Google Maps Reviews Scraper

Google Maps Reviews Scraper

Point it at a Google Maps place — a URL, a place ID, or just the business name — and get every review back as a row: star rating, the full review text, when it was posted, who wrote it (with their own review and photo counts), the owner's reply, and how many photos the review carries.

Sorted newest first. No Google account, no cookies, no API key, nothing to log into.

Price

$0.35 per 1,000 reviews, plus a $0.0000125 start fee per GB of run memory (this Actor runs at 1 GB, so that is $0.0000125 per run — a hundredth of a cent).

That is the cheapest Google Maps reviews scraper on the market. Here is the comparison, taken from Apify's own store API on 2026-08-14, counting only the per-result event and ignoring the decoy apify-default-dataset-item charges rivals attach beside their real price:

The most-installed rival, compass, does not charge per review at all — it bills you Apify platform usage, so what a run costs depends on how long its browser sits there. This Actor charges a flat $0.00035 per review and nothing else, so a 5,000-review job costs $1.75 whether it takes one minute or ten.

What is free. The uncharged sample row you get from an empty input is free. Every diagnostic row (ok: false, with an errorCode) is free — blocks, empty places, bad URLs. Reviews dropped by your keyword or date filter are free, because they are never pushed. You only pay for a review row that actually landed in the dataset.

Input

Everything is optional. Give it at least one of startUrls, placeIds or searchStrings.

{

  "searchStrings": ["Katz's Delicatessen New York"],

  "maxReviews": 100,

  "maxPlaces": 1,

  "reviewsSort": "newest",

  "language": "en"

}

The field names match compass/google-maps-reviews-scraper where they overlap (startUrls, placeIds, maxReviews, reviewsSort, reviewsStartDate, reviewsFilterString, language, personalData), so an existing input JSON can be pasted straight in.

Output

A real row from a real run (run mg8e1KaEV80M11Pu5, 12 newest reviews of Katz's Delicatessen):

{

  "ok": true,

  "charged": true,

  "placeName": "Katz's Delicatessen",

  "placeUrl": "https://www.google.com/maps/place/Katz's+Delicatessen/@40.722233,-73.987429,17z/data=!3m1!4b1!4m6!3m5!1s0x89c2598f7ff4aa09:0x313547e757cb8cea!8m2!3d40.722233!4d-73.987429!16zL20vMDN0eF9o",

  "placeTotalScore": 4.5,

  "placeReviewsCount": 39526,

  "reviewsSort": "newest",

  "sortApplied": true,

  "reviewId": "Ci9DQUlRQUNvZENodHljRjlvT2treGRHUnhjMmxhWWs0MGMzQTRha3RYZW5keldGRRAB",

  "reviewerName": "Vanesa Herrera",

  "reviewerUrl": "https://www.google.com/maps/contrib/116379031429699768301/reviews",

  "reviewerId": "116379031429699768301",

  "reviewerNumberOfReviews": 5,

  "reviewerNumberOfPhotos": null,

  "isLocalGuide": false,

  "rating": 2,

  "relativeDate": "3 hours ago",

  "publishedAtDate": "2026-08-15T01:00:12.328Z",

  "text": "If you're in a group larger than six and want to eat at the table where the servers serve you, don't mention it because it will take a long time to get your table... In the end, the food was okay, but nothing extraordinary.",

  "ownerReply": null,

  "ownerReplyRelativeDate": null,

  "reviewPhotosCount": 0,

  "reviewPhotoUrls": []

}

Coverage on that run: 12 of 12 rows carried placeName, placeTotalScore, placeReviewsCount, rating, relativeDate, reviewerName, reviewerUrl, reviewerNumberOfReviews and text, and 0 of 12 texts were left truncated with the "... More" preview. On a larger run of the same place (100 reviews, run M5qRtD9Rzymcff0pt) 43 of the 100 rows also carried an ownerReply — the rest simply have no reply — and the longest review body came back at over 1,900 characters.

Rows arrive newest first when sortApplied is true; that run's first rows were "3 hours ago", "an hour ago", "21 minutes ago".

Honest limits

• publishedAtDate is derived, not exact. Google only publishes a relative label

("a month ago"), so this Actor converts that label to a timestamp at scrape time. relativeDate is the raw label Google gave; use it when you need the ground truth. reviewsStartDate filtering works off the derived timestamp, so it is accurate to roughly the granularity Google shows.

• sortApplied tells you the truth about sorting. Google's sort control is a menu that

occasionally does not open in a headless session. When that happens the run still returns reviews, but in Google's default "most relevant" order, and sortApplied is false on every row so you can see it rather than guess.

• Google soft-throttles by IP. Instead of an error it sometimes serves a cut-down place page

with no Reviews tab at all — the page footer says "You're seeing a limited view of Google Maps". This Actor detects that page, rotates to a different exit address and retries up to three times. If all three fail you get an uncharged BLOCKED diagnostic row, not a silent empty dataset.

• placeReviewsCount is the number Google prints on the reviews pane, which is not always the

same as the headline count on the overview card — Google counts star-only ratings differently in the two places. Treat it as Google's own figure for the review list you are paging, not as an audited total.

• Ceiling is 2000 reviews per place. Google's own pane stops serving more long before that on

most places; very large places will return what the pane gives, then stop.

• One place per search phrase. A searchStrings entry resolves to Google's top match. If you

need a specific branch, pass its place URL or place ID.

• No reply threads, no review edits history, no reviewer email. Google does not publish them.
• Very short reviews can come back with an empty text. Google Maps lets people leave a star

rating with no words. Those rows still carry the rating, reviewer and date, and text is empty rather than padded with the card's chrome.

• Runs use a real browser. Google's reviews are not served by any public JSON endpoint — its

internal listugcposts RPC returns HTTP 403 to every anonymous request (verified across four TLS fingerprints, four Google country hosts, with and without cookies and session tokens). A short run therefore has a fixed browser cost; runs of roughly 25 reviews or more are where the per-review price does its job.

FAQ

Do I need a Google account or an API key? No. Nothing is logged into. It reads the public Google Maps place page.

How much does 10,000 reviews cost? $3.50, plus the start fee of one hundredth of a cent per run.

Can I get only reviews from the last month? Yes — set reviewsStartDate to 30 days. Older reviews are dropped before they are pushed, so you are not charged for them.

Can I scrape several places in one run? Yes. Put multiple entries in startUrls, placeIds or searchStrings. maxReviews applies per place; maxPlaces caps how many are visited.

What happens if a place has no reviews? You get one uncharged diagnostic row with errorCode: "NO_RESULTS" and the run finishes as succeeded. You are not billed for it.

What happens if I run it with empty input? You get exactly one labelled sample row (_sample: true, charged: false) showing the output shape, and the run succeeds. Nothing is charged except the start fee.

Is this really the cheapest Google Maps reviews scraper? At $0.35 per 1,000 reviews, yes — every other Google Maps reviews scraper on Apify with a real per-result price charges more (see the table above), and the free-tier ones bill you platform compute instead, which on a browser-driven review crawl runs higher than $0.35 per 1,000.

Can I send the results to Google Sheets / a webhook? Results land in a standard Apify dataset, so every Apify integration works: API, webhooks, Zapier, Make, or a direct CSV/JSON/Excel export from the run page.