averageUserRating, userRatingCount	Apple's official aggregate
currentVersion, currentVersionReleaseDate	latest release info
reviewsReturned, totalReviewsFetched, newReviewsCount	reported / fetched / new-since-last-run counts
reviews[]	each with reviewId, author, rating, title, content, appVersion, voteCount, voteSum, updatedAt, country
perVersionBreakdown[]	per app version: version, reviewCount, avgRating, topComplaintThemes[]
themeBreakdown[]	theme, count, sampleSnippets[] (from real text)
ratingDelta	averageUserRatingChange, userRatingCountChange, versionChanged, previousVersion vs last run
changeDigest	newReviewsCount, isFirstRun, spikingThemes[], negativeShare
isNegativeSpike	boolean alert when the 1–2★ share jumps past your threshold
scrapedAt	ISO timestamp of the record

A table view (Overview) summarizes app, rating, version, new reviews and spike flags.

Sample output (one dataset item, trimmed)

{
  "appId": "310633997",
  "kind": "software",
  "role": "primary",
  "country": "us",
  "trackName": "WhatsApp Messenger",
  "sellerName": "WhatsApp Inc.",
  "genres": ["Social Networking"],
  "averageUserRating": 4.68,
  "userRatingCount": 12873440,
  "currentVersion": "25.x.x",
  "currentVersionReleaseDate": "2026-06-18T12:00:00Z",
  "reviewsReturned": 37,
  "totalReviewsFetched": 120,
  "newReviewsCount": 37,
  "reviews": [
    {
      "reviewId": "1234567890",
      "author": "someuser",
      "rating": 1,
      "title": "Broken after update",
      "content": "Crashes every time I open a chat now…",
      "appVersion": "25.x.x",
      "voteCount": 14,
      "voteSum": 12,
      "updatedAt": "2026-06-21T08:42:00Z",
      "country": "us"
    }
  ],
  "perVersionBreakdown": [
    { "version": "25.x.x", "reviewCount": 19, "avgRating": 2.8, "topComplaintThemes": [{ "theme": "crash", "count": 11 }, { "theme": "login", "count": 6 }] }
  ],
  "themeBreakdown": [
    { "theme": "crash", "count": 11, "sampleSnippets": ["app keeps crashing right after the latest update…"] },
    { "theme": "login", "count": 6, "sampleSnippets": ["can't verify my number since updating…"] }
  ],
  "ratingDelta": { "averageUserRatingChange": -0.01, "userRatingCountChange": 2231, "versionChanged": false, "previousVersion": "25.x.x" },
  "isNegativeSpike": false,
  "changeDigest": { "newReviewsCount": 37, "isFirstRun": false, "spikingThemes": [{ "theme": "crash", "was": 4, "now": 11 }], "negativeShare": 0.32 },
  "scrapedAt": "2026-06-21T09:00:00Z"
}

---

Example use cases

1. Indie dev — release regression triage. Schedule daily with appIds: ["YOUR_APP_ID"], minRating: 1, maxRating: 2. Every morning you get only the new 1–2★ reviews, grouped by the version they landed on — instantly see if your latest release spiked "crash" or "login" complaints.

2. ASO / app-marketing agency — competitive watch. appIds: ["CLIENT_ID"], competitorIds: ["RIVAL_1","RIVAL_2","RIVAL_3"], weekly cadence, countries: ["us","gb","de"]. Side-by-side rating deltas and theme breakdowns across storefronts for your weekly client report.

3. Podcast network — audience feedback monitor. podcastIds: ["SHOW_ID_A","SHOW_ID_B"], weekly. Track complaint themes and feature requests per show, with a digest of what's newly spiking.

---

Example input

{
  "appIds": ["389801252", "https://apps.apple.com/us/app/slack/id618783545"],
  "competitorIds": ["310633997"],
  "countries": ["us"],
  "maxPages": 10,
  "sinceLastRun": true,
  "includeAggregates": true
}

---

Pricing

App Review Radar is pay-per-result — you're billed per dataset item produced (one result = one app/show × country, per run), not for compute minutes. Because it runs on Apple's free zero-auth APIs with no proxies or browser, runs are fast and lightweight (see the Pricing tab for the current rate).

That makes your cost predictable up front:

What you monitor	Results / run	Daily for a month
1 app, 1 country	1	~30 results
5 apps, 1 country	5	~150 results
5 apps + 3 competitors, 3 countries	24	~720 results

It scales linearly — add an app, a competitor, or a storefront and you add exactly one result each. The only other cost is a few seconds of Apify platform compute per run, which is minimal by design.

