Google Hotels Prices Scraper & API - OTA Rates, Hotel Details avatar

Google Hotels Prices Scraper & API - OTA Rates, Hotel Details

Pricing

from $4.00 / 1,000 hotel price rows

Go to Apify Store
Google Hotels Prices Scraper & API - OTA Rates, Hotel Details

Google Hotels Prices Scraper & API - OTA Rates, Hotel Details

Google Hotels API: hotel prices by city, hotel name or URL for your dates. One row per hotel: nightly rate, stay total, taxes, deals, availability, rating. Optional OTA rates (Booking.com, Expedia, Agoda, official site): rate parity, price comparison, price tracking. Free: address, phone, amenities.

Pricing

from $4.00 / 1,000 hotel price rows

Rating

0.0

(0)

Developer

Tedj MEABIOU

Tedj MEABIOU

Maintained by Community

Actor stats

1

Bookmarked

293

Total users

158

Monthly active users

4 hours ago

Last modified

Features

Limited permissions

Share

Google Hotels prices scraper for exact dates and guests: the lowest nightly rate and the stay total for every hotel in a search, then every booking site's rate for each hotel — Booking.com, Expedia, Agoda, Hotels.com, Trip.com and the hotel's official site — with free-cancellation flags and deep links. Run it once for a price check, on a schedule for hotel price tracking and rate parity, or with a date sweep to build a price calendar and your own hotel price history. No Google API key, no login, no browser: it reads the same public Google Hotels results any visitor sees, through Apify's proxies, and returns clean JSON.

You pay only for priced rows. Sold-out hotels, empty searches and status rows cost nothing, and the price per row is stated on this page and identical on every plan.

Connect this Actor to your AI agent

Open the MCP setup and small sample call for this Actor, or choose a related toolkit for a larger workflow. The setup includes the input schema and current pricing events.

For a per-run dollar ceiling, use the hosted MCP call-actor tool with callOptions.maxTotalChargeUsd outside the Actor input. After the run finishes, retrieve the dataset with get-dataset-items; a run response is not the data itself. Check the live schema and pricing before a large or recurring run.

Read the hotel and booking-site offer rows

For cross-site price comparison, use includeOffers: true (the default). The dataset is a flat list of three row types, not a hotel object with nested offers:

Row typeMeaning and useful fields
hotelOne hotel's lowest price for the stay: entity_id, check_in, check_out, name, nightly, total, currency. Its source can be null; read offer rows for individual booking sites.
offerOne booking site's rate: source, official, nightly, total, currency, link, partner_url, cancellation fields and the same stay keys. Optional values and links can be null.
statusCounts, error and offer_errors for a place/property and stay. Never charged. Check these before interpreting missing offers.

Join hotel and offer rows on (entity_id, check_in, check_out). Do not join solely on hotel_id, which can be null. For each hotel/stay, collect the matching offer rows to build a nested comparison for your user. Keep different currencies and stay dates separate. official: true means an official-site offer is actually present; its absence does not prove there is no direct rate or no parity violation.

Read the declared dataset fields, an observed hotel-and-offer dataset, and the example's input, date and limits. The example contains historical prices, not a current quote. A sampled output schema from an integration can omit optional offer fields; use the declared contract and actual returned rows together.

includeOffers: false omits the offer rows and costs at most $0.004 per delivered priced hotel. With offers enabled, maxOffersPerHotel: N for N greater than zero bounds each hotel/stay at $0.004 + N × $0.002. Ten hotels with eight offers each cost at most $0.20 in listed events. A cap keeps the cheapest offers and can omit the official site; use a suitable row cap and a separate callOptions.maxTotalChargeUsd run ceiling, then report incomplete comparisons accurately.

Google 酒店价格、酒店比价与 OTA 房价:中文调用与语言设置

查询城市或具体酒店的指定入住日期,获得酒店价格、税费与可订状态。需要多渠道比价时开启 includeOffers;只需要酒店行时关闭,避免多余的报价行费用。

language 设置显示语言;country 是搜索者市场,不是酒店所在地;currency 独立设置币种。中文界面示例 language="zh-CN";不要因为目的地在中国就自动改变搜索市场。

调用需求参数与结果约定
小规模验证使用下面 JSON;先检查结果和状态,再提高上限
语言与市场按上面的真实字段选择,不要把中文说明当成自动翻译承诺
费用控制maxTotalChargeUsd 是 Apify 运行选项,不是输入 JSON 字段;按定价页的计费事件设置上限
数据完整性检查状态行和 SUMMARY;空结果、来源错误与完整数据集是不同情况
{
"queries": [
"hotels in Lisbon"
],
"checkIn": "30 days",
"checkOut": "32 days",
"adults": 2,
"currency": "USD",
"maxHotels": 1,
"includeOffers": false,
"offerLevel": "sources",
"sweepDays": 1,
"language": "zh-CN",
"includeDetails": false,
"sessions": 1
}

Hotelpreise / prix des hôtels / precios de hoteles:使用同一组日期、人数和币种比较价格。

This localized quick start uses the existing input contract. API users should send the JSON as actor input and set maxTotalChargeUsd separately in Apify run options. Source coverage and translation depend on the platform, not the language of this guide.

What does Google Hotels Prices Scraper do?

Google Hotels is a metasearch engine: for a hotel and a stay it shows the rates of every booking site it knows and the hotel's own site. This scraper reads that comparison for the dates, occupancy, currency and market you choose and turns it into three kinds of rows:

  • hotel rows — up to 200 hotels per place search (Google's results, paged) or exactly the properties you list: lowest nightly rate, stay total, deal note ("19% less than usual"), star class, rating and review count, coordinates, website, Google Maps link and the Google Hotels entity id.
  • offer rows — for every priced hotel, each booking source's rate: nightly, total for the stay, free-cancellation flag and deadline, the source's link, and for official sites the booking-engine URL with your dates already filled in. Optionally every room type and rate.
  • status rows — one per place or property and stay: how many hotels, how many priced, how many offers, any error. Never charged.

It is the data behind a Google Hotels price tracker, a hotel price monitoring workflow, a rate shopping tool or a rate parity check — without the SaaS subscription, and with the raw hotel price data in your own dataset.

Google Hotels scraper for AI agents: which input answers which request?

Use it when the user wants hotel prices from Google Hotels for given dates: a city or area, a named hotel, or a list of hotels; one price per hotel or every booking site's rate; once or on a schedule. Do not use it for flights, for hotel stays of more than 6 adults (vacation rentals take up to 16) or more than one room, or for a rate straight from one booking site's own pages: every rate is the one Google Hotels lists. Review text comes from the reviews add-on (includeReviews, from 2026-12-01).

The user asks forInput
Hotel prices in a city for dates{"queries": ["hotels in Lisbon"], "checkIn": "30 days", "checkOut": "33 days", "includeOffers": false}
The 30 cheapest 4-star and 5-star hotelsadd "maxHotels": 30, "sortBy": "lowest_price", "hotelClass": ["4", "5"]
Pool, free cancellation, under 250 a night, rated 4.5+add "amenities": ["pool"], "freeCancellation": true, "maxPrice": 250, "minRating": "4.5"
Address, phone, amenities, photos of each hoteladd "includeDetails": true (no extra charge)
What each booking site charges for a named hotel, rate parity{"hotelNames": ["Hyatt Regency Lisbon"], "checkIn": "…", "checkOut": "…"} (includeOffers is already true)
Only Booking.com, Expedia and the hotel's own siteadd "offerSources": ["Booking.com", "Expedia", "official"] (only those offer rows are billed)
Every room type with its cancellation deadlineadd "offerLevel": "rooms"
Which check-in day is cheapest, a price calendar"sweepDays": 13 with "includeOffers": false
Track prices daily without editing the input"checkIn": "30 days", "checkOut": "32 days" and an Apify schedule
The same hotels every daypass the entity_id values of a first run in hotels
Hotels the user has as Google Maps links, share links or place idsput them in hotels: "ChIJ…", "place_id:ChIJ…", the Maps link or a https://maps.app.goo.gl/… share link — resolved exactly, never by name
Prices as a phone user sees them"device": "mobile"
Prices as a user in Japan sees them, names in Japanese"country": "jp", "language": "ja", "currency": "JPY"
A family"adults": 2, "childrenAges": [5, 9]
Apartments, houses or villas (vacation rentals)"vacationRentals": true with "minBedrooms": 2, "amenities": ["kitchen", "pool"], up to 16 adults
Prices as a visitor in the UK sees them (taxes included)"country": "gb", "currency": "GBP", "sameCountryIp": true
A price for every check-in date of the next 11 months"calendarNights": 330 (price calendar add-on, from 2026-12-01)
A spending limitthe Apify run option maxTotalChargeUsd; maxHotels and maxOffersPerHotel bound the rows

includeOffers defaults to true, and it is where the bill is decided: a hotel row alone is one event, a hotel with every booking site's rate is about 19 (measured on 174 hotels: 18 offer rows on average, 4 to 36). Set it false whenever the user asks for one price per hotel. With false the hotel row still carries the lowest nightly rate, but not which site sells it: that is in the offer rows.

