Facebook Reviews Scraper

Every review on any public Facebook page, newest first — the text, whether the reviewer recommends the page or not, Facebook's review tags, the reviewer's name and profile, the date, reactions and comments — plus the page's overall recommend percentage on every row. Follows the cursor for all of them. No login, no cookies, no browser.

Why Use This Scraper?

• All of them, not the first page. Facebook server-renders three reviews and loads the rest on scroll. Reading the page and stopping is why other review scrapers return "only the first 2 of 59". This one walks Facebook's own pagination query until it has what you asked for — measured 30 of 30 on a page with 201, 30 of 30 on a page with 76, ids unique, dates back to 2019.
• Recommends or doesn't, on every row. Facebook replaced star ratings with a recommend verdict in 2018. recommends is true/false on every review, from Facebook's own rendered verdict.
• Facebook's review tags. Child-friendly, Fast delivery, Cosy atmosphere — the structured tags reviewers pick, where present.
• The page's own summary on each row. 94% recommend (201 reviews), parsed into pageRecommendPercent and pageReviewCount, so a batch across pages needs no second lookup.
• Stop at a date. onlyReviewsNewerThan halts paging at the first older review, so you are not charged for history you did not ask for.
• A page with no reviews says so. no_reviews, not_found, login_wall — as a row, not charged, rather than a silently shorter dataset.

Overview

One row per review, 23 fields. Ask for 50 reviews across three pages and you get up to 150 rows, each carrying pageUrl, pageName and inputUrl.

Supported Inputs

Input	Example
Reviews tab URL	https://www.facebook.com/copperkettleyqr/reviews
Page URL	https://www.facebook.com/copperkettleyqr
Vanity name	copperkettleyqr — in the pages field
Numeric id	100064027242849 — in the pages field, or profile.php?id=… as a URL
Any host or tab	m.facebook.com, /about, /photos — normalised

Use Cases

• Reputation monitoring — every negative review on your own or a client's pages, with the text and the date, on a schedule.
• Competitor research — what customers praise and complain about across a set of local businesses.
• Lead qualification — recommend rate and review volume for every page a search returned.
• Sentiment datasets — review text with a binary label, tags and engagement, at scale.

How It Works

Each input is normalised to the page and its About tab is streamed — about 100 KB — for the page id, name, category and the overall recommend summary. The reviews connection is then paged through Facebook's own refetch query, three reviews per request, following the cursor until resultsLimit, the date cut-off, or the end. Responses are the multi-line format Facebook streams to its own client — the first review on line one, the rest as their own lines, the cursor last — and are read as such. The recommend verdict is Facebook's rendered title for the review, pinned to English by the request's Accept-Language.

Pure HTTP over a residential proxy. No browser, no login.

Input Configuration

Field	Type	Default	Notes
startUrls	array	—	Page or reviews-tab URLs.
pages	array	—	Vanity names or numeric ids. Either field, or both; de-duplicated on the page.
resultsLimit	integer	50	Reviews per page, newest first.
onlyReviewsNewerThan	string	—	YYYY-MM-DD. Paging stops at the first older review.
language	string	en-US	Accept-Language for the page fetch. The verdict is always read in English.
proxy	object	—	Non-paying sample runs only. Paying runs use a built-in residential proxy.

Example input

{
  "startUrls": [{ "url": "https://www.facebook.com/copperkettleyqr/reviews" }],
  "resultsLimit": 100,
  "onlyReviewsNewerThan": "2025-01-01"
}

Output Overview

One flat row per review. A page that yields none produces { inputUrl, error, message } and is not charged.

Output Sample

A real row from a live run — a negative review, chosen on purpose:

{
  "reviewId": "10159367163365044",
  "url": "https://www.facebook.com/jayna.nicholson/posts/pfbid0FDW1nctxFdgK9T2Rzx9Vannn68uxzLE4WRhfGLgvL98AdEjrhJKX42BfeFTDHPPhl",
  "text": "Normally the food from here is really good but we got pizza last night and it was somewhere in the range of sub par to just really not good. The whole thing was barely cooked and was so soggy. The crust was literally wet. Maybe it’s because it’s the deluxe 🤷‍♀️ \nAnyway, I don’t think we’ll be getting pizza from here again. Definitely not worth the price.",
  "recommends": false,
  "verdictText": "Jay Enn doesn't recommend The Copper Kettle Restaurant.",
  "tags": [],
  "createdAt": "2024-04-11T17:40:55.000Z",
  "createdTimestamp": 1712857255,
  "reviewerName": "Jay Enn",
  "reviewerId": "pfbid0CdLWRx8qk8cuqmSdwKbWkZ72kYoKST7ZefipmWrcdRTGvnN6MmX9qZHpUU9kqUE6l",
  "reviewerUrl": null,
  "reactionsCount": 1,
  "commentsCount": 3,
  "photoUrls": [],
  "pageUrl": "https://www.facebook.com/copperkettleyqr",
  "pageName": "The Copper Kettle Restaurant",
  "pageId": "100064027242849",
  "pageCategory": "Pizza place",
  "pageRatingText": "94% recommend (201 reviews)",
  "pageRecommendPercent": 94,
  "pageReviewCount": 201,
  "inputUrl": "https://www.facebook.com/copperkettleyqr/reviews",
  "scrapedAt": "2026-09-17T12:46:30.484Z"
}

