Booking.com Reviews Scraper

Pull every guest review from any Booking.com property at scale — full review text, headlines, 1–10 guest scores, traveler segments, room types, stay dates, partner replies, and guest-uploaded photos — straight into a clean dataset, one review per row. Pipe in raw property URLs OR drop in the entire output of our Booking.com Hotel Scraper and start collecting reviews on the same run. Built for hospitality competitive-analysts, market-research firms, hotel-investment due-diligence teams, and SaaS reputation tools that need fresh, structured Booking.com review data without wrestling with rotating filters, multilingual pagination, and embedded reply threads.

Why This Scraper?

• $1.30 per 1,000 reviews — flat PPR. No compute charges, no per-property surcharge, no enrichment upcharges. One row = one review = one charge.
• Native pipe from the Booking.com Hotel Scraper — paste the upstream actor's dataset rows directly into startUrls. Any object with a url key is auto-detected; every other field is ignored. No glue script, no manual URL extraction.
• Five-axis filtering on Booking's native facets — 5 sort orders (Most Relevant, Newest, Oldest, Score High-to-Low, Score Low-to-High), 6 non-overlapping score buckets (Wonderful 9–10, Good 7–8, Okay 5–6, Poor 3–4, Very Poor 1–2, plus All), 26 review languages, 6 traveler segments (Families, Couples, Solo, Groups of Friends, Business, plus All), and 5 review-season windows keyed on the date the review was posted (Mar–May, Jun–Aug, Sep–Nov, Dec–Feb, plus Any time of year).
• 30 fields per review row — review headline, "Liked"/"Disliked" body, 1–10 guest score, helpful-vote count, language code, traveler segment, anonymous flag, guest display name, guest country (ISO-2 + display name), avatar URL, room-type ID + name, nights stayed, stay status, check-in / check-out month, hotel reply text, photo array ({ id, urls.thumbnail, urls.large, kind }), Booking review ID, and a deep-link reviewUrl anchored to the review on the property page.
• 26 review languages, regional variants included — English, German, French, Spanish, Italian, Dutch, Portuguese, Polish, Russian, Japanese, Korean, Chinese, Arabic, Turkish, Swedish, Norwegian, Danish, Finnish, Czech, Hungarian, Romanian, Greek, Hebrew, Thai, Vietnamese, Indonesian. Picking Spanish also brings back Mexican and Latin-American Spanish, English also brings back US English, Portuguese also brings back Brazilian Portuguese — variants Booking's own language menu leaves out. Multi-select for cross-market sentiment work; leave empty to collect every language.
• Partner reply capture, free — toggle Include Hotel Replies on (default) and every row carries the property's official partnerReplyText. Critical for reputation-management SaaS measuring response rate, response speed, and tone across a portfolio.
• Guest-uploaded photos in every row — Booking exposes original-resolution large and thumbnail URLs; both come back as a photos array per review for hotel-amenity computer-vision pipelines.
• Predictable cost ceiling per property — Max Reviews Per Property defaults to 1,000 (worst case $1.30 per property), with 0 opting in to "all available". Booking mega-properties can carry 10,000+ reviews; the cap stops cost runaways before they happen.
• Tolerant input parsing — handles plain URL strings, bare { "url": "..." } objects, and full upstream dataset rows with extra fields. Rows where url is missing or null are quietly skipped instead of crashing the run.

Use Cases

Hospitality Market Research

• Benchmark guest sentiment across an entire city's 4-star inventory.
• Compare review-language mix to validate which markets a property is winning.
• Quantify month-over-month review velocity and average score for a competitor set.

Reputation Management & Hotel SaaS

• Power dashboards that surface fresh negative reviews to ops teams within hours.
• Audit a chain's partner-reply coverage — measure response rate, lag, and tone.
• Feed verbatim "Liked" and "Disliked" text into NLS / topic-modeling pipelines.

Investment Due-Diligence

• Pull 10 years of reviews on an acquisition target to validate brand health.
• Detect renovation impact: split scores by check-in date, before vs after a known refurbishment window.
• Surface the top recurring complaints driving sub-7 scores in a target portfolio.

OTA & Meta-search Aggregation

• Refresh reviews nightly across a curated catalog of 10,000+ hotels.
• Augment an in-house listings index with multilingual review snippets.
• Cross-check Booking review counts against TripAdvisor / Google for inventory parity.