---

Why it's reliable & cheap

• Apple official APIs, zero auth — no Cloudflare, no reCAPTCHA, no fingerprinting. The same infrastructure that powers apps.apple.com.
• Pure HTTP GETs — no headless browser, no proxies, minimal compute → low cost per run, high margin on pay-per-result.
• Built-in request throttle (min 1.2s between calls) + exponential backoff on HTTP 429/503 to stay within Apple's per-IP limits.
• No mock data, ever — every field is derived from the live JSON for the IDs you supply.

---

Limitations

Please read these before scheduling — they reflect exactly what the code does and what Apple's public APIs expose.

• Review count / pagination cap. Reviews come from Apple's iTunes RSS Customer Reviews feed, which exposes at most 10 pages of ~50 newest reviews (≈500 reviews) per app, per country. maxPages (1–10) caps this further. There is no way to page beyond that limit through these endpoints.
• Newest-only, no historical guarantee. The RSS feed returns only the most recent reviews, sorted most-recent-first. Older reviews roll off and cannot be back-filled. This actor is a forward-looking monitor, not a full historical review exporter.
• Per-store rate limits. Apple applies per-IP rate limits (~20 req/min in practice). The actor enforces a minimum 1.2s gap between requests plus exponential backoff on HTTP 429/503, so large runs (many apps × many countries) take proportionally longer rather than failing.
• Country / language scoping. Reviews and aggregates are per storefront. You only get data for the countries you list, and each storefront returns reviews in that region's language. averageUserRating / userRatingCount come from iTunes Lookup for the requested country and reflect Apple's own (rounded) aggregate.
• Theme tagging is keyword-based, not ML/sentiment. themeBreakdown, perVersionBreakdown.topComplaintThemes, and isNegativeSpike are computed by matching your (or the default) keyword groups against real review text and by 1–2★ share — there is no machine-learning sentiment model or LLM involved.
• Static HTTP fetch bounds. Pure HTTP GETs against Apple's public JSON endpoints — no headless browser. If an ID is invalid, region-restricted, or exposes no reviews for a storefront, that target still emits one dataset record with reviews: [] and null aggregates (the run does not fail).
• Delta state depends on the key-value store. sinceLastRun, ratingDelta, changeDigest, and isNegativeSpike compare against state saved in the actor's default key-value store (RUN_STATE). The first run (or any run after the store is reset) is treated as a full snapshot with changeDigest.isFirstRun: true and no rating delta.
• Podcasts coverage. Apple Podcasts shows are monitored through the same RSS review pipeline and only return data where Apple exposes customer reviews for that show/storefront.
• No valid targets. If you supply no resolvable appIds / podcastIds / competitorIds / searchTerms, the run does not crash — it emits a single marker record (status: "NO_TARGETS") and exits cleanly.

---

FAQ

Do I need an Apple developer account or any API key? No. It uses Apple's public, zero-auth endpoints — just supply the app/show IDs and run.

Will Apple block or rate-limit it? It uses the same official APIs that power apps.apple.com, with a built-in request throttle (minimum 1.2s between calls) and exponential backoff on HTTP 429/503, so it stays within Apple's per-IP limits and keeps running unattended. Very large multi-app / multi-country runs simply take longer as a result.

How do I find an app's ID? Paste the full App Store URL into appIds — the actor extracts the numeric ID for you (e.g. https://apps.apple.com/us/app/slack/id618783545). Or put a name in searchTerms and it resolves the ID via iTunes Search.

Does it really track Apple Podcasts too? Yes — put show IDs/URLs in podcastIds; they run through the same review + theme + delta pipeline as apps.

How is "what changed since last run" calculated? State is persisted in the Apify key-value store between runs, so each scheduled run emits only new reviews and flags spiking complaint themes via changeDigest and isNegativeSpike — no duplicate noise.

Can I watch multiple countries at once? Yes — list storefront codes in countries (e.g. ["us","gb","de"]); each app × country is a separate result with its own reviews and deltas.

What's the best way to run it? On an Apify schedule — daily for release-regression triage, weekly for competitive reports — with sinceLastRun: true, and let the digest do the watching for you.

---

Tip: run it on a daily or weekly schedule and let the sinceLastRun digest do the watching for you.