Agoda Scraper

Scrape Agoda accommodation reviews from a hotel URL and export Agoda review data to JSON, CSV, or Excel. This Agoda API alternative collects public ratings, review text, stay details, reviewer labels, management replies, and the score breakdowns shown on the accommodation page.

Give the actor one accommodation URL, choose the review providers and optional filters, and set a result limit. Each dataset row represents one review. The fields are stable across runs, so you can compare new records with earlier exports by accommodation ID, provider ID, and review ID.

What data can you extract from Agoda?

The output is one flat record per public review. Dates retain the timestamp and offset returned by the site. A value can be null when that review or accommodation does not expose it.

Field	Description	Example
hotelId	Numeric accommodation ID	49487
hotelName	Accommodation name	Ibis Budapest City
hotelUrl	Accommodation page URL	https://www.agoda.com/ibis-budapest-city/hotel/budapest-hu.html
reviewPageUrl	Public reviews page URL	https://www.agoda.com/ibis-budapest-city/reviews/budapest-hu.html
hotelReviewId	Review ID	1064241007
rating	Review rating on the displayed 0–10 scale	8.8
ratingText	Displayed rating label	Excellent
reviewTitle	Review headline	Excellent
reviewComments	Main comment section	One of the staff went out of the way...
reviewPositives	Positive feedback section	Comfortable room and helpful service
reviewNegatives	Negative feedback section	Slow check-in
reviewText	Nonempty comment sections joined into one text value	Overall, pleasant service!
reviewDate	Review publication timestamp	2026-01-04T10:11:00+07:00
checkInDate	Stay check-in timestamp	2025-12-27T00:00:00+07:00
checkOutDate	Stay check-out timestamp	2025-12-31T00:00:00+07:00
reviewerCountryName	Publicly displayed country	Singapore
reviewerName	Publicly displayed reviewer label	Amalina
reviewerGroupName	Traveler group	Group
reviewerRoomTypeName	Room type shown with the review	Room with 2 single beds
reviewerLengthOfStay	Stay length in nights	4
reviewProviderText	Displayed source of the review	Agoda
reviewProviderId	Numeric provider ID	332
reviewLanguage	Original review language code	en
responseDate	Management response timestamp	2026-01-24T13:01:06+07:00
responderName	Public responder label	Ibis Budapest City
responseText	Management reply	Thank you very much for your review...
originalTitle	Original headline when a translation is shown	null
originalComment	Original comment when a translation is shown	null
helpfulVotes	Helpful vote count	0
agodaAggregateReviewScore	Accommodation score from Agoda reviews	8.4
agodaAggregateDetailedReviewScore	Agoda category IDs and scores	cleanliness: 8.7
bookingAggregateReviewScore	Accommodation score from Booking.com reviews	8.7
bookingAggregateDetailedReviewScore	Booking.com category IDs and scores	location: 9.0
agodaReviewsCount	Agoda review total displayed on the page	174
bookingReviewsCount	Booking.com review total displayed on the page	1758
combinedReviewScore	Combined score displayed on the page	8.6
combinedReviewsCount	Combined review count displayed on the page	2073

The two detailed score fields contain arrays of objects. Each object has an ID, such as cleanliness, facilities, location, roomComfort, staffPerformance, or valueForMoney, and a numeric score. Categories differ by property and provider. The count fields describe the accommodation page, not the number of rows in one run.

How to scrape Agoda reviews

1. Open a public accommodation page and copy its URL. A normal URL contains /hotel/ and ends in .html. A public /reviews/ URL for the same property also works.
2. Paste that URL into Accommodation URL. Choose the maximum number of reviews. The default is 100; a smaller number is useful for a first export.
3. Keep both review providers selected, or select the provider you need. Add rating, language, and date filters if the downstream report needs a narrower set.
4. Run the actor. The dataset contains one row for each accepted review, up to the result limit and the records the public review service makes available.
5. Download the dataset in the format your workflow uses. Keep the review ID and provider ID when storing the records so repeated runs can be deduplicated.

The input accepts one property at a time. To monitor a portfolio, run it once per accommodation URL and combine the exported rows. Keeping each run tied to a single property also makes failures and unusually low result counts easier to investigate.

Input

Parameter	Type	Default	Description
hotelUrl	String, required	None	Public Agoda accommodation or reviews URL.
maxReviewsPerHotel	Integer	100	Maximum rows to return. Zero continues until pagination ends, within a 100-page safety limit.
enabledProviders	Array	Agoda and Booking.com	Providers whose review rows may be returned.
rating	Array	Empty	Rating bands to keep: exceptional, excellent, very_good, good, below_expectation. Empty means all.
language	String	all	Original language code, such as en, de, or ko; all keeps every language.
cutoffDate	Date string	Empty	Keep reviews later than this YYYY-MM-DD date. The date itself is excluded.
sortBy	String	recent	recent, ratingHigh, ratingLow, or helpful.
aggregateRatings	Boolean	true	Include accommodation-level scores, breakdowns, and counts in every row.
proxyConfiguration	Object	Hosted proxy	Connection settings for the hosted run.