Trend & Sentiment Monitoring

• Track sentiment by season — compare summer vs winter review tone in seasonal markets.
• Compare solo-traveler vs family vs business guest segments on the same property.
• Monitor a single hotel's score distribution week-over-week to flag service drift.

Travel Editorial & Content

• Generate "What guests really say about Le Meurice" articles from verified Booking text.
• Curate "Top 25 family-friendly hotels in Lisbon" lists from filtered review datasets.
• Pull guest photos to enrich hotel review write-ups with real-stay imagery.

Getting Started

1. Minimal — one property, all defaults

Returns up to 1,000 most-relevant reviews from a single property, every language, every traveler type, every score, every season. Hotel replies are included.

{
    "startUrls": [
        { "url": "https://www.booking.com/hotel/fr/le-meurice.en-gb.html" }
    ]
}

2. Mid — newest negative reviews only

Triage the 100 most recent low-scoring reviews to find emerging service issues. Useful for ops dashboards and reputation alerts.

{
    "startUrls": [
        { "url": "https://www.booking.com/hotel/gb/the-savoy.en-gb.html" }
    ],
    "sortBy": "newest",
    "scoreFilter": "poor",
    "maxReviewsPerProperty": 100
}

3. Advanced — pipe the Hotel Scraper output, then narrow by language and traveler

Run the Booking.com Hotel Scraper over a destination first, then drop its dataset rows straight into this actor's startUrls. The url key on each row is read; every other field (rating, price, address, etc.) is ignored. Below, only English-speaking solo-traveler reviews are collected, capped at 200 per property.

{
    "startUrls": [
        { "url": "https://www.booking.com/hotel/gb/the-savoy.en-gb.html", "name": "The Savoy", "rating": 9.2 },
        { "url": "https://www.booking.com/hotel/fr/le-meurice.en-gb.html", "name": "Hotel Le Meurice", "rating": 9.0 },
        { "url": "https://www.booking.com/hotel/it/danieli.en-gb.html", "name": "Hotel Danieli", "rating": 8.9 }
    ],
    "languages": ["en"],
    "customerType": "solo",
    "maxReviewsPerProperty": 200,
    "includeReplies": true
}

Input Reference

Property URLs

Parameter	Type	Default	Description
startUrls	array	[ "https://www.booking.com/hotel/fr/le-meurice.en-gb.html" ]	Booking.com property URLs to scrape reviews from. Accepts plain URL strings, bare { "url": "..." } objects, OR rows from another actor's dataset (any object with a url key — every other field is ignored).

Pagination & Limits

Parameter	Type	Default	Description
maxReviewsPerProperty	integer	1000	Cap on reviews collected per property. Set to 0 to pull every available review (popular hotels can carry several thousand). Pricing is per review row, so this is your main cost lever.

Sorting & Filtering

Parameter	Type	Default	Description
sortBy	string	relevant	How Booking.com orders reviews before collection. Options: Most Relevant, Newest first, Oldest first, Highest score first, Lowest score first. Most Relevant is Booking.com's own ranking, centred on the property's overall guest rating, so a capped run clusters around that rating. Use Newest first or Oldest first for a representative cross-section (see Tips).
scoreFilter	string	all	Limit to one Booking score bucket. Guest scores are whole numbers 1–10, so the buckets never overlap. Options: All, Wonderful (9–10), Good (7–8), Okay (5–6), Poor (3–4), Very poor (1–2).
languages	array	[] (all)	Multi-select review languages. Choose any subset of: English, German, French, Spanish, Italian, Dutch, Portuguese, Polish, Russian, Japanese, Korean, Chinese, Arabic, Turkish, Swedish, Norwegian, Danish, Finnish, Czech, Hungarian, Romanian, Greek, Hebrew, Thai, Vietnamese, Indonesian. Each choice also brings back its regional variants — see the language field in the output table.
customerType	string	all	Filter by traveler segment. Options: All travelers, Families, Couples, Solo travelers, Groups of friends, Business travelers. Business travelers filters on the purpose of the trip; the customerType column in the output reports the guest group instead, so those rows come back as families / couples / solo / groups.
timeOfYear	string	all	Filter by the season the review was posted in, not the season of the stay. Options: Any time of year, March – May (spring), June – August (summer), September – November (fall), December – February (winter).

Output Options

Parameter	Type	Default	Description
includeReplies	boolean	true	Attach the property's official reply text to each review when available. Adds partnerReplyText at no extra cost, plus the companion partnerReplyDate column that Booking.com currently leaves empty.

