Google Play & App Store Reviews | AppReview Radar

Turn public app reviews into a feedback shortlist for your next product decision. Read Android and iOS reviews in one run, filter by stars and words, and export CSV plus a Markdown report with source evidence.

Built for indie app founders, product teams, ASO specialists, and agencies comparing competitor feedback. No store login, LLM key, paid upstream API, or proxy setup is required.

Try a small run

Choose the ChatGPT demo preset, or paste this into JSON input:

{
  "appPreset": "chatgpt",
  "country": "gb",
  "language": "en",
  "maxReviews": 20,
  "maxReviewsPerApp": 10,
  "maxPagesPerApp": 1,
  "maxRating": 3
}

Click Start, then open the dataset and the Output tab. It links results.csv and feedback-report.md. The run requests up to 20 saved reviews across Android and iOS; it may return fewer or zero because filters apply to a bounded source window. Check DIAGNOSTICS for source failures or page limits. The separate OUTPUT record is the complete JSON report.

Step-by-step guide · Hướng dẫn tiếng Việt

What you get

• One schema for two stores: review ID, store, app ID, requested locale, rating, review text, date/version when supplied, public reply where exposed, source URL, and matched topic evidence.
• Filters before billing: rating band, ISO date cutoff, included words, excluded words, topic matches, and an explicit feature-request phrase option.
• Transparent topic tags: stability, performance, billing, login, ads, privacy, and feature-request mentions using English/Vietnamese literal rules.
• Ready-to-use exports: dataset/CSV, Markdown feedback report, and TOPICS with per-topic counts and evidence.
• Repeat-review deduplication: pass KNOWN_REVIEW_IDS into your next input to skip saved reviews before output and charging.
• Review edit monitoring: supply SNAPSHOT and changedOnly to detect exposed same-ID content changes while skipping unchanged review events.
• Release/support filters: match source app versions or visible Google Play developer replies. Apple RSS reply status remains unavailable.

Topic matches are rules, not AI analysis or a verified bug diagnosis. Read the original evidence: a request phrase does not prove intent or demand. The report uses saved reviews only; it is not representative polling or the app's overall rating.

Useful recipes

Compare complaints about two apps: set your Android package IDs or iOS numeric IDs, maxRating: 2, and the same country/page settings. Group exported rows by app; compare the original text and available versions rather than treating different sample sizes as market share.

Inspect login problems: choose topicFilters: ["login"], optionally add reviewedAfter, and download the report. Topic keywords match literal text and can miss paraphrases or match a negated phrase.

Find feature-request mentions: set featureRequestsOnly: true. Results must contain a supported request phrase; they are candidates for human review, not verified feature requests.

Skip reviews already exported: copy the previous KNOWN_REVIEW_IDS array into knownReviewIds. Your workflow must pass that state explicitly; attaching a schedule does not do it automatically. This mode skips same-key edits too.

Track review edits with a snapshot

Use previousSnapshot plus changedOnly: true, and leave knownReviewIds empty. The first run produces SNAPSHOT; copy it to the next run's previousSnapshot with the same country/language. The Actor compares exposed title/text, rating, app version and developer reply. It saves NEW to that baseline or UPDATED reviews; matching UNCHANGED reviews are skipped before output and billing.

A NEW key may be an old review not yet in your supplied snapshot. Source windows are partial; missing reviews never become deletion claims. Only saved rows advance the baseline. Snapshots retain at most 10,000 hashes and 2 MiB; oldest entries can be evicted with snapshotTruncated: true. Same-key reviews in knownReviewIds still take priority and are skipped without edit checking. Default changedOnly: false saves/bills matching unchanged rows.

Snapshot workflow and caveats · Hướng dẫn theo dõi review đã sửa

Main inputs

Input	Purpose
appPreset	none or Android + iOS chatgpt demo
googlePlayApps, appStoreApps	App IDs or standard store URLs; combined maximum 10 apps
country, language	Requested store country and Google Play language hint; not reviewer residence/language verification
maxReviews	Global saved cap; default 100, maximum 2,000
maxReviewsPerApp	Saved cap per app; default 100, maximum 1,000
maxPagesPerApp	Source-page cap per app; default 3, maximum 10
minRating, maxRating	Inclusive 1–5-star range
includeKeywords, excludeKeywords	Case-insensitive substring lists; any excluded match removes the review
topicFilters	Any selected topic must match, in addition to other filters
featureRequestsOnly	Require an explicit supported request phrase
reviewedAfter	ISO cutoff; unknown review dates are excluded
knownReviewIds	Saved keys to skip, up to 10,000
previousSnapshot, changedOnly	Caller-supplied review fingerprints and optional suppression of unchanged rows
appVersionKeywords	Source-version substring filter; missing versions excluded when set
replyFilter	all, with-reply, or without-visible-reply; reply-specific filters exclude Apple RSS

Preset apps and supplied apps are combined and deduplicated. App names are not searched automatically; use an ID or a store listing URL.

Pricing

Launch rate: $0.50 per 1,000 saved matching reviews, one review-saved event per saved row. Confirm the live rate in Pricing. No Actor start fee and no extra review events for report/topic evidence.

At that rate, 20 saved reviews generate $0.01 in review events. Source failures, filtered reviews, duplicates, and supplied known IDs do not generate review events. Platform usage follows Apify's applicable plan and pricing model. Set a maximum total charge for the run.

Important limits

• Public web/feed data only. This is not the official authenticated Google Play Developer or App Store Connect API.
• Source windows are incomplete. The reader scans at most 10 pages per app, and available pages may contain fewer reviews or be empty. Google Play is bounded by your page cap and what the source returns. No full-history guarantee.
• Apple RSS does not expose developer replies. Missing source values stay null; labels are not verified app metadata.
• The requested country is not verified reviewer residence. Language is a Google Play hint, not a guaranteed review-language filter.
• Failures are explained in DIAGNOSTICS; they are not silently treated as an app having no reviews.
• Known-key history rolls at 10,000 saved keys and must be passed explicitly. Same-key edited reviews are skipped.
• Snapshot edit tracking is separate from known-key skipping and also needs explicit state handoff. Only exposed bounded fields are compared; changes outside fetched windows or beyond the text limit cannot be recovered.
• No login, private data access, CAPTCHA bypass, proxy rotation, or automated review posting.

Source policy and field caveats · Open-source repository

This independent tool is not affiliated with OpenAI, Google, or Apple. The ChatGPT preset is a public-app example, not an endorsement.