A realistic request for feedback from both displayed providers is:

{
  "hotelUrl": "https://www.agoda.com/ibis-budapest-city/hotel/budapest-hu.html",
  "maxReviewsPerHotel": 80,
  "enabledProviders": ["Agoda", "Booking.com"],
  "sortBy": "recent",
  "language": "all",
  "aggregateRatings": true
}

Rating bands use the review's numeric score: exceptional is 9 or above; excellent is 8 to below 9; very_good is 7 to below 8; good is 6 to below 7; and below_expectation is below 6. Multiple selected bands are combined. Filters are applied to the returned records, so a request for a narrow band may yield fewer rows than the maximum even when the property has more reviews overall.

The language setting checks the original language reported with the review. A translated comment can therefore appear in English while its original language is another code. Use originalComment and originalTitle when the site supplies them and the original wording matters.

Output example

This item is from a successful hosted run for the example accommodation. The management reply and category arrays are shortened here; the dataset row contains their complete values.

{
  "hotelId": 49487,
  "hotelName": "Ibis Budapest City",
  "hotelUrl": "https://www.agoda.com/ibis-budapest-city/hotel/budapest-hu.html",
  "reviewPageUrl": "https://www.agoda.com/ibis-budapest-city/reviews/budapest-hu.html",
  "hotelReviewId": 1064241007,
  "rating": 8.8,
  "ratingText": "Excellent",
  "reviewTitle": "Excellent",
  "reviewComments": "One of the staff went out of the way to attend to our queries. Overall, pleasant service! ",
  "reviewPositives": null,
  "reviewNegatives": null,
  "reviewText": "One of the staff went out of the way to attend to our queries. Overall, pleasant service!",
  "reviewDate": "2026-01-04T10:11:00+07:00",
  "checkInDate": "2025-12-27T00:00:00+07:00",
  "checkOutDate": "2025-12-31T00:00:00+07:00",
  "reviewerCountryName": "Singapore",
  "reviewerName": "Amalina",
  "reviewerGroupName": "Group",
  "reviewerRoomTypeName": "Room with 2 single beds equipped with the new b edding",
  "reviewerLengthOfStay": 4,
  "reviewProviderText": "Agoda",
  "reviewProviderId": 332,
  "reviewLanguage": "en",
  "responseDate": "2026-01-24T13:01:06+07:00",
  "responderName": "Ibis Budapest City",
  "responseText": "Thank you very much for your excellent review...",
  "originalTitle": null,
  "originalComment": null,
  "helpfulVotes": 0,
  "agodaAggregateReviewScore": 8.4,
  "agodaAggregateDetailedReviewScore": [{"id": "cleanliness", "score": 8.7}],
  "bookingAggregateReviewScore": 8.7,
  "bookingAggregateDetailedReviewScore": [{"id": "location", "score": 9.0}],
  "agodaReviewsCount": 174,
  "bookingReviewsCount": 1758,
  "combinedReviewScore": 8.6,
  "combinedReviewsCount": 2073
}

The hosted check returned 80 rows for this accommodation. All 80 had an accommodation ID, review ID, rating, publication date, displayed provider, reviewer label, accommodation name, and combined review text. These are observations from that run, not guarantees for every property or future date.

Use cases

• A hotel operations team can review recent low scores and their associated comment text before deciding which service issue needs attention. The response fields show whether a public reply is already present.
• A travel researcher can compare public satisfaction across properties using the same numeric rating scale, while preserving provider IDs to avoid mixing sources without attribution.
• A reputation analyst can record daily review IDs, count newly observed comments, and separate new negative feedback from older records.
• A property manager can compare category scores for cleanliness, location, facilities, service, and value for money with the text of individual stays.
• A localization team can review original language codes and available original wording when translated comments are shown in a common display language.
• A data team can export tidy rows into a warehouse, join them to an accommodation table by hotelId, and use hotelReviewId with reviewProviderId as a repeat-run key.

Review data is best interpreted as customer feedback. A high score does not establish an objective measure of property quality, and the aggregate scores can change as new reviews are added or the site changes its display logic.

How much does it cost to scrape Agoda?

The configured result price is $1.00 per 1,000 delivered reviews, or $0.001 per dataset row. The charge is based on rows delivered, so a limit of 100 is an upper bound of $0.10 for that run. A run that returns fewer rows has a lower result charge. Platform usage is included in the result price.

Delivered reviews	Result charge
80	$0.08
500	$0.50
1,000	$1.00
5,000 across multiple runs	$5.00

These examples are arithmetic, not a promise that every property exposes that many records. In the hosted check, the 80-row request delivered 80 rows. If you are building a large collection, start with a smaller limit on a representative property and inspect both the row count and the provider mix before setting larger limits.