Output

One row per review, flat schema for clean dataset table display.

Example

{
    "hotelUrl": "https://www.booking.com/hotel/fr/le-meurice.en-gb.html",
    "hotelId": 55361,
    "hotelName": "Hotel Le Meurice",
    "hotelCountryCode": "fr",
    "reviewId": "6340bcc03d6b080b",
    "reviewUrl": "https://www.booking.com/hotel/fr/le-meurice.en-gb.html#tab-reviews?review=6340bcc03d6b080b",
    "reviewDate": "2026-04-18T15:17:40Z",
    "title": "Excellent value for your money",
    "positiveText": "Location is great, the staff were extremely helpful and the room was spotless.",
    "negativeText": "There aren't any negatives for me staying here.",
    "score": 10,
    "language": "en",
    "helpfulVotesCount": 4,
    "isApproved": true,
    "customerType": "solo",
    "anonymous": false,
    "guestName": "Carollyn",
    "guestCountryCode": "us",
    "guestCountryName": "United States",
    "avatarUrl": "https://q-xx.bstatic.com/xdata/avatars/...",
    "roomTypeId": "5536114",
    "roomTypeName": "Standard Double Room",
    "numNights": 2,
    "stayStatus": "stayed",
    "checkInDate": "2026-04-12",
    "checkOutDate": "2026-04-14",
    "partnerReplyText": "Thank you for your wonderful review, Carollyn — we look forward to welcoming you back to Le Meurice.",
    "partnerReplyDate": null,
    "photos": [
        {
            "id": "12345678",
            "urls": {
                "thumbnail": "https://q-xx.bstatic.com/xdata/images/xphoto/square80/12345678.jpg",
                "large": "https://q-xx.bstatic.com/xdata/images/xphoto/max1280x900/12345678.jpg"
            },
            "kind": "guest"
        }
    ],
    "scrapedAt": "2026-05-01T19:16:14Z"
}

Hotel Context Fields

Field	Type	Description
hotelUrl	string	The property URL exactly as you submitted it, so rows join straight back onto your input list.
hotelId	integer	Booking.com internal property ID.
hotelName	string	Hotel display name.
hotelCountryCode	string	ISO-2 country code (e.g. fr, gb, it).

Core Review Fields

Field	Type	Description
reviewId	string	Booking-side review identifier. Stable across runs.
reviewUrl	string	Deep-link that opens this review on the property's own page. When you submit an older Booking.com address for a property, the link points at the page that address now leads to.
reviewDate	string (ISO 8601)	When the review was submitted to Booking.
title	string	Review headline written by the guest.
positiveText	string	"Liked" body.
negativeText	string	"Disliked" body.
score	integer (1–10)	Reviewer's overall score for the stay. Booking asks each guest for a single whole-number rating from 1 to 10, so this is never a decimal. The decimal scores shown on Booking (for example 8.7) are the property average across all reviews, not one guest's score.
language	string	Two-letter code for the language the review was written in (en, de, fr, ...). Booking marks regional variants with its own codes instead: xu US English, xb Brazilian Portuguese, xa Latin-American Spanish, xm Mexican Spanish, xt Traditional Chinese. These arrive with the parent language you filtered on, so a run filtered to English returns both en and xu rows.
helpfulVotesCount	integer	Number of "Helpful" upvotes.
isApproved	boolean	Booking-side moderation flag.

Guest Fields

Field	Type	Description
customerType	string	Guest group Booking recorded for the stay: families, couples, solo, or groups. This is a different question from the customerType input filter, which selects on trip purpose, so this column never reads business.
anonymous	boolean	Whether the guest chose to remain anonymous.
guestName	string	Display name (or Anonymous when anonymous=true).
guestCountryCode	string	Reviewer's home-country ISO-2 code.
guestCountryName	string	Reviewer's home-country display name.
avatarUrl	string | null	Reviewer avatar URL when public.

Stay Fields

Field	Type	Description
roomTypeId	string	Booking room-type ID for the booked stay.
roomTypeName	string	Human-readable room type (e.g. "Deluxe Double Room").
numNights	integer	Length of stay in nights.
stayStatus	string	stayed, cancelled, etc. (Booking taxonomy).
checkInDate	string (YYYY-MM-DD)	Date the guest checked in.
checkOutDate	string (YYYY-MM-DD)	Date the guest checked out.