Key Output Fields

Field	Notes
reviewId / url	Facebook's post id for the review and its permalink.
text	The review.
recommends	true / false, from Facebook's verdict. verdictText is the sentence it came from.
tags	Facebook's review tags where the reviewer picked any — Child-friendly, Fast delivery. Empty otherwise.
createdAt / createdTimestamp	ISO 8601 and epoch seconds.
reviewerName / reviewerId / reviewerUrl	The reviewer. The URL is built from the id when Facebook lists no vanity name.
reactionsCount / commentsCount	Engagement on the review.
photoUrls	Photos attached to the review, signed CDN links.
pageName / pageId / pageUrl / pageCategory	The page.
pageRatingText / pageRecommendPercent / pageReviewCount	94% recommend (201 reviews) and its parts.
error / message	Present only on a page-level row: no_reviews, no_reviews_in_range, not_found, login_wall, blocked, invalid_url.

FAQ

Do I need a Facebook account or cookies? No. Every figure in this README came from logged-out requests with no cookies of any kind.

Why is there no star rating? Facebook removed stars in 2018. A review is a recommend or a don't-recommend, and that is what recommends carries. pageRecommendPercent is the page-level equivalent of a rating.

Can I get every review on a page? Yes — set resultsLimit above the page's pageReviewCount. Facebook serves three per request, so a page with 200 reviews takes about 70 requests, which is a minute or two.

Why do some reviews have no tags or photoUrls? Most reviewers don't add them. Empty arrays mean none were attached, not that extraction failed.

Am I charged for a page with no reviews? No. You get one row saying so, and it is free.

Support

Open an issue on the Actor's Issues tab. Bug reports with a run id and the page URL get looked at first.

Explore More Scrapers

• Facebook Page Contact Details — email, phone, address, hours for any page
• Facebook Pages & Contacts — find pages by keyword and city, with contacts
• Facebook Comments Scraper — comment threads on any public post
• Facebook Reels Scraper — every reel on a page with play counts

🤖 For AI Agents & LLM Apps

Purpose: Return every review on public Facebook pages — text, a binary recommend label, tags, reviewer, date, engagement — with the page's overall recommend rate, from page URLs (startUrls) or vanity names and ids (pages).

Minimal tested input:

{ "startUrls": [{ "url": "https://www.facebook.com/copperkettleyqr/reviews" }], "resultsLimit": 30 }

Output: array of flat objects, one per review. Fields: reviewId, url, text, recommends, verdictText, tags, createdAt, createdTimestamp, reviewerName, reviewerId, reviewerUrl, reactionsCount, commentsCount, photoUrls, pageUrl, pageName, pageId, pageCategory, pageRatingText, pageRecommendPercent, pageReviewCount, inputUrl, scrapedAt.

Agent-relevant behaviour: recommends is the label; there is no numeric rating per review on Facebook. Rows arrive newest first per page. A row carrying error is a page-level failure, not a review. pageReviewCount is Facebook's own total, so resultsLimit >= pageReviewCount means all of them. photoUrls expire within hours.

Billing: per review row. Page-level failure rows are free.

⚠️ Disclaimer

This Actor reads only publicly available Facebook reviews — no login, no cookies, no private content. Reviews are their authors' words. You are responsible for using the output lawfully, including under the GDPR, the CCPA and Facebook's terms where they apply to you. Nothing here is legal advice.

SEO Keywords

facebook reviews scraper, scrape facebook reviews, facebook page reviews, facebook recommendations scraper, facebook business reviews, facebook review data, facebook reviews export, facebook reviews to csv, facebook reviews api, facebook reviews sentiment, facebook negative reviews, facebook page rating, facebook review monitoring, facebook reviews no login, apify facebook reviews scraper