Rows come in three types. hotel: name, nightly, total, total_base, taxes, fees, total_with_taxes, deal, stars, rating, reviews, lat, lng, entity_id, google_url. offer: source, official (true for the hotel's own site, present only when Google lists it), nightly, total, free_cancel, free_cancel_until, link, and with offerLevel: "rooms" also room and max_guests. status: one per place or hotel and stay with hotels, priced, offers, spread_km, place (Google's name for the place it searched) and error, never charged. An empty result is a status row that says why, not a failed run. The actor returns rows, not conclusions: "which day is cheapest" is a group-by on check_in over the hotel rows.

Google Hotels price tracking with your own data

Google's built-in hotel price tracking emails you when a price moves, but it picks the sources, keeps the history and gives you no export. Scheduling this scraper gives you the same signal as data you own:

  • Track a city or a shortlist daily. Schedule the run (Apify Console → Schedules, or the API) with relative dates such as "checkIn": "30 days", "checkOut": "32 days": every run prices the stay 30 days out, forever. Each run appends rows with fetched_at, so the dataset becomes your hotel price history.
  • Price alerts. A Google Hotels price alert you control: connect the run to n8n, Make or Zapier through Apify's integrations and alert on nightly below a threshold, or on a specific source dropping below the official site.
  • Rate calendars. sweepDays: 6 prices the same length of stay for each of the next seven check-in days in one run — a price calendar per hotel, or "which weekend is cheapest".

Google Hotels' own price history and price alerts stay inside Google; this is how you get the numbers out.

Rate parity and rate shopping for hotels

For revenue managers the offer rows are the point. One run answers, per property and stay: what is the official site charging, what is every OTA charging, which rate is refundable until when, and what is each source's stay total. That is a rate parity check across every channel Google lists, at the price of a few cents per hotel.

  • Put your properties and your competitive set in hotels (Google Hotels URLs — find them on google.com/travel/hotels — or the entity_id from an earlier run). No search is needed: one offers call per property and stay.
  • Use offerLevel: "sources" (default) for one row per source with its lowest rate — the rate shopping view — or offerLevel: "rooms" for every room type and rate.
  • Sources are keyed by source_id, and the property's own listing carries official: true, so a parity breach is a one-line comparison: any row where official is false and nightly is below the official row.

The same rows serve a hotel competitor rate shopping tool, hotel price intelligence dashboards, OTA rate comparison and channel-manager checks.

Input

FieldMeaning
queriesFree-text places exactly as you would type them into Google Hotels: "hotels in Lisbon", "hotels near Times Square", "5-star hotels Bangkok Sukhumvit". Google's results are paged 20 at a time; the scraper pages on until maxHotels distinct hotels.
hotelsSpecific properties without a search: Google Hotels URLs (…/travel/hotels/entity/ChkI…) or entity tokens from an earlier run, Google Maps links that carry the place (…!1s0x…:0x… in the link, or https://maps.google.com/?cid=…, the maps_url of a row) and Google Maps place ids (ChIJ…, also place_id:ChIJ…) and Maps share links (https://maps.app.goo.gl/…, followed through their redirect). One offers call per property and stay; a Maps link or place id adds one property-page call per run, which resolves it exactly (never by name) and fills star class, rating and coordinates. A place that is not a hotel is reported in its status row and costs nothing; a Maps link without the place in it is looked up by the name it carries.
hotelNamesSpecific properties by name: "Hyatt Regency Lisbon" (add the city), a Google Maps place link (looked up by the name in it) or a place id (resolved exactly). One lookup per name and run; a name nothing matches costs nothing and is reported in its status row, and only a property whose name matches yours is billed. These rows also carry star class, rating, reviews and coordinates.
checkIn, checkOutYYYY-MM-DD, or relative to the run date for schedules: "30 days", "6 weeks", "+30d", today, tomorrow. Nights = the difference. Past check-ins are rejected before anything is fetched. Left out, a run prices a 2-night stay 30 days out (a checkIn alone: one night) and says so on the run card.
sweepDaysAlso price the same length of stay starting on each of the next N days (rate calendars).
vacationRentalsSearch Google Hotels' Vacation rentals tab instead of hotels (place searches): apartments, houses and villas, each a hotel row with max_guests, bedrooms, bathrooms, rental_info and amenities. See Vacation rentals.
minBedrooms, minBathroomsVacation rentals only: at least this many bedrooms / bathrooms (Google's own filters, 0+ to 5+).
adults, childrenAgesOccupancy per room; prices scale exactly as on Google. Hotels take up to 6 adults: Google Hotels prices no hotel stay for a larger party (asked for 7 or more it returns no price and no booking site at all), so a scraper that accepts more can only return empty rows. Vacation rentals take up to 16: Google then lists the rentals that sleep that many.
currency, language, countryAny ISO currency for every price; Google UI language (hl); market (gl) — prices and available sources differ by market.
maxHotelsDistinct hotels per place and stay, up to 200 (one list call per 20).
sortByGoogle Hotels' own sort for place searches: relevance (default), lowest_price, highest_rating, most_reviewed. With maxHotels: 50 that is the 50 cheapest or the 50 best-rated hotels of a city. Google sorts prices on the basis of the country the request comes from — the pre-tax total_base from a US IP, the all-in total_with_taxes elsewhere. With lowest_price each search's rows then come cheapest first by the all-in price (total_with_taxes), numbered by price_rank; rank keeps Google's own position.
freeCancellationOnly hotels with a free-cancellation rate for your dates. Google then quotes the refundable rate where it has one. Hotels given by URL or name are held to it too: their row quotes the cheapest free-cancellation rate (and is free, without a price, when there is none) and only refundable offer rows come; a site whose terms Google does not state is left out.
minPrice, maxPriceNightly price range in your currency (decimals are fine: a maximum rounds up, a minimum down). Google filters on the price it displays (the nightly field), which is before taxes from a US IP and all-in from most other countries; filter again on total_with_taxes when the limit must include taxes.
minRating3.5, 4.0 or 4.5 — minimum Google guest rating. 4 reads as 4.0; a value between Google's steps (4.2) uses the step below it.
hotelClassStar classes to keep, e.g. ["4", "5"].
amenitiesHotels that have all of: free_wifi, free_breakfast, restaurant, bar, kid_friendly, pet_friendly, free_parking, parking, ev_charger, room_service, fitness_center, spa, pool, indoor_pool, outdoor_pool, air_conditioned, wheelchair_accessible, beach_access, all_inclusive.
specialOffersOnly hotels Google marks as a deal for your dates; the deal field carries the note ("23% less than usual").
includeOffersFetch every source's rate per hotel (one extra call per hotel).
offerLevelsources: one row per source, lowest rate (about 18 rows per hotel, 4 to 36). rooms: every source × room × rate with room names and cancellation deadlines (about 50–100 rows per hotel).
maxOffersPerHotel0 = every source; N = only the N cheapest rows per hotel (cost control).
offerSourcesOnly these booking sites' offer rows, e.g. ["Booking.com", "Expedia", "official"]; official is the hotel's own website. A site matches by the start of its name, any case ("booking" matches Booking.com; "Hotels.com" does not match Hotelscombined.com). Applied before maxOffersPerHotel; only kept rows are billed.
includeDetailsAdds what Google's property page says about each hotel: address, phone, place_id (Google Maps), description_long, rating_breakdown (5 to 1 stars), review_aspects (Service, Location, Breakfast… with positive and negative mention counts), external_ratings (Tripadvisor, Trip.com…), amenities and amenities_unavailable (about 120 of them, with notes such as Wi-Fi (free), Parking (extra charge), Front desk (24 hour)) and photos. One small extra call per hotel, at no extra charge. It also completes hotels given by URL with star class, rating and coordinates.
devicedesktop (default) or mobile: ask Google as a phone does. Google can show phones other prices and booking sites than computers: measured 2026-11-17, Hyatt Regency Lisbon was $241 with 12 booking sites on mobile and $250 with 10 on desktop from one network, while from Apify's US datacenter IPs the two hotels checked were within $0.30. Same rows, same price per row.
sameCountryIpAsk Google through Apify residential IPs in the country market instead of US datacenter IPs, so prices, sort order and the price filter are what a visitor there sees (the all-in price in most of Europe and Asia). No extra charge; a run is about twice as slow.
proxyConfigurationApify datacenter proxies work; keep the default.
sessions, perIpParallel proxy sessions and requests per second per IP (defaults 4 and 0.5).