Hotel Reply, Photos & Metadata

Field	Type	Description
partnerReplyText	string | null	Hotel's official response to the review when present. null if no reply or includeReplies=false.
partnerReplyDate	string | null	Reply timestamp. Booking.com exposes the field but has left it empty on every review measured, so treat this column as reserved: it fills itself if Booking ever starts publishing reply dates.
photos	array of { id, urls: { thumbnail, large }, kind }	Guest-uploaded photos. Empty array when none.
scrapedAt	string (ISO 8601)	UTC timestamp of extraction.

Tips for Best Results

• Pipe the Booking.com Hotel Scraper output here directly. Run a destination + filter search in the Hotel Scraper, then drop the output dataset URL straight into startUrls in this actor — every row's url is read automatically; everything else is ignored. No glue script needed for end-to-end "destination → reviews" pipelines.
• Sort order decides which slice of the score range you get, so pick it before you cap. Booking.com's "most relevant" ranking is centred on the property's own overall rating, which means a capped run keeps the reviews nearest that score and drops both ends. Measured at a 300-review cap: relevant returned just 3 of the 10 possible scores on Atlantis The Palm (8/9/10), Park Plaza Westminster Bridge (7/8/9) and Excalibur Las Vegas (7/8/9), with no review below 7 anywhere and no perfect 10 on the last two. newest on the same three properties at the same cap returned all 10 scores, including 20, 39 and 60 reviews below 7. Use relevant for the reviews Booking.com itself surfaces first, newest for score histograms, complaint mining and sentiment work, score_asc to triage the worst-rated reviews for a reputation dashboard, and a cap of 0 when you want the complete distribution.
• For multi-language sentiment work, run separate jobs per language and join on reviewId. This keeps each dataset narrowly scoped and avoids paying for languages you won't analyze. Group each language with its variants when you tally the results.
• Cap maxReviewsPerProperty at 1,000 unless you genuinely need every review. Mega-properties on Booking can carry 10,000+ reviews — at $1.30 per 1,000 that's $13 per property if uncapped. The default 1,000 keeps spend predictable.
• Combine scoreFilter=very_poor with customerType=business to surface the highest-impact complaints from corporate guests — these reviews often drive the biggest churn risk for hotel sales teams. The rows come back tagged with the guest group (couples, solo, and so on), because that column answers a different question than the filter does.
• A whole-number score is correct, not a rounding error. Booking collects one 1–10 rating per guest. Averaging score across a property gives you the decimal figure Booking displays on the listing.
• Don't discard the xu, xb, xa, xm and xt rows. They are Booking's codes for US English, Brazilian Portuguese, Latin-American Spanish, Mexican Spanish and Traditional Chinese. On a Las Vegas property, US English alone was 15,611 of 31,018 reviews, so a pipeline that keeps only en throws away most of its English data. Map them onto the parent language before you aggregate.
• Measure reply coverage from partnerReplyText, not from a date. Booking publishes the reply itself but not when it was written, so partnerReplyDate arrives empty; response-rate reporting works fine off the text, response-speed reporting does not.

Pricing

$1.30 per 1,000 reviews — flat per-result price.

No compute charges — you only pay per review returned.

Reviews	Cost
100	$0.13
1,000	$1.30
10,000	$13.00
100,000	$130.00

One dataset row = one review = one charge. maxReviewsPerProperty is a per-property cap, so a run over 20 property URLs at the default 1,000 can return 20,000 rows; set the cap to the number you actually want per hotel, and use the scoreFilter / languages / customerType / timeOfYear filters to narrow the volume before the run starts.

Integrations

Export data in JSON, CSV, Excel, XML, or RSS. Connect to 1,500+ apps via:

• Zapier / Make / n8n — Workflow automation
• Google Sheets — Direct spreadsheet export
• Slack / Email — Notifications when new reviews land
• Webhooks — Trigger custom APIs on run completion
• Apify API — Full programmatic access from any language

Legal & Ethical Use

This actor is designed for legitimate market research, hospitality intelligence, reputation management, investment analysis, and travel content creation. It collects only publicly visible review content from Booking.com property pages. Users are responsible for complying with applicable laws — including data-protection regimes that govern personal information in guest names and avatars — and with Booking.com's Terms of Service. Do not use the data to harass individuals, build replica review platforms, generate spam, or for any unlawful purpose.