Google Hotel Reviews Scraper

Give it Google Hotels links or hotel names and it returns every guest review Google shows for each hotel as clean JSON: rating, text in the original language plus Google's translation, date, trip type and party (vacation or business; solo, couple, family, friends), the rooms/service/location scores, the hotel's reply, and the partner-site reviews Google lists next to its own (Tripadvisor, Trip.com, ...). Each hotel also gets a free summary row with its star histogram and Google's topic breakdown. $0.54 per 1,000 reviews; you only pay for the reviews you get back.

It is for hotel and revenue teams who watch their own reviews and their competitive set, reputation and review-analytics tools that need Google's hotel reviews in bulk, and analysts who study what guests say about a city's hotels. It reads Google's own data endpoint (no browser): up to 200 reviews per request, so the 4,761 reviews of a large hotel take under a minute.

Use with AI agents

Returns Google hotel reviews for hotel links or names: one row per review (rating, text, translation, date, trip type, aspect scores, owner reply, source site) plus a free hotel summary row (rating, histogram, topics with positive and negative mentions, partner sites). Agents can call it through the Apify MCP server (https://mcp.apify.com, open source as apify/actors-mcp-server) as arcadium/google-hotel-reviews: the input schema is read from the actor, the result is a dataset of JSON rows.

Input example for "the latest 50 Google reviews of the Ritz Paris":

{
  "hotels": ["Ritz Paris"],
  "maxReviewsPerHotel": 50
}

And for "what do guests complain about at these two hotels, newest first, Google reviews only":

{
  "hotels": ["https://www.google.com/travel/hotels/entity/ChUImcGKtLfWgL9OGgkvbS8wNTZuODEQAQ", "Hotel Adlon Kempinski Berlin"],
  "sort": "lowest",
  "reviewSource": "google",
  "maxReviewsPerHotel": 100
}

What comes back: review rows (charged), one free hotel-summary row per hotel, one free coverage row per hotel (Google's stated total vs delivered, and why reading stopped) and a free error row for a hotel that could not be read. Filter on the rowType field. An error row carries rowType: "error", a code (invalid_input, not_found, no_match, filter_unsupported, blocked or unavailable), a message and the input it is about. Retry on blocked and unavailable later; do not retry the other codes with the same input.

Cost of the calls above: 50 reviews at $0.54 per 1,000 is about $0.03; 200 reviews about $0.11. Each run adds a $0.0005 start fee (512 MB). Paid Apify plans pay less per review.

What you get

Review row (rowType: "review"), the shared review shape of all our review actors:

Field	What it holds
reviewId	GoogleHotels___<Google post id>, the same in every language. Use it to deduplicate between runs.
productName, productUrl, entityId	The hotel, its Google Hotels link and Google's token for it (reuse it in hotels).
ratingOverall	1 to 5. Partner ratings are scaled to 5 (raw value in extras.rawRating, extras.rawScale).
ratings	The scores the guest gave per aspect: rooms, service, location (1 to 5), when given.
text	The review as written, in its original language. null for rating-only reviews.
date	Approximate date YYYY-MM-DD. Google shows "3 months ago"; extras.datePrecision says how exact it is.
vendorResponse	The hotel's reply in its original language (greetings and signatures with names are redacted).
reviewUrl	Partner reviews: the review on Tripadvisor, Trip.com, ... Google reviews have no public link (null).
extras.source, extras.sourceType	Google or the partner site; google or partner.
extras.textTranslated, extras.translatedTo	Google's translation into your language when the review is in another language.
extras.tripType, extras.travellerType	vacation / business; solo / couple / family / friends, when the guest said.
extras.visitMonth, extras.highlights, extras.topicComments	Month of the stay (2026-09); tags the guest picked ("Luxury", "Great view"); short answers per topic ({ topic: "Rooms", text: "Spotless" }).
extras.ownerResponseDate, extras.ownerResponseTranslated	Approximate reply date; Google's translation of the reply.
extras.dateText, extras.edited, extras.photoCount	Google's own date wording, whether the review was edited, number of guest photos (no photo links).