Example: hotel prices in a city for a stay

{ "queries": ["hotels in Lisbon"], "checkIn": "2026-10-03", "checkOut": "2026-10-06", "adults": 2, "currency": "USD", "maxHotels": 60 }

Example: the 40 cheapest 4–5-star hotels rated 4.5+ with free cancellation

{ "queries": ["hotels in Lisbon"], "checkIn": "30 days", "checkOut": "32 days", "maxHotels": 40, "sortBy": "lowest_price", "hotelClass": ["4", "5"], "minRating": "4.5", "freeCancellation": true, "maxPrice": 350, "includeOffers": false }

Filters are Google Hotels' own and are applied by Google before anything is fetched, so you pay only for hotels that match. They apply to place searches (queries); hotels you name in hotels or hotelNames are always returned.

Example: hotel data with contact details, rating breakdown and review topics

{ "queries": ["hotels in Lisbon"], "checkIn": "30 days", "checkOut": "32 days", "maxHotels": 50, "includeDetails": true, "includeOffers": false }
{ "type": "hotel", "name": "Altis Grand Hotel", "stars": 5, "rating": 4.7, "reviews": 6419, "nightly": 172.45, "total": 344.9, "total_base": 344.9, "taxes": 38.6, "fees": 0, "total_with_taxes": 383.5,
"address": "R. Castilho 11, 1269-072 Lisboa, Portugal", "phone": "+351 21 310 6000", "place_id": "ChIJmYhOlXAzGQ0RmcenLVMx37g", "website": "https://www.altishotels.com/altis-grand-hotel/",
"rating_breakdown": [{ "stars": 5, "percent": 77, "reviews": 4811 }, { "stars": 4, "percent": 19, "reviews": 1221 }],
"review_aspects": [{ "aspect": "Service", "mentions": 611, "positive": 568, "negative": 29, "score": 0.79 }, { "aspect": "Location", "mentions": 332, "positive": 308, "negative": 7, "score": 0.77 }],
"external_ratings": [{ "source": "Tripadvisor", "rating": 4.3, "scale": 5, "reviews": 2082 }],
"amenities": ["Wi-Fi (free)", "Indoor pool", "Spa", "Fitness center", "Restaurant", "Bar", "Room service (24 hour)", "Parking (extra charge)", "Pet-friendly (extra charge)", "Airport shuttle"], "amenities_unavailable": ["Outdoor pool"] }

The detail fields cost nothing extra: a hotel row is the same price with or without them. Amenity names are in English whatever language is set to; an amenity Google lists that this actor has no verified name for is left out, never guessed.

Example: track named hotels, no URLs needed

{ "hotelNames": ["Hyatt Regency Lisbon", "Altis Grand Hotel Lisbon"], "checkIn": "30 days", "checkOut": "32 days", "currency": "EUR" }

Example: rate parity for your own properties, scheduled daily, always 30 days out

{ "hotels": ["https://www.google.com/travel/hotels/entity/ChkIhLCQwvjO2IYWGg0vZy8xMXE0bTZieDkyEAE"], "checkIn": "30 days", "checkOut": "32 days", "sweepDays": 6, "currency": "EUR", "offerLevel": "sources" }

Example: cheapest weekend in the next two months for a family

{ "queries": ["family hotels Orlando near Disney"], "checkIn": "2026-10-02", "checkOut": "2026-10-04", "sweepDays": 56, "adults": 2, "childrenAges": [5, 9], "currency": "USD", "maxHotels": 20, "includeOffers": false }

Hotel prices in Asia-Pacific: Singapore, Hong Kong, China, Taiwan, Japan, Korea and Southeast Asia

Every market Google Hotels serves works the same way: set country (the market), language and currency, and search in any language. Prices, deal notes, star-class labels and free-cancellation dates come back as that market shows them, in the one currency you asked for. Checked on 2026-09-20 in all of the markets below — 20 of 20 hotels priced in each.

MarketcountrylanguagecurrencyBooking sites Google lists there (besides Booking.com, Agoda, Trip.com and the hotel's own site)
Singaporesgen or zh-CNSGDKlook, Wego, AirAsia MOVE, Skyscanner, hutchgo
Hong Konghkzh-HK or enHKDWego, hutchgo, Qantas Hotels
Mainland Chinacnzh-CNCNYHotels.com, Expedia, Qantas Hotels
Taiwantwzh-TWTWDezTravel 易遊網, Lion Travel 雄獅旅遊, Klook
JapanjpjaJPYExpedia.co.jp, Wego
South KoreakrkoKRWTripbtoz, Creatrip, Klook, Wego
IndonesiaididIDRAirAsia MOVE, KAYAK, Hotelscombined
ThailandththTHBTraveloka, Klook, Wego, EaseMyTrip
Malaysiamyms or enMYRKlook, Expedia, Qantas Hotels
VietnamvnviVNDKAYAK, Hotelscombined, HomeToGo

Queries and hotel names can be written in the local script: "新加坡酒店", "香港酒店", "上海酒店", "台北飯店", "東京 ホテル", "서울 호텔", "hotel di Jakarta", "โรงแรมในกรุงเทพ". hotelNames understands Chinese, Japanese, Korean and Thai names too — "香港半島酒店", "上海外滩华尔道夫酒店", "パーク ハイアット 東京", "서울신라호텔" — and a name Google lists in another script ("台北君悅酒店" → Grand Hyatt Taipei) is matched as well.

Example: Singapore hotels in SGD, cheapest 4–5-star first

{ "queries": ["hotels in Singapore"], "country": "sg", "language": "en", "currency": "SGD", "checkIn": "30 days", "checkOut": "32 days", "maxHotels": 40, "hotelClass": ["4", "5"], "sortBy": "lowest_price" }

Example: Hong Kong hotels in HKD, in Traditional Chinese

{ "queries": ["香港酒店"], "hotelNames": ["香港半島酒店"], "country": "hk", "language": "zh-HK", "currency": "HKD", "checkIn": "30 days", "checkOut": "32 days", "maxHotels": 40 }

Example: rate parity for hotels in Shanghai, Tokyo and Seoul, every source, daily

{ "hotelNames": ["上海外滩华尔道夫酒店", "Park Hyatt Tokyo", "서울신라호텔"], "currency": "USD", "checkIn": "30 days", "checkOut": "31 days", "offerLevel": "sources" }

简体中文 — 谷歌酒店价格爬虫

谷歌酒店(Google Hotels)价格爬虫:按入住日期和人数抓取酒店价格,包括每晚最低价、整段住宿总价,以及 Booking.com、Agoda、携程 Trip.com、Klook 和酒店官网等每个预订网站的报价,并标明是否可免费取消。适用于酒店价格监控、酒店比价、价格追踪和渠道价格一致性(rate parity)检查。支持中文搜索词和中文酒店名称,支持人民币(CNY)、新加坡元(SGD)、港币(HKD)等币种,无需 API 密钥,按实际获得的价格数据付费。

繁體中文 — Google 飯店價格爬蟲

Google 飯店(Google Hotels)價格爬蟲:依入住日期與人數擷取飯店與酒店價格,包含每晚最低價、整段住宿總價,以及 Booking.com、Agoda、Trip.com、Klook、易遊網、雄獅旅遊和飯店官網等各訂房網站的報價與免費取消條件。適用於酒店價格監控、飯店比價、價格追蹤與通路價格一致性檢查。支援中文搜尋與中文酒店名稱,支援港幣(HKD)、新台幣(TWD)、新加坡幣(SGD)。

日本語 — Google ホテル料金スクレイパー

Google ホテルの料金を日付と人数を指定して取得します。1泊あたりの最安値、滞在合計、Booking.com・Agoda・Trip.com・Expedia・公式サイトなど予約サイトごとの料金と無料キャンセルの有無を出力。ホテル料金の監視、料金比較、価格追跡、レートパリティの確認に。日本語の検索語とホテル名、日本円(JPY)に対応。

한국어 — 구글 호텔 가격 스크레이퍼

구글 호텔(Google Hotels)에서 날짜와 인원에 맞는 호텔 가격을 수집합니다. 1박 최저가, 총 숙박 요금, Booking.com·아고다·Trip.com·공식 홈페이지 등 예약 사이트별 요금과 무료 취소 여부를 제공합니다. 호텔 가격 모니터링, 가격 비교, 가격 추적에 적합하며 한국어 검색어와 호텔 이름, 원화(KRW)를 지원합니다.

Bahasa Indonesia, ไทย, Tiếng Việt

Harga hotel dari Google Hotels untuk tanggal dan jumlah tamu tertentu: harga termurah per malam, total menginap, dan harga tiap situs pemesanan (Agoda, Trip.com, Booking.com, AirAsia MOVE) — untuk memantau harga hotel dan membandingkan harga hotel, dalam Rupiah (IDR) atau Ringgit (MYR). ราคาโรงแรมจาก Google Hotels ตามวันที่เข้าพัก: ราคาต่ำสุดต่อคืนและราคาจากทุกเว็บไซต์จองโรงแรม สำหรับเปรียบเทียบราคาโรงแรมและติดตามราคาโรงแรมเป็นเงินบาท (THB). Giá khách sạn từ Google Hotels theo ngày lưu trú: giá thấp nhất mỗi đêm và giá của từng trang đặt phòng, để so sánh giá khách sạn và theo dõi giá khách sạn bằng đồng (VND).

