Google Play Reviews Scraper

Collect Google Play Store reviews and ratings for any app, across country storefronts. Give it an app name or a Google Play link and it returns a clean, one-row-per-review dataset: the star rating, review body, author, date, the app version reviewed, thumbs-up count, any developer reply, and the storefront country. Reviews are storefront-specific, so you can sweep many countries to gather the full picture for a single app.

Why this scraper

• Reviews are the primary output: rating, body, author, date, app version, thumbs-up, developer reply, country, and language on every row.
• Two ways in: search by app name, or paste one or more Google Play app links (a bare package id like com.spotify.music works too).
• Multi-country sweep: pass a list of countries, or use "all" for a broad built-in set of major markets.
• Sort by newest, most helpful, or by rating.
• Filter by minimum and maximum star rating.
• Choose the review language / locale.
• Optional app-metadata enrichment: developer, average score, ratings count, installs, category, icon, price, and description attached to each review.
• 20+ output fields, more than typical alternatives, at a predictable per-result price.

Data you get

Sample shape, values are illustrative placeholders, not from a live review.

Field	Example
reviewId	00000000-0000-0000-0000-000000000000
appId	com.example.app
appName	Sample App
country	us
language	en
rating	5
body	Full review text appears here.
author	Reviewer Name
authorId	100000000000000000000
authorImage	https://play-lh.googleusercontent.com/…
thumbsUp	12
appVersion	1.0.0
reviewDate	2026-01-01T00:00:00+00:00
reviewTimestamp	1780000000
reviewUrl	https://play.google.com/store/apps/details?id=com.example.app&reviewId=…
developerReply	Thanks for the feedback!
developerReplyDate	2026-01-02T00:00:00+00:00

With enrichment enabled, each review also carries: appDeveloper, appDeveloperEmail, appDeveloperWebsite, appScore, appRatingsCount, appInstalls, appCategory, appContentRating, appPrice, appCurrency, appIconUrl, appUpdated, appDescription, appStoreUrl.

How to use

Search by app name:

{
  "mode": "search",
  "queries": ["Instagram", "Spotify"],
  "appsPerQuery": 1,
  "countries": ["us"],
  "sortBy": "newest",
  "maxItems": 100
}

Sweep several countries for one app via its link:

{
  "mode": "url",
  "urls": ["https://play.google.com/store/apps/details?id=com.instagram.android"],
  "countries": ["us", "gb", "de", "jp"],
  "maxItems": 500
}

Only high-rated reviews, with app details attached:

{
  "mode": "url",
  "urls": ["https://play.google.com/store/apps/details?id=com.instagram.android"],
  "countries": ["us"],
  "minRating": 4,
  "fetchDetails": true,
  "maxItems": 200
}

Multiple apps at once (bare package ids work too):

{
  "mode": "url",
  "urls": [
    "com.instagram.android",
    "com.spotify.music"
  ],
  "countries": ["us"],
  "maxItems": 200
}

Input parameters

Parameter	Type	Default	Description
mode	string	search	"search" (by app name) or "url" (paste app links).
queries	array	["Instagram"]	App names or keywords (search mode).
appsPerQuery	integer	1	Top matching apps to take per search term.
urls	array	(example)	Google Play app links or package ids (url mode).
countries	array	["us"]	Country codes to collect from. Empty uses the link's country or us. "all" sweeps major markets.
language	string	en	Review language / locale code.
sortBy	string	newest	"newest", "mostHelpful", or "rating".
minRating	integer	(none)	Keep reviews at or above this star rating.
maxRating	integer	(none)	Keep reviews at or below this star rating.
fetchDetails	boolean	false	Attach app metadata to each review.
maxItems	integer	20	Total review cap. 0 means no limit.
maxPages	integer	(none)	Optional safety bound on review pages read per country (about 40 reviews per page). Leave empty to walk every page Google Play serves — the run still stops at Max reviews.
resumeFromRunId	string	(none)	Run id (or dataset id) of a previous run of this Actor. Reviews it already collected are skipped, so this run only fetches new ones.
incrementalMode	boolean	false	Recurring monitoring: remembers what a previous run of the SAME search saw and marks each review NEW / UPDATED / UNCHANGED / REAPPEARED / EXPIRED.
stateKey	string	(none, auto)	Name this monitoring campaign explicitly. Leave empty to derive one automatically from mode, queries/urls, countries, language, sortBy, minRating/maxRating, and fetchDetails (maxItems/maxPages and delivery options like mcpConnectors do NOT affect it).
emitUnchanged	boolean	false	Also return (and bill for) reviews unchanged since the last run.
emitExpired	boolean	false	Also return (and bill for) a tombstone row for reviews the previous run saw that this run no longer finds. Only reported after a run that fully re-walked every tracked app/country stream — never after a capped or resumed run.
proxy	object	Apify proxy	Connection settings.