Output example:

{
  "rowType": "review",
  "platform": "googlehotels",
  "productName": "Ritz Paris",
  "entityId": "ChUImcGKtLfWgL9OGgkvbS8wNTZuODEQAQ",
  "reviewId": "GoogleHotels___Ci9DQUlRQUNvZENodHljRjlvT25GSlZXbzJkV1Z0VTNwdmVGTXRVblUxY2s5aE1WRRAB",
  "ratingOverall": 4,
  "ratings": { "rooms": 3, "service": 4, "location": 3 },
  "date": "2026-10-16",
  "text": "이곳은 아름다운스파 수영장 레스토랑 노을을 볼수있는곳 까지있어서 너무아름다워요 고급프랑스느낌",
  "vendorResponse": "Cher [name],\n\nThank you for taking the time to share your wonderful feedback. ...\n\nKind regards,\n\n[name]\n Directrice d'hôtel, Ritz Paris",
  "language": null,
  "extras": {
    "source": "Google",
    "sourceType": "google",
    "dateText": "6 days ago",
    "datePrecision": "day",
    "textTranslated": "This place is so beautiful, with a lovely spa, swimming pool, restaurant, and even a spot to watch the sunset. It has a luxurious French feel.",
    "translatedTo": "en",
    "tripType": "vacation",
    "travellerType": "couple",
    "visitMonth": "2026-10",
    "highlights": ["Luxury", "Romantic"],
    "ownerResponseDate": "2026-10-17",
    "photoCount": 0
  }
}

Hotel summary row (rowType: "hotel-summary", free, one per hotel; switch off with includeHotelSummary: false):

Field	What it holds
rating, reviewCount, ratingHistogram	Google rating, number of Google reviews, share and count per star.
topics	Google's topic breakdown: { name, shortName, mentions, positive, negative, sentiment, polarity } for topics such as Location, Service, Breakfast, Couples, Safety.
partnerSites	Review sites Google lists for the hotel with their own rating and review count: { name, providerId, rating, ratingScale, reviewCount, url }.
hotelClass, address, countryCode, latitude, longitude	Star class and location.

{
  "rowType": "hotel-summary",
  "productName": "Ritz Paris",
  "rating": 4.6,
  "reviewCount": 4649,
  "ratingHistogram": { "5": { "percent": 86, "count": 3931 }, "1": { "percent": 5, "count": 238 } },
  "topics": [
    { "name": "Bar or lounge", "shortName": "Bar", "mentions": 150, "positive": 110, "negative": 22, "sentiment": 0.502, "polarity": 1 }
  ],
  "partnerSites": [
    { "name": "Tripadvisor", "providerId": 100532569, "rating": 4.5, "ratingScale": 5, "reviewCount": 1073, "url": "https://www.tripadvisor.com/Hotel_Review-g187147-d188728-r1078308347" }
  ]
}

Input

{
  "hotels": ["https://www.google.com/travel/hotels/entity/ChUImcGKtLfWgL9OGgkvbS8wNTZuODEQAQ", "Park Hyatt Tokyo"],
  "maxReviewsPerHotel": 500,
  "sort": "newest",
  "reviewSource": "all",
  "language": "en",
  "dateFrom": "6 months"
}

Field	What it does
hotels	Google Hotels / Travel links, entity tokens (Ch...), hotel names with the city, or Google Maps place links (read by name). A name Google cannot match to a hotel gives a free no_match row.
maxReviewsPerHotel	Reviews per hotel, up to 50,000.
sort	newest, most_helpful, highest, lowest (Google's own orders).
reviewSource	all, google (Google's own reviews) or partners (the Tripadvisor, Trip.com, ... reviews Google shows).
topic	Only reviews about one of the hotel's topics, e.g. Location, Breakfast, Couples.
dateFrom	Only reviews from this date on (YYYY-MM-DD or 30 days, 6 months). Approximate, see Limitations.
language, country	Interface language (translations and date wording) and market.
includeHotelSummary	The free summary row per hotel. On by default.
onlyNewReviews, stateKey, markEdited	Monitoring: only reviews not delivered before under the same key; optionally re-deliver edited ones.
includeErrorRows, includeCoverageRows	The free error and coverage rows. On by default.