Output

hotel row (the Console's Hotels and Offers by source views show the same rows as tables; the dataset schema documents every field):

{ "type": "hotel", "query": "hotels in Lisbon", "check_in": "2026-10-03", "check_out": "2026-10-06", "nights": 3, "adults": 2, "children": 0, "currency": "USD",
"rank": 1, "name": "The Central House Lisbon Baixa", "hotel_id": "10088532857338925820", "entity_id": "ChkIg-b2ismUj7M1Gg0vZy8xMWg3MThreGg1EAE",
"stars": 2, "class_label": "2-star hotel", "rating": 4.3, "reviews": 727, "lat": 38.7131, "lng": -9.1368, "types": ["hostel"],
"nightly": 81.81, "nightly_display": "$82", "total": 245, "total_display": "$245", "price_basis": "all_in", "deal": "19% less than usual", "source": null,
"website": "https://www.thecentralhousegroup.com/", "maps_url": "https://maps.google.com/?cid=3848859359889388291", "thumbnail": "https://lh3.googleusercontent.com/gps-cs-s/…",
"google_url": "https://www.google.com/travel/hotels/entity/ChkIg-b2ismUj7M1Gg0vZy8xMWg3MThreGg1EAE", "total_results": 15000, "fetched_at": "2026-08-25T14:03:11+00:00" }

offer row (one per source with offerLevel: "sources"; per room and rate with "rooms"):

{ "type": "offer", "name": "Hyatt Regency Lisbon", "entity_id": "ChoI4cjBy-KGurGyARoNL2cvMTFtdnFfdncydhAB", "check_in": "2026-10-03", "check_out": "2026-10-06", "nights": 3, "adults": 2, "children": 0, "currency": "USD",
"source": "Booking.com", "source_id": 184, "official": false, "room": null, "nightly": 569.35, "total": 1708.05, "max_guests": null,
"free_cancel": null, "free_cancel_until": null, "link": "https://www.google.com/travel/lodging/clk?pc=…", "partner_url": "https://www.booking.com/hotel/pt/hyatt-regency-lisbon.html?checkin=2026-10-03&checkout=2026-10-06…", "lowest": true, "price_basis": "all_in", "fetched_at": "2026-08-25T14:03:14+00:00" }

Room-level rows carry the free-cancellation deadline three ways: free_cancel_until as Google words it ("Oct 25", "26. Okt.", "10月22日"), free_cancel_until_date as an ISO date (2026-10-25) and free_cancel_until_time ("11:59 PM"). The date is derived only where the wording cannot be misread — English, German, French, Spanish, Italian, Portuguese, Dutch, Chinese and Japanese — and left empty otherwise. Rows with "lowest": true are each source's cheapest rate, with Google's own cancellation terms for that rate when Google states them (about a third of sites): free_cancel true with its deadline, or false; null when Google does not say. offerSources keeps the rows of the sites you name, e.g. ["Booking.com", "Expedia", "official"] (official = the hotel's own website); only those rows are billed. Official-site rows carry the hotel's booking engine URL with your dates in partner_url. A source is keyed by source_id — a name such as Holidu or Booking.com can appear twice for one property when Google lists both its official listing (official: true) and its marketplace rate. Properties given in hotels are answered from their offers page: the hotel row carries name, headline nightly rate, stay total and that rate's tax breakdown (total_base, taxes, fees, total_with_taxes, the same fields a place search gives); a Google Maps link or place id also brings star class, rating, coordinates and the Maps link from the property page, and includeDetails adds them for a URL.

status row (never charged):

{ "type": "status", "query": "hotels in Lisbon", "hotel": null, "check_in": "2026-10-03", "check_out": "2026-10-06", "hotels": 25, "priced": 24, "offers": 310, "duplicates": 0, "offer_errors": 0, "spread_km": 1.2, "place": "Lisbon", "error": null, "note": null, "fetched_at": "2026-08-25T14:03:20+00:00" }

place is Google's own name for the place it searched ("Times Square" → "New York", "Algarve" → "Faro District"). When Google cannot place a query at all, it answers with hotels from anywhere in the market's country; the run then delivers none of them, bills nothing for that query and says so in error (not found: Google could not place 'qqqqzzzz9999' and answered with unrelated hotels — nothing charged …).

What each price includes: price_basis. Every priced hotel, offer and calendar row says whether its nightly and total are all_in (taxes and fees included) or before_taxes (mandatory fees included, taxes not), read from Google's own breakdown of that answer: Google picks the basis by the country of the IP that asks and by the destination, so it is never assumed: from Apify's default datacenter IPs Paris came back all-in in one run (bU6K0Jj9uMk2t4wSC) and before taxes in another (qgxbyT0YWxK240htZ), with Times Square, the Algarve and Tokyo before taxes in both. total_with_taxes is the all-in price on either basis. note on a status row says what that search or hotel left out and why: a pasted Google URL read as its search text (and that its own dates and filters were not applied), a search Google listed no hotels for, how many booking sites came with room-level rates, a refundable rate without a tax breakdown.

Free extras on every row

These come in the answers the actor already reads, so they cost nothing and add no request:

  • photos — up to 10 photo URLs from each hotel's card in a place search (9 on every card checked). Change the size suffix of a URL (=s287-w287-h192… → =s1200) for a larger image. includeDetails gives 9 from the property page for hotels given by URL.
  • nearby — what Google's card lists around the hotel, with travel times: [{"type": "airport", "name": "Humberto Delgado Airport", "drive": "28 min", "transit": "28 min"}, {"type": "transit", "name": "Pç. Figueira", "walk": "2 min", "line": "Bus"}, {"type": "attraction", …}, {"type": "restaurant", …}] (on 18 of 20 cards in a Lisbon sample).
  • typical_price_low, typical_price_high, price_level — Google's own price insight for a hotel given by URL, name, place id or link: its typical nightly range for your dates and whether today's price is low, typical or high against it (Altis Grand Hotel: $175 now, typically $182–$287 → low).
  • Each booking site's cancellation terms on its lowest-rate offer row, when Google states them (about a third of sites): free_cancel with free_cancel_until.
  • place on every place search's status row: Google's own name for the place it searched.
  • total_base, taxes, fees, total_with_taxes on every priced hotel row, and star class, rating, coordinates, check-in and check-out times on every row from a place search or a name.

Speed

Measured on 2026-11-17 from the moment the run was started to the first row in the dataset and to the end of the run:

RunFirst rowWhole run
The input form's defaults (20 hotels in Lisbon with every booking site's rate, 357 rows)4.4 s18 s
20 hotels in Paris with every booking site's rate6.2 s18 s
3 hotels by name with every booking site's rate5.1 s11 s
60 hotels in Paris with the 2 cheapest booking sites each (177 rows)—38 s (42 s before 1.15)
The most-used comparable scraper on the Store, 20 hotels in Paris with prices52 s61 s

Measured on 2026-11-18 with the same input on 1.16 and 1.17 started at the same moment, so Google and the proxy pool were in the same mood for both:

Run1.161.17
100 hotels in Paris with the 3 cheapest booking sites each (386 rows)140 s50 s
10 hotels by name with the 3 cheapest booking sites each26 s19 s
10 hotels by URL, two of them vacation rentals56 s, 7.8 MB26 s, 3.3 MB

The first row of a 100-hotel search still comes after about 16 s: its five pages are read one after another from one IP (Google's page tokens belong to the IP that issued them) before the booking sites are asked.

A run that plans 60 or more Google calls (60 hotels with offers, or a long date sweep) and leaves sessions at its default spreads them over 8 IPs instead of 4, each still paced at 0.5 requests per second, and the run card shows what is done so far every 15 seconds while it works. Most of a large run is the per-IP pacing Google tolerates; raise sessions (up to 16) for very large sweeps, and give runs of an hour or more a longer timeout.

Vacation rentals

vacationRentals: true turns a place search into Google Hotels' Vacation rentals tab: apartments, houses and villas from the rental sites Google lists (Vrbo, HomeToGo, Holidu, Casamundo, Booking.com and others). Each rental is a hotel row with its nightly price, stay total, tax breakdown, rating, coordinates and Google Hotels link, plus what its card says:

  • max_guests, bedrooms, bathrooms — numbers from the card;
  • rental_info — Google's essential info as shown, in the page language: ["Entire apartment", "Sleeps 4", "2 bedrooms", "1 bathroom", "3 beds", "861 sq ft"];
  • amenities — the amenities the card names: ["Kitchen", "Free Wi-Fi", "Washer", "Air conditioning", …] (a card names only part of a rental's amenities).