Tips for best results

Use the property's public accommodation URL, including its city path and .html suffix. Search result pages and account pages are not accepted as accommodation URLs. The input removes query parameters, so referral codes and tracking fragments are unnecessary.

For regular monitoring, sort by most recent and save the last seen review IDs. The cutoffDate input can narrow results to reviews after a specific day, but the service may interleave review providers. Do not assume that one provider's first row is the newest record across every source. Compare dates in your stored data when building a single timeline.

If you only need one displayed provider, select it in enabledProviders. The actor still requests the public review feed and filters records before output, so a single-provider request may inspect pages containing records from other providers. A small result limit may therefore take more than one page to fill.

Use reviewText for full-text indexing because it includes the nonempty main, positive, and negative sections. Keep the separate sections if your analysis distinguishes praise from complaints. The site sometimes stores a Booking.com review mainly in positive and negative fields while leaving reviewComments empty.

Enable aggregateRatings when a report needs accommodation-level context next to each comment. Disable it when you do not need score values. The score fields become null in that case. Counts and scores refer to the page snapshot at run time and may change independently of individual review records.

Limits

The actor reads public review data for one accommodation per run. It does not read booking accounts or private guest data. Some accommodations have no public comments, and some reviews omit a stay date, room name, country, response, or original-language text. Missing values are returned as null where possible.

The displayed total review count can be much larger than the rows exposed by the public paginated endpoint. On the tested accommodation, the page displayed more than two thousand combined reviews, while a limited request for 80 records succeeded. A probe of later pages returned empty responses, so a high maxReviewsPerHotel value must be treated as a cap, not an assurance that the cap is reached. The actor stops when the endpoint returns no comments or no new review IDs and has a 100-page safety boundary.

Rating, language, date, and provider filters are applied after pages are fetched. If the matching share is small, the result count can be below the requested maximum. Changing the sort order can expose a different subset, but it is not guaranteed to unlock all historical reviews.

A successful HTTP response can occasionally contain no comments even when the property has reviews. The actor retries an empty page with a fresh connection. If the first page is still empty, it checks a second public review endpoint and then rotates the proxy connection once before failing. A property whose page explicitly lists zero reviews can finish with an empty dataset. An empty response for a property that lists reviews is not silently reported as a completed scrape.

Automated workflows

A scheduled run can write each new review to a review table keyed by hotelId, reviewProviderId, and hotelReviewId. A second process can compare reviewDate against the previous run, route low ratings for manual review, and retain the unmodified text fields for auditability. Use the dataset export endpoint when your workflow needs JSON or CSV, and use a webhook to react when a run finishes.

For a portfolio, keep a list of accommodation URLs in your own workflow and start one run for each URL. Record the input URL, start time, run status, and delivered row count. Those four values make it easier to detect a property whose public page changed or whose review feed temporarily returned fewer records.

FAQ

Can I scrape Agoda reviews without a login?

Yes. This actor uses public accommodation and review pages. It does not ask for account credentials, booking references, or private booking information. The hotel URL must be publicly reachable.

Can I include both Agoda and Booking.com reviews?

Yes. Both displayed providers are selected by default. Each row includes reviewProviderText and reviewProviderId so you can separate them later. The provider totals in the aggregate fields are page-level counts, while the dataset contains only records returned during this run.

Why is reviewComments empty on some records?

Some reviews put their feedback in reviewPositives and reviewNegatives. reviewText combines available sections for a convenient searchable field. Keep the separate fields if the distinction matters to your analysis.

Why did I receive fewer reviews than maxReviewsPerHotel?

The maximum is a stop limit. Public pagination can end before it is reached, and optional filters discard nonmatching records. A property can also have fewer public comments than its displayed aggregate review count. Check the provider mix, try another sort order, or use a less restrictive filter when you need a broader sample.

What does language filter?

It checks the original language code attached to the review. A comment can be displayed in translation while reviewLanguage still records the source language. Original wording is included only when the site provides originalTitle or originalComment.

Can I schedule exports or open the results in a spreadsheet?

Yes. Schedule runs at the interval you need, then download the dataset as CSV or Excel. Retain the two review identifiers and the accommodation identifier so later exports can be merged without double-counting.

Is it legal to use the data?

The actor reads public information. You are responsible for checking the site's terms, applicable privacy rules, and the permitted use of review and reviewer data in your jurisdiction. Avoid using public reviewer labels to identify or contact individuals.

Do I need to configure a proxy?

The hosted run uses its configured connection automatically. Most users can keep the default. If a page or endpoint temporarily returns no records, the actor has a first-page recovery path and reports a failure when recovery does not produce records.

Integrations

Export the default dataset to JSON, CSV, or Excel; retrieve it through the dataset API; or trigger a downstream workflow when a run finishes. Scheduling supports recurring monitoring. For spreadsheet reports, keep one sheet for raw rows and another for property-level summaries so repeated aggregate scores are not counted as separate accommodations.

Support

Open an issue on the Issues tab; we reply within 24h and add fields on request.