Input names of other Google review actors are accepted as they are (hotelUrls, startUrls, startUrl, hotelNames, maxReviews, maxItems, reviewsSort, sortBy, reviewsOrigin, reviewsStartDate, languageCode, countryCode), so an existing input can be pasted in.

Use from code

Python:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("arcadium/google-hotel-reviews").call(run_input={
    "hotels": ["Hotel Arts Barcelona", "W Barcelona"],
    "maxReviewsPerHotel": 300,
    "sort": "newest",
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    if row.get("rowType") == "review" and row["ratingOverall"] <= 2:
        print(row["productName"], row["date"], row["extras"].get("textTranslated") or row["text"])

JavaScript:

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: '<YOUR_APIFY_TOKEN>' });
const run = await client.actor('arcadium/google-hotel-reviews').call({
  hotels: ['Bellagio Las Vegas'],
  onlyNewReviews: true,
  stateKey: 'bellagio-daily',
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items.filter((r) => r.rowType === 'hotel-summary').map((r) => r.topics.slice(0, 5)));

Pricing

You pay per review delivered, at the price of your Apify plan. Summary, coverage and error rows and repeats in only-new mode are free.

Apify plan	Per 1,000 reviews
Free	$0.54
Starter (Bronze)	$0.41
Scale (Silver)	$0.36
Business (Gold)	$0.27
Platinum	$0.18
Diamond	$0.14

Plus a one-off $0.001 per GB of memory when a run starts.

Set a maximum cost per run in the Apify run options and the actor stops before it goes over.

Use cases

• Reputation monitoring: new Google reviews of your hotels every morning (onlyNewReviews), with the reply status of each.
• Competitive set: compare topic sentiment (breakfast, service, location) of your hotel with the five hotels next door.
• Review analytics: thousands of reviews per hotel with trip type and aspect scores, translated into one language.
• Reply quality: which reviews got an answer, how fast (extras.ownerResponseDate), in which language.

Limitations

• Google shows review dates only as "3 months ago" or "a year ago", so date is approximate: to the day for the last week, then to the week, month or year (extras.datePrecision). The date filter and only-new mode work with these dates. Dates are read for English, German, French, Spanish, Italian, Portuguese and Dutch; other language values leave date empty (the original wording is in extras.dateText).
• Partner sites appear with the sample of reviews Google shows (e.g. 109 of 1,073 Tripadvisor reviews of a large hotel), not their full review list. For all Tripadvisor or Booking.com reviews use a source actor.
• Google does not state the review language; language is null. A translated review has extras.translatedTo.
• Topic filters use Google's topic list of each hotel; mention counts in the summary also count reviews the list does not return, so a topic filter can return fewer reviews than the topic's mentions.
• The actor reads Google over Apify's shared datacenter proxy. When Google rate-limits that pool, the run retries with short pauses and, if the limit persists, stops and reports the remaining hotels as free blocked error rows. Run again later.

Privacy

Reviewer names, profile links, avatars and photo links are never collected into the output. The reviewer's display name is only used, in memory, to remove it from the review and reply texts; greetings and signatures in replies ("Dear ...", "Kind regards, ...") are redacted to [name], and emails and phone numbers are removed from all texts.

FAQ

How do I find a hotel's entity token? It is the entityId of any row and the Ch... part of a Google Hotels link (/travel/hotels/entity/<token>). A hotel name with the city works too; the actor checks that Google's match shares a word with your name and reports no_match otherwise.

Why are there more reviews than Google's review count? reviewCount counts Google reviews only. With reviewSource: "all" the partner-site reviews Google shows come on top.

Why is text in Korean when I asked for English? text is always the original. The English version is in extras.textTranslated.

Something looks wrong in the output. Who do I tell? Open an issue on the actor page with the run link. We read them all.