Filters are the rentals tab's own: minBedrooms and minBathrooms (0+ to 5+), amenities (kitchen, hot_tub, patio_or_deck, outdoor_grill, crib, fireplace, pool, free_wifi, air_conditioned, kid_friendly, pet_friendly, fitness_center), minRating and sortBy. minPrice/maxPrice are applied by this actor to each rental's nightly price, because Google's own price filter on this tab keeps only a few dozen of thousands (measured 2026-11-17: 14,853 Lisbon rentals, 10 left under $150 a night and 39 under $1,000); rentals outside the range are skipped and never billed. Groups of up to 16 adults work here (Google lists the rentals that sleep that many). The hotel-only filters hotelClass, freeCancellation and specialOffers are refused for free. A rental row costs the same as a hotel row, and includeOffers adds each listing site's rate as offer rows (some rentals list none).

{ "queries": ["Lisbon"], "vacationRentals": true, "checkIn": "30 days", "checkOut": "35 days", "adults": 4, "minBedrooms": 2, "amenities": ["kitchen", "air_conditioned"], "sortBy": "lowest_price", "maxPrice": 250, "maxHotels": 40, "includeOffers": false }

Prices as a local visitor sees them

Google decides what to display as the price from the country of the IP that asks and from the destination, not from the market you choose: from Apify's default datacenter IPs Paris came back all-in in one run (bU6K0Jj9uMk2t4wSC) and before taxes in another (qgxbyT0YWxK240htZ), Times Square, the Algarve and Tokyo before taxes (with mandatory fees), and visitors in most of Europe and Asia see the all-in price; Google sorts and price-filters on that same figure. Every priced row says which basis it is on in price_basis (all_in or before_taxes), and every row carries total_base, taxes, fees and total_with_taxes whatever the IP, so most comparisons need nothing more. Some booking sites are also listed only to visitors in their own country: Traveloka came back for a Bali villa through Indonesian IPs (run tO4c87bVCfiTtM9TP) and not from the default US datacenter IPs (run iFozZhFzBsY4cvt5h). A run for another market's country therefore puts a sameCountryIp tip in its SUMMARY, and on its run card when its prices came back before taxes. When the displayed figures, Google's order or its price filter must be exactly a local visitor's, set sameCountryIp: true: the run asks Google through Apify residential IPs in the country market ("gb", "de", "jp"…). There is no extra charge per row; such a run is slower (40 s instead of 18 s for 20 hotels with offers, measured 2026-11-17). Since 1.18 it asks each hotel's booking sites once: Google's one-price all-in answer was taken for a cut-down rate ladder and asked twice more (20 extra residential calls for 10 London hotels, run rJ4hT3h1ZBreP9KTg); 5 London hotels with 68 booking-site rows then took 7 s and 0.9 MB (run iBFiBVnPVglgCckxg).

{ "queries": ["hotels in London"], "country": "gb", "currency": "GBP", "sameCountryIp": true, "checkIn": "30 days", "checkOut": "32 days", "maxHotels": 20 }

Price calendar and reviews add-ons (optional)

Two add-ons follow every hotel a run delivers — from a city search, a URL in hotels or a name in hotelNames. Both are off by default; a run that does not ask for them returns exactly the rows and bill it always did. maxItems still caps the hotel rows, and the add-ons only cover the hotels delivered.

Price calendar — calendarNights (1-330) check-in dates from calendarFrom (default: checkIn), or calendarFrom to calendarTo. Each date is priced for the same stay length (checkIn to checkOut) and guests, in one call per hotel however many dates: Google prices about 330 days ahead (checked 2026-11-17: 322 and 329 of 330 dates priced on two hotels, the last one 330 days out), so calendarNights: 330 is a hotel's whole bookable year. One calendar row per hotel and date: nightly, total, total_with_taxes, currency, and three markers — no_rate (no price for the stay length asked), min_nights (Google prices this check-in only for a longer stay: a minimum stay, not a sold-out night) and sold_out (no booking site has a rate, not even one or two nights longer, while later dates are priced). Dates Google does not price for the asked length are asked again one and two nights longer, one call each and only when such dates exist; on 2026-11-18, 8 of 10 Lisbon vacation rentals and 2 of 5 Lisbon hotels had dates that need 2 or 3 nights. Rows without a price are free, including minimum-stay dates: a longer stay's price is never billed as a one-night rate. The Console's Price calendar view shows them as a table.

{ "hotelNames": ["Hyatt Regency Lisbon"], "queries": [], "checkIn": "14 days", "checkOut": "16 days", "includeOffers": false, "calendarNights": 30 }
{ "type": "calendar", "name": "Hyatt Regency Lisbon", "entity_id": "ChoI4cjBy-KGurGyARoNL2cvMTFtdnFfdncydhAB", "check_in": "2026-11-30", "check_out": "2026-12-02", "nights": 2, "currency": "USD",
"nightly": 210.48, "nightly_display": "$210", "total": 421, "total_base": 363.51, "taxes": 57.45, "fees": 0, "total_with_taxes": 420.97, "sold_out": false, "no_rate": false }

Reviews — includeReviews: true adds up to maxReviewsPerHotel (default 50) reviews per hotel, in reviewsSort order (newest by default, or most_relevant, highest_rating, lowest_rating). reviewsMinRating, reviewsMaxRating and reviewsMaxAgeDays filter before billing: a review they drop is never charged. One review row per review: review_id, source (Google, Tripadvisor…), rating, text (translated into the run's language when Google translates it, the original in text_original), date_text as shown ("3 weeks ago") and date derived from it — approximate, because Google shows only a review's age — trip_type (Business, Vacation), travel_group (Family, Couple, Friends, Solo), author and owner_reply. The Reviews view lists them.

{ "hotels": ["https://www.google.com/travel/hotels/entity/ChoI4cjBy-KGurGyARoNL2cvMTFtdnFfdncydhAB"], "checkIn": "30 days", "checkOut": "32 days", "includeOffers": false, "includeReviews": true, "maxReviewsPerHotel": 100, "reviewsMaxRating": 2, "reviewsMaxAgeDays": 90 }

Prices: $0.0004 per calendar night with a price (a 30-night calendar is $0.012 per hotel) and $0.001 per review, with Store discounts on higher Apify plans. These prices take effect on 2026-12-01 — Apify gives existing users 14 days' notice of a new price. Until then a run that asks for an add-on delivers everything else as usual, skips the add-on, says so on the run card and charges nothing for it.

Inputs from other scrapers work too

An input written for another Google Hotels scraper runs here unchanged: their field names are read as ours, the run's SUMMARY record lists each one under aliased_fields, and a name this actor has no use for is listed under ignored_fields (and on the run card) instead of failing the run. Our own field always wins when both are given.

Their fieldRead as
searchQueries, locations, location, q, searchQuery, destination, searches[].locationqueries
hotelUrls, propertyTokens, property_token, entity, startUrlshotels (URLs, tokens, Maps links, place ids) or hotelNames (names)
checkInDate, check_in_date, checkin, start / checkOutDate, check_out_date, checkoutcheckIn / checkOut
nights, los, days (with a check-in date)the stay length: checkOut = check-in + that many nights
maxResults, maxResultsPerSearch, maxHotelsPerLocation, limit, resultsWanted, maxItemsPerQuery, max_results, max_pages (× 20)maxHotels (above 200 held to 200; 0, their "no limit", leaves ours)
countryCode, gl, market / languageCode, hlcountry / language
children (a count) / children_ageschildrenAges (a count without ages is priced as children aged 10)
includePrices, includeBookingPrices, includeProviderPrices / includeRoomRatesincludeOffers / offerLevel: "rooms"
onlyDealsspecialOffers
min_price, minPricePerNight / max_price, maxPricePerNightminPrice / maxPrice
minGuestRating, guestRating, guest_ratingminRating (Google's step at or below the value)
minStars / stars, hotel_classhotelClass (that class and above / the classes listed)
maxReviews / reviewSort, reviews_sort_bymaxReviewsPerHotel / reviewsSort
priceWindowEnd, lastCheckInDate / dateRangeDayssweepDays (every check-in date up to that day / that many dates)
priceCalendarTo / days with propertyTokens (the calendar scraper's shape)calendarTo / calendarNights
vacation_rentals, type: "vacation_rentals", propertyTypes: ["vacationRental"] / bedrooms, bathroomsvacationRentals / minBedrooms, minBathrooms
proxyConfigproxyConfiguration
device (desktop, mobile)device (the same)
providers, bookingSites, otasofferSources (only these booking sites; official = the hotel's own website)

A Google Travel or Google search URL pasted as a place (https://www.google.com/travel/search?q=hotels%20in%20Lisbon&ts=…) is searched as its text, and the status row, the run card and url_queries in SUMMARY say that the URL's own dates and filters were not applied: set them in the input. A hotel's own link in queries is priced as that hotel.