Incremental mode (recurring monitoring)

Turn on incrementalMode and schedule the Actor to run on the same app/country/filter set (or set stateKey yourself for an explicit campaign name). Each run compares against what the previous run of the SAME campaign saw and adds changeType (NEW, UPDATED, UNCHANGED, REAPPEARED, or EXPIRED), changedFields, firstSeenAt, and lastSeenAt to every row. By default only NEW/UPDATED/REAPPEARED rows are returned — turn on emitUnchanged/emitExpired to also see (and pay for) the rest.

What counts as a change: rating, review text, author, thumbsUp, app version, and developer replies are all compared — every field a review actually carries is real data a monitoring run exists to report. One thing is deliberately excluded from change detection (though still shown on every row):

• App-metadata fields attached when fetchDetails is on (appScore, appRatingsCount, appInstalls, appDeveloper, ...) — these describe the APP, not the individual review, and change on their own as other users interact with the app. Excluding them stops one unrelated review's helpful-vote or the app's aggregate rating from flipping every other review to UPDATED in lockstep.

thumbsUp was suspected of ticking up independently of the review (like a vote counter) and was excluded in an earlier draft on that guess. It was measured directly: two live fetches of the same popular app's top-40 most-helpful reviews (Gmail, com.google.android.gm), 1810s (30.2 minutes) apart, diffed field-by-field. Result: zero drift — every one of the 40 thumbsUp values (ranging 2–3318, including reviews as recent as days old) was byte-identical across both passes, as was every other field. The guess wasn't supported by data, so thumbsUp is now treated as ordinary review data and included in change detection. This was tested at a 30-minute window, not the actor's real incremental cadence (typically daily/weekly) — if a future run surfaces thumbsUp-only UPDATED noise at that longer cadence, revisit _VOLATILE_FIELDS in src/incremental.py.

Combining resumeFromRunId with incrementalMode is only for bootstrapping a brand-new campaign from an earlier non-incremental run's dataset; once a campaign has its own tracked state, combine them again and the run fails fast (remove resumeFromRunId or use a different stateKey). emitExpired reviews are only ever reported after a complete, uncapped, non-resumed scan — a partial run can't tell "deleted" apart from "not reached yet".

Output example

Sample shape, values are illustrative placeholders.

{
  "reviewId": "00000000-0000-0000-0000-000000000000",
  "appId": "com.example.app",
  "appName": "Sample App",
  "country": "us",
  "language": "en",
  "rating": 5,
  "body": "Full review text appears here.",
  "author": "Reviewer Name",
  "authorId": "100000000000000000000",
  "authorImage": "https://play-lh.googleusercontent.com/…",
  "thumbsUp": 12,
  "appVersion": "1.0.0",
  "reviewDate": "2026-01-01T00:00:00+00:00",
  "reviewTimestamp": 1780000000,
  "reviewUrl": "https://play.google.com/store/apps/details?id=com.example.app&reviewId=…",
  "developerReply": null,
  "developerReplyDate": null
}

Send results into your apps (MCP connectors)

You can optionally pipe results into the apps you already use through Model Context Protocol (MCP) connectors. Authorize a connector under Apify, Settings, API & Integrations, then select it in the input. For Notion, set a parent page URL and each review is written as a page. Other connectors receive a best-effort write.

The connector receives a condensed, human-readable summary of each review (a heading plus the key fields and body text), not the full JSON. The complete record always stays in the Apify dataset. Leave the connector field empty to skip this step; it never changes the dataset output.

Plan requirement

Runs on any Apify plan. For very large multi-country sweeps, a proxy with more exit rotation can help.