Their spellings of our values are read too: sortBy price_low, lowestPrice, rating, highestRating, mostReviewed; reviewsSort relevant, most_recent, highest_score, lowest_score; minRating 4; amenities freeWifi, Free Wi-Fi, Gym; device mobile. null for any optional field means "not given", currency may be lowercase and dates may be full ISO datetimes (2026-12-20T00:00:00Z). Apify checks the input's types before a run starts, so a few shapes are still refused there: whole numbers for hotelClass (send ["4", "5"] or use minStars: 4), a number for minRating (send "4.5"), and a single string where a list is expected (queries, hotels, hotelNames, childrenAges; query takes one place as a string). Google has no rooms control (it prices one room for the whole party), so rooms is ignored, and so are request timeouts. An includeVacationRentals switch (true = rentals mixed into hotel results) is ignored too: Google's hotel results already mix some in, and vacationRentals: true searches rentals only. A days field means the stay length unless the input is shaped like the calendar scraper's (propertyTokens, start, los), where it is the number of check-in dates.

How much does it cost?

Pay per event, only for priced rows: a hotel row (a hotel with a price for your dates) and an offer row (one source's rate). Sold-out hotels, empty searches and status rows are free, and a hotel and stay is billed once per run even when overlapping searches return it twice. A city search of 60 hotels with source-level offers is a few hundred rows; a daily parity check of ten properties with seven check-in days is about 70 hotel rows plus their offers. Apify's platform usage for a run of that size is a fraction of a cent — the runs above cost $0.001–$0.003 in compute. The exact per-row prices are on this page's pricing tab, and maxOffersPerHotel caps the offer rows when you only need the cheapest few sources. The optional add-ons bill their own rows: $0.0004 per price-calendar night with a price and $0.001 per review.

Google Hotels API or scraper: how this compares

Google has no public Google Hotels API for reading prices: its Hotel APIs (Hotel Prices, Hotel Ads, Book on Google) are partner feeds for hotels that send prices to Google. Reading prices back means a scraper or a SERP API vendor.

  • SerpApi Google Hotels API, DataForSEO, SearchApi, Bright Data sell a Google hotel price API per search. They return the search page; per-source rates and room-level offers usually need a second product or are not exposed, and you pay per search whether or not it had results.
  • This scraper returns hotels and every source's rate as flat rows, bills per priced row, handles paging to 200 hotels, occupancy including children's ages, any currency and market, relative dates for schedules, and property mode for your own hotels. It runs inside Apify, so scheduling, storage, webhooks and integrations come with it.
  • Writing your own with Python ("scrape google hotels", "scrape hotel prices", "web scraping hotel prices"): Google Hotels renders prices through internal RPC calls, not HTML, and the payload formats change. The point of this actor is that someone keeps that working and reports failures as status rows instead of silent zeros.

How it compares with the other Google Hotels scrapers on the Store

Read from the input schemas and public run statistics of the 31 other Google Hotels actors on Apify Store that had two or more users in the 30 days to 2026-11-17:

What to checkThis actorThe 31 others
Relative dates ("30 days"), so a schedule never goes staleYes11 of 31
Hotels by name, no URL or token to look upYes — also in Chinese, Japanese, Korean and Thai6 of 31
Hotels by Google Maps link or place id (ChIJ…), resolved exactlyYes5 of 31
Hotels by Maps share link (maps.app.goo.gl/…)Yes3 of 31
Prices as a phone sees themYes (device: "mobile")2 of 31
Photos, nearby places and Google's typical price range, free on the rowYesphotos are a paid or separate lookup in 7 of 31
The same stay across many check-in dates in one runYes (sweepDays up to 60, and the price calendar add-on up to 330 dates)9 of 31
Vacation rentalsYes, with Google's bedroom, bathroom and amenity filters14 of 31 (bedroom filter: 3 of 31)
Prices from residential IPs in the market's countryYes (sameCountryIp)6 of 31 (3 built in, 3 through their proxy setting)
Google's free-cancellation filterYes10 of 31
Google's own sort (price, rating, reviews), held across pagesYes16 of 31
Children's ages, which change the priceYes14 of 31
Hotel reviewsYes (add-on, from 2026-12-01)8 of 31
Runs that succeeded over the last 30 days2,426 of 2,455 (98.8%); 1 failed, 2 timed out, 26 stopped by the userfrom 80% to 100%; 3 of 31 below 90%

Two things no schema shows, so check them on any scraper you try. Complete rate ladders: about one proxy IP in eight gets a cut-down answer from Google — 6 or 7 booking sites instead of 20 to 26, with the lowest rate wrong; this actor recognises those answers, retires the IP and asks again. Complete result pages: Google's page tokens belong to the IP that issued them, so a scraper that spreads one search over several proxies can silently skip hotels; this actor keeps each search on one session. Taxes: every priced hotel row, also one given by URL, name or place id, carries base, taxes, fees and the all-in total separately, because what Google displays as the price depends on the country of the IP that asks.

If you searched for "Google Hotels API free" or a hotel price comparison API: there is a free tier on Apify, no API key from Google is needed, and empty results are never charged. As a hotel rates API it is one POST with the dates and a dataset of rows back.

Where other scrapers fail, and what this actor does instead

From the public run statistics, issues and changelogs of the other Google Hotels scrapers on the Store (30 days to 2026-11-18). The last column names the unit test or the cloud run that checks our side.

Where others failWhat this actor doesChecked by
Runs that fail on blocks and dead proxies. The two scrapers with the most users failed 960 and 460 runs in 30 days; one documents that some countries' datacenter IPs "have no available IPs and every request fails".On Google's rate limit the run holds and slows down. After 3 refusals in a row, including a proxy that refuses the connection (HTTP 407) or keeps dropping it, the rest of the run moves to Apify residential IPs in your proxy's country. 1 failed run of 2,455 in 30 days.test_three_straight_refusals_move_the_run_to_the_residential_fallback, test_a_dead_proxy_moves_the_run_to_the_residential_fallback_without_holding_it, test_the_residential_fallback_is_set_up_at_the_first_escalation_in_the_proxy_country; cloud run ejlr2tIyAv7Ms0U0P (a Portuguese datacenter proxy without IPs: moved to residential, priced)
Stay dates not respected. One scraper's "Number of Days not respected" issue has been open since February; another scraper's users reported "some dates are missing".Google's echo of every stay is checked: an answer for other dates is refused and never billed. Every stay of a sweep gets a priced row, a free unpriced row or a status row.test_silent_date_fallback_is_refused_rather_than_billed, test_sweep_days_bills_each_stay_separately, test_unpriced_hotels_are_delivered_free
Hotels from the wrong place. When Google cannot place a query it answers with hotels from anywhere in the country, and those get billed.Google says so in its answer: that query ends free with a not found status row, and every status row names the place Google searched (place).test_a_query_google_cannot_place_is_reported_free_instead_of_billing_hotels_from_anywhere; cloud run 26h2ALs0av9Z6P04T; 15 cloud runs of bare city names on 2026-11-18 all answered the right city (e.g. runs fmwp1pfTF58QnidUi Paris, k4DaNT2CiDkcb2zWc Springfield)
Minimum-stay nights shown as sold out in price calendars.Such dates carry min_nights (2 or 3 in practice), are not marked sold out and are free.test_a_minimum_stay_is_told_from_a_sold_out_night_and_stays_free (a real Lisbon rental: 13 two-night and 2 three-night minimum dates out of 46); run 8VM0fwRr4KXYG0upM on a priced private copy: 78 minimum-stay dates free, the 83 priced nights billed exactly
"Free cancellation only" for a named hotel. An open request on one scraper since October 2025.freeCancellation holds hotels given by URL or name to refundable rates: the hotel row quotes the cheapest refundable rate and only refundable offers come.test_free_cancellation_holds_hotels_given_by_url_or_name_to_refundable_rates; cloud run crvA8CFXsheAw2Ic7 (4 hotels: each row quotes its cheapest refundable site rate)
Sorted searches whose rows are not in order. One scraper fixed "sortBy is now applied to the rows you get" on 2026-11-14: Google sorts on the asking IP's price basis, before taxes from a US IP.With sortBy: "lowest_price" each search's rows come cheapest first by the all-in price, numbered by price_rank; rank keeps Google's own position.test_a_lowest_price_search_delivers_its_rows_cheapest_first_and_keeps_googles_rank; cloud run oaGBZWMd85Ofzqzac (30 Lisbon hotels)
Paying for booking sites you do not watch.offerSources keeps only the sites you name; the others are neither delivered nor billed.test_offer_sources_keep_only_the_named_sites_and_bill_only_them; cloud run tZgrpWMMTMMRSZzrt (3 hotels × 7 nights: only Booking.com, Expedia and the hotels' own sites)
Short rate ladders. About one proxy IP in eight gets 6 or 7 booking sites instead of 20 to 26, and a wrong lowest rate.The cut-down answer is recognised, that IP is retired and the hotel is asked again.test_a_short_answer_retires_that_ip_and_the_complete_answer_is_returned
An empty first page, about 1 run in 85, reported as "no results".Asked again from another IP before anything is reported.test_an_empty_first_page_is_asked_again_from_another_ip_and_a_truly_empty_search_stays_empty
Long runs that people give up on. One scraper had 25 aborted and 41 timed-out runs in 30 days.Calls go to whichever proxy session is free, so one slow IP no longer holds a run back; names are looked up side by side. 100 hotels in Paris with offers took 50 s on 1.17 against 140 s on 1.16, started at the same moment.cloud runs 7fYR7XSsADfUP8UhL and r4eGLjVuxIhuG9vsr; test_an_unpinned_call_takes_whichever_session_is_free_so_one_slow_ip_does_not_hold_back_the_run, test_names_are_looked_up_side_by_side
Big runs killed at the timeout. One scraper had 41 timed-out runs in 30 days.No new work starts in the run's last ~10%: it ends SUCCEEDED with everything it delivered and charged.test_near_the_run_timeout_no_new_job_starts_and_the_run_ends_green_with_what_it_charged

Questions users ask about other scrapers

These are the most frequent complaints in the reviews and issue threads of the other Google Hotels scrapers and of the Booking.com, Expedia and Tripadvisor price scrapers (156 items read on 2026-11-18). Each answer names the unit test or the cloud run that checks it.

"I asked for a hotel by name and was billed for another one." A name is matched on the words that say which property it is, not on "hotel", "villa", "inn", "suites" or the town, so a neighbour that shares only those is never billed. When Google lists no such property, or several properties fit the name (four Hilton-brand hotels for "Hilton Lisbon"), the name ends as a free not found status row that says why; pass the hotel's Google Maps link or place id in hotels, which name exactly one property. Before 1.18, "Abia Villa Legian" billed "Wyanna Villa LEGIAN" (run iFozZhFzBsY4cvt5h) and the invented "Zorblax Villa Legian" billed it too (run C5kmj5CHDbZK69Ndk); on 1.18 the same input billed only Abia Villa & Suite Legian, given by its place id, and reported the name not found, free (run EsRdsczVyVN1Mgp0G), and four invented names were all not found and free (run qZfk57x06WibhglaA). Checked by test_a_name_never_bills_a_neighbour_that_shares_only_villa_and_the_town and test_a_name_google_lists_no_property_for_is_not_found_instead_of_billing_a_neighbour (Google's own answers to 10 names).

"The price is not what the website shows." Google displays a price either all-in or before taxes, depending on the country of the IP that asks and on the destination: from Apify's default datacenter IPs Paris came back all-in in one run (bU6K0Jj9uMk2t4wSC) and before taxes in another (qgxbyT0YWxK240htZ). Every priced hotel, offer and calendar row says which in price_basis, read from Google's own breakdown of that answer, and total_with_taxes is always the all-in price: in run qgxbyT0YWxK240htZ each of the 12 hotels in Paris, Times Square, Tokyo and the Algarve said before_taxes, and in run GDSf06etJ6pJfd6SZ every one of 65 booking-site rows carried its hotel's basis. Checked by test_every_priced_row_says_whether_its_price_includes_taxes.

"A booking site I see in my browser (Traveloka, a local site) is missing." Google lists some sites only to visitors in their country: Traveloka came back for a Bali villa through Indonesian IPs (sameCountryIp: true, run tO4c87bVCfiTtM9TP) and not from the default US datacenter IPs (run iFozZhFzBsY4cvt5h). A run for another market's country carries that tip in its SUMMARY, and on its run card when its prices came back before taxes. Checked by test_a_market_seen_from_the_proxys_country_gets_the_same_country_ip_tip_where_it_changes_the_prices.

"I asked for 3 results and was billed for 20 (or 1,000)." maxHotels caps what each search delivers and bills, and another scraper's limit, resultsWanted, maxItemsPerQuery, max_results or maxResults is read as it (above 200 it is held to 200). maxItems and the run's spending limit are exact. Before 1.18 limit: 3 was ignored and 20 hotels were billed (run 0o1hicwaxeBh6Dar2); on 1.18 the same input billed 3 (run Fuj9Cq3B25xrgL7jN), and resultsWanted: 2 and max_results: 2 billed 2 each (runs I1EbLb0kWJvBE5d5b, jBp8XGRXIkdrzHpnx). Checked by test_another_scrapers_result_limit_caps_the_hotels_billed.

"The filters in my search URL were ignored." A Google Travel or Google search URL pasted into queries is searched as its text, and the status row, the run card and SUMMARY say that the URL's own dates and filters were not applied: set them in the input (checkIn, hotelClass, freeCancellation, minRating…). Before 1.18 the URL itself was searched and Google answered with 3 Dallas hotels, billed (run yhp5hDc3Jnc1yYCcY); on 1.18 the same input searched "hotels in Lisbon" and billed 3 hotels in Lisbon (run Ph37zvqg1VsofgRcc), and a URL with its own date and filter parameters said they were not applied (run euYzgsAjmKoSPeNFL). A search Google names no place for is refused free: "hotels near Times Square" from a datacenter IP was once answered with hotels in Phoenix (run HKpcH2koJgGrgw38n); in run QuIWTfUNLWqmfE0ep Google placed it in New York and it was delivered, while "hotels" named no place and was refused, free. Checked by test_a_pasted_google_travel_url_is_searched_as_its_text_and_says_its_filters_were_not_applied and test_an_answer_that_names_no_place_is_refused_free.

"Hotels given by URL come back with empty columns." A hotel given by URL gets its star class, rating, review count, coordinates, hotel id, Maps link and photos from the property page's own small call, asked once per run beside its rates, at no extra charge; includeDetails adds address, phone and amenities. Before 1.18, 12 of 13 such columns were null (run HXugqR8nR4sbBh8Ck); on 1.18 three hotels given by URL came back with all of them (run 8WqhOoLhW0VC1H2Zr). Checked by test_a_hotel_given_by_url_gets_its_class_rating_and_coordinates_without_include_details.

"I can't get the room names, or every site's room rates." offerLevel: "rooms" gives room names, rate plans, guests and cancellation deadlines for the 3 to 5 sites Google features with room-level rates (153 rows for 4 of Hyatt Regency Lisbon's 24 sites, run LddS2CfY7MiuOd9Ud); the other sites stay one lowest-rate row each, and the status row says how many sites came with rooms ("room-level rates came from 4 of 18 booking sites", run raKMKwxhDBQTVVAgp). Checked by test_status_rows_say_what_the_run_left_out.

"Only one length of stay, or one date instead of 365." Any stay length works (days: 3 in another scraper's input priced 3 nights on every row, run eRLUnHoLw6MiJZfLH), sweepDays prices up to 61 check-in dates in one run (run yveyVctEFCAjG7aRz: 8 stays of 3 nights for 2 hotels, 16 of 16 priced), and the price calendar add-on prices up to 330 check-in dates in one call per hotel from 2026-12-01; until then a run that asks for it says it was skipped and charges nothing for it (run Fnpqz5CTaRdEa6Bdh).

"Runs fail or get blocked." Google's rate limit makes the run hold and slow down, and three refusals in a row, or a proxy that refuses the connection, move the rest of the run to residential IPs in the proxy's country (run jmmh7bslxnGcuCbjZ: a Portuguese datacenter proxy without IPs, moved and priced). 1 failed run of 2,455 in 30 days.

"It is too slow for an API call." One hotel with every booking site's rate through run-sync-get-dataset-items took 5.4, 3.8 and 6.6 s from the call to the rows (runs GAH702sWSra13X0CC, c0tdPm2ni8xp5nzKL, rYVIfyLwgCk3pxBhX), and 200 hotels in Lisbon took 27 s (run g5tahgSWuSpIf3rD2).

How to use it from Python, JavaScript, curl, n8n, Make or an AI agent

Run it from the Apify Console with the form, or programmatically. Replace <TOKEN> with your Apify API token.

Python:

from apify_client import ApifyClient
client = ApifyClient("<TOKEN>")
run = client.actor("kestrel/google-hotels-prices").call(run_input={
"queries": ["hotels in Lisbon"], "checkIn": "2026-10-03", "checkOut": "2026-10-06", "adults": 2, "currency": "USD", "maxHotels": 40})
rows = list(client.dataset(run["defaultDatasetId"]).iterate_items())
hotels = [r for r in rows if r["type"] == "hotel"]; offers = [r for r in rows if r["type"] == "offer"]
cheapest = min((h for h in hotels if h["nightly"]), key=lambda h: h["nightly"])
print(cheapest["name"], cheapest["nightly_display"], cheapest["total_display"])

JavaScript / Node.js:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: '<TOKEN>' });
const run = await client.actor('kestrel/google-hotels-prices').call({ hotels: ['https://www.google.com/travel/hotels/entity/ChkIhLCQwvjO2IYWGg0vZy8xMXE0bTZieDkyEAE'], checkIn: '30 days', checkOut: '32 days', currency: 'EUR' });
const { items } = await client.dataset(run.defaultDatasetId).listItems();
const official = Math.min(...items.filter(x => x.type === 'offer' && x.official).map(x => x.nightly));
const parityBreaches = items.filter(o => o.type === 'offer' && !o.official && o.nightly < official);
console.log(parityBreaches.map(o => [o.name, o.source, o.nightly]));

curl:

curl -X POST "https://api.apify.com/v2/acts/kestrel~google-hotels-prices/run-sync-get-dataset-items?token=<TOKEN>" \
-H "Content-Type: application/json" \
-d '{"queries": ["hotels in Lisbon"], "checkIn": "2026-10-03", "checkOut": "2026-10-06", "maxHotels": 20, "includeOffers": false}'
  • n8n, Make, Zapier: use the Apify node/app with this actor's id, or a webhook on run finished, then read the dataset — a hotel price tracker or alert flow is three nodes.
  • AI agents and MCP: the actor is available through the Apify MCP server and Apify AI; the README, input schema and flat row types are written so an agent can ask "hotels in Lisbon under $150 from Oct 3 to 6 with free cancellation" and filter the rows.
  • Google Sheets, BigQuery, Airtable: export the dataset from the run page or through Apify's integrations.

The scraper reads publicly displayed prices — the same figures any visitor sees without logging in — and stores no personal data: rows describe hotels, sources and rates, not people. Collecting public pricing for analysis and monitoring is common practice in the hotel industry (every rate shopping tool does it), but terms of service and local law differ; check that your use complies, and use the data for your own analysis rather than republishing it as Google's.

Limits and honest notes

  • A place search pages through Google's ranking 20 at a time up to maxHotels (200); neighbourhood or landmark queries add coverage beyond that.
  • Prices are what Google Hotels shows for the market (country) and currency you choose; whether taxes and fees are included follows Google's display rules for that market. The total is the stay total Google shows.
  • Sources per hotel vary by market and day: 18 on average, 4 to 36, measured on 174 hotels in September 2026; room level averages about 65 rows per hotel and reaches about 150. Sold-out hotels appear as hotel rows without a price and are not charged.
  • Long runs end before their timeout, not at it. In the last ~10% of the run's timeout (Apify's run option, 3,600 s by default) no new search, hotel or add-on is started; the hotels in flight are delivered with their offers, and the run ends SUCCEEDED with time_hit: true and the number of stays not started in its SUMMARY record and on the run card. A 60-day sweep of 200 hotels with every booking site takes about two hours: give such a run a longer timeout.
  • Google changes its internal formats occasionally; a daily canary run watches for that, the actor is rebuilt within days, and every run reports status rows so a failure is visible rather than a silent zero.
  • A place Google cannot place is refused, free. Google answers every place search with hotels; when it cannot place the query it searches the market's whole country instead, and says so in its answer. Since 1.16 the run reads that, delivers none of those hotels and reports the query as not found on its status row (checked on 2026-11-18: 6 of 6 nonsense queries in two markets flagged, 0 of 40 real ones — cities, regions, landmarks, misspellings, Chinese names). A real but vague place is answered for what Google understood: place on the status row is Google's own name for it ("Springfield" → Springfield, Massachusetts from a US market), and spread_km is the median distance of the hotels from their centre (a city is single digits, a region tens, a country a few hundred). Add the country or region when a name is ambiguous.
  • A name is matched on the words that say which property it is, never on "hotel", "villa", "inn", "suites" or the town: a name Google lists no such property for, or that several properties fit ("Hilton Lisbon": four Hilton-brand hotels), ends as a free not found status row that says why, and the run card counts it as not found, not as a failed job. Pass the Google Maps link or place id of such a hotel in hotels: they name exactly one property.
  • A query Google names no place for is refused, free. From Apify's US datacenter IPs, "hotels" and "hotels near Times Square" were answered with hotels near the proxy's own location, in Phoenix, and Google named no place for either (run HKpcH2koJgGrgw38n; 50 of the 52 real places in that run named theirs, and of the other two, "Pariss" was flagged by Google and "hotels near Times Square" is the Phoenix answer). Since 1.18 such an answer is reported not found on the status row and nothing is charged: name the place ("hotels near Times Square, New York"). Queries such as "cheap hotels" that Google does place ("New York", from the proxy's location) are delivered, and place says where.
  • A hotel name in queries ("Hyatt Regency Lisbon") is answered by Google with that one property, not a list: the status row says so, nothing is charged, and the name belongs in hotelNames.

FAQ

Does it need a Google Hotels API key or a Google account?

No. There is no public Google Hotels API for reading prices; the scraper uses the public results page through Apify's proxies. No login, cookies or API key.

Can I get hotel price history and price alerts from Google Hotels?

Google's own tracker keeps history and alerts inside Google. Schedule this scraper (daily, with relative dates) and every run appends dated rows to your dataset — that is your price history — and an Apify integration (n8n, Make, Zapier, webhook) turns a drop into an alert.

How is this different from Google's "Track hotel prices" feature?

Google's feature emails you about the listings it chooses to watch and does not let you pick sources or export data. This scraper returns every source's rate for the exact dates, occupancy and currency you set, as rows you own.

Which booking sites are covered?

Whatever Google Hotels lists for that property and market: Booking.com, Expedia, Hotels.com, Agoda, Trip.com, Priceline, Kayak, eDreams, Vio.com, the official site and dozens of regional OTAs. The source name and source_id come straight from Google. To watch only some of them, list them in offerSources (["Booking.com", "Expedia", "official"]): the other sites' rows are neither delivered nor billed.

Does it include taxes and fees?

Every priced row says so in price_basis: all_in (taxes and fees included) or before_taxes (mandatory fees included, taxes not). The nightly and total fields are what Google displays, and Google picks that basis by the country of the IP that asks and by the destination, not by the market you ask for: from Apify's default datacenter IPs Paris came back all-in in one run (bU6K0Jj9uMk2t4wSC) and before taxes in another (qgxbyT0YWxK240htZ), so no rule of thumb such as "a US IP means before taxes" holds, and the actor reads the basis from Google's own breakdown of each answer instead (checked by test_every_priced_row_says_whether_its_price_includes_taxes). Every priced hotel row — from a place search, a Google Hotels URL, a name or a Google Maps place id — also carries that breakdown: total_base, taxes, fees (service charges, resort fees) and total_with_taxes, where total_base + taxes + fees = total_with_taxes (checked on 120 hotels in six markets). The same hotel can show $221 or $253 for the same stay depending on the proxy: compare on total_with_taxes (or total_base) and the proxy stops mattering. Booking-site offer rows are displayed on their answer's basis, which their price_basis gives; Google sends no breakdown per site. A hotel row that quotes a refundable rate other than Google's headline rate has no breakdown, and its status row says so.

How many hotels can one run return?

Up to 200 per place search, or as many properties as you list in hotels, times the number of stays (sweepDays). A run stops cleanly when it reaches the spending limit you set on the run.

Why do prices differ from what I see in my browser?

Google prices depend on market, currency, occupancy, dates, the device and the country your browser asks from. Match country, currency, adults, childrenAges and the dates, compare the total for the stay rather than a headline nightly figure, and check price_basis: a browser in Europe usually shows the all-in price (total_with_taxes) where a US IP shows it before taxes. sameCountryIp: true asks Google as a visitor in your country does, local booking sites included.

Can I use it as a hotel price scraper for Booking.com or Expedia specifically?

Yes for prices: offerSources: ["Booking.com"] returns only Booking.com's rate per hotel and stay (and bills only those rows), and likewise for Expedia, Agoda or Hotels.com — without scraping each site separately. For reviews or room descriptions use a Booking.com scraper or Expedia scraper built for that site.

Is there a free tier?

Yes: Apify's free plan covers small runs, and this scraper never charges for a run that returns no priced rows.

Last verified working: 2026-11-18.

Ready-made runs

Each link opens this actor with the input already filled in — press Start and it runs. Every one is capped, so what it bills is on the page before you run it. What each returns is spelled out at mtedj.github.io/kestrel-actors-examples/recipes.html.

Hotel prices are one side of a trip. These read the rest of it, and they share the same row shape, the same pay-per-delivered-row billing and the same scheduling story:

  • Agoda Prices Scraper — the same stay priced on Agoda, by destination or property, for a second channel in a rate comparison.
  • Booking.com Scraper — Hotel Prices — the same stay priced on Booking.com itself, by destination or property, no browser.
  • Hotel Price Tracker — only want the lowest rate per property and check-in date, cheap enough to run every morning? It runs the same engine and bills one row per property and date over a window of up to 60 dates, ranked, with a maxPrice alert that filters before billing.
  • Hotel Rate Parity Checker — the same comparison as one row per property and stay with the parity math done: cheapest and dearest source, spread, and which OTAs undercut the official site by how much.
  • Google Flights Scraper — airfare for a route and date, with a cheapest-day sweep.
  • Airbnb Scraper — Airbnb prices, availability calendars and occupancy.
  • Airbnb Reviews Scraper — every guest review of an Airbnb listing, with translations.
  • Agoda Reviews Scraper — Agoda hotel reviews with separate positives and negatives.

All of them bill per delivered row, never charge for rows a filter or a spending limit removed, and write an Apify dataset you can export to CSV, Excel or JSON.