Facebook Business Page Recommendation & Review Scraper

Get Facebook business page reviews by direct URL, business name search, or Facebook Page ID — no login or cookies required. Returns review text, Recommended/Not Recommended status, author details, reaction breakdowns, structured recommendation tags, business owner replies, and page-level aggregate stats like follower count and overall recommendation percentage.

What this actor does

• Scrapes reviews from any public Facebook business page — restaurants, hotels, retailers, airlines, and services
• Three input modes: paste a page or /reviews URL, search by business name + city, or supply a Facebook Page ID (numeric) or page username
• Filters results by date range, recommendation status (Recommended / Not Recommended), or keyword in the review text
• Extracts review photos, structured recommendation tags (e.g. "Outdoor dining", "Fast delivery"), reaction breakdowns (Like/Love/Wow/Haha/Sad/Angry/Care counts), and top comments including business/owner replies
• Attaches page details to each review — page name and ID, plus follower count, total review count and recommendation percentage when Facebook exposes them
• Empty/unavailable fields are omitted from each record rather than returned as null

Output

One record per review. Fields are only present when Facebook actually has that data — see the FAQ for why some fields (like reviewPhotos) are frequently absent.

Field	Example
reviewId	"UzpfSTEwMDAy..."
text	"Amazing food and great service!"
isRecommended	true
reviewTags	["Outdoor dining", "Fast delivery"]
createdAt	"2024-03-15T10:22:00+00:00"
timestamp	1710498120
authorName	"Julia Odan"
authorProfileUrl	"https://www.facebook.com/julia.odan"
authorProfileImageUrl	"https://scontent.fbcdn.net/..."
authorIdType	"pfbid" or "numeric" — see FAQ
authorReviewCount	12 — only on reviews delivered in Facebook's older review format
reactionCount	3
reactionBreakdown	[{"type": "Like", "count": 4}, {"type": "Haha", "count": 1}]
shareCount	0
commentCount	2
reviewPhotos	["https://scontent.fbcdn.net/..."]
topComments	[{"authorName": "McDonald's", "text": "Thanks for the feedback!", "createdAt": "2024-03-16T09:00:00+00:00", "isOwnerReply": true, "authorIdType": "numeric", "commentPermalinkUrl": "https://facebook.com/...", "replyCount": 2}]
reviewUrl	"https://www.facebook.com/.../posts/..."
page.name	"McDonald's"
page.id	"100064458773195"
page.url	"https://www.facebook.com/McDonalds" — when available
page.category	"Fast Food Restaurant" — present only when Facebook's page data exposes it
page.address	"110 N Carpenter St, Chicago, IL" — present only when Facebook's page data exposes it
page.phone	"+1 630-623-6196" — present only when Facebook's page data exposes it
page.website	"https://www.mcdonalds.com" — present only when Facebook's page data exposes it
page.totalReviewCount	4821 — when available
page.recommendationPercentage	87 — when available
page.followerCount	82000000 — read from the follower label Facebook displays (e.g. "82M"), so it is rounded the same way
recordType	"review"
scrapedAt	"2026-09-02T12:00:00+00:00"

The dataset can also contain error records with "recordType": "error", holding only error (a description) and scrapedAt. One is written when an input item returns no reviews at all or still fails after three attempts, and the run writes a single error record and stops when no input items are given for the chosen mode, the dates are invalid, or the proxy can't be set up. If reviews were fetched but your filters matched none of them, no error record is written.

Input

Field	Type	Default	Description
mode	string	url	url / search / pageId — how to specify the Facebook page
startUrls	array	[]	Facebook page URLs (mode=url); /reviews is added automatically when missing
searchQueries	array	[]	Business name + location, e.g. "Starbucks Dallas TX" (mode=search)
pageIds	array	[]	Numeric Facebook Page IDs, e.g. "100064458773195", or page usernames such as "mcdonalds" (mode=pageId)
maxReviews	integer	100	Maximum total reviews across all input pages (1–500)
maxReviewsPerPage	integer	–	Cap per individual page, for an even spread across multiple pages (1–500). When empty, only maxReviews applies
sortBy	string	most_recent	most_recent keeps the page's default order; most_helpful is attempted but usually has no effect (see FAQ)
filterByRecommendation	string	all	all / recommended / not_recommended
filterByDateFrom	string	–	Only reviews on or after this date (YYYY-MM-DD). Reviews without a date are kept
filterByDateTo	string	–	Only reviews on or before this date (YYYY-MM-DD)
filterByKeyword	string	–	Only reviews containing this word or phrase (case-insensitive)
cookies	array	[]	Optional logged-in session cookies to unlock more than ~10-15 reviews per page
proxyCountry	string	US	Exit country of the built-in residential proxy: US / GB / CA / AU / IE

Example: scrape reviews from a direct URL

{
  "mode": "url",
  "startUrls": [{ "url": "https://www.facebook.com/mcdonalds/reviews" }],
  "maxReviews": 100
}

Example: find a business by name and city

{
  "mode": "search",
  "searchQueries": ["Starbucks Seattle WA"],
  "maxReviews": 50
}

Example: only negative reviews from the last 90 days

{
  "mode": "url",
  "startUrls": [{ "url": "https://www.facebook.com/IHOP/reviews" }],
  "filterByRecommendation": "not_recommended",
  "filterByDateFrom": "2026-06-01",
  "maxReviews": 100
}

Example: multiple pages by numeric Page ID, capped per page

{
  "mode": "pageId",
  "pageIds": ["100064458773195", "100044057870366"],
  "maxReviewsPerPage": 25,
  "maxReviews": 50
}

Use Cases

• Reputation monitoring — track new Recommended/Not Recommended reviews and business owner replies for a brand or franchise location
• Competitor analysis — compare recommendation percentages, review volume, and recurring complaint keywords across competing businesses
• Local business research — pull reviews for restaurants, hotels, or service providers before choosing where to eat, stay, or book
• Customer feedback analysis — export review text and recommendation tags (e.g. "Fast delivery", "Disappointing food") for sentiment or NLP pipelines
• Franchise/multi-location tracking — scrape reviews across many location-specific Page IDs in a single run to compare performance
• Market research — measure follower counts and recommendation trends across an industry vertical

FAQ

Does this require a Facebook account or cookies? No. The actor scrapes publicly visible reviews on public business pages without logging in. Cookies are an optional input for unlocking more than the ~10-15 reviews per page that Facebook shows to logged-out visitors.

Is this an official Facebook or Meta product? No. This is an independent, third-party actor that reads publicly available Facebook business page data. It is not affiliated with, endorsed by, or connected to Meta Platforms, Inc.

Why does Facebook use Recommended/Not Recommended instead of star ratings? Facebook business pages now use a binary Recommended/Not Recommended system, so there is no per-review star rating. The page-level page.recommendationPercentage is Facebook's own figure and is not recalculated from the reviews returned here, so it may not match a simple count of them. Use isRecommended for per-review analysis.

Can I get the business's replies to reviews? Yes. When a business reply is among the comments Facebook shows on a review, it appears inside topComments with isOwnerReply: true. The flag is set when the commenter's account ID matches the page's own ID, not by comparing display names.

How many reviews can I get per page? maxReviews and maxReviewsPerPage each go up to 500. Without cookies, Facebook usually stops serving new reviews to a logged-out visitor well before that (commonly around 5-15 per page) — this is a Facebook-side limit, not a scraper failure; the run's status message notes it when it happens. Supply cookies for a logged-in session to raise this ceiling.

Do I need to configure a proxy? No separate setup is needed — a residential proxy is always used for access to Facebook. proxyCountry just lets you pick which country that proxy exits from.

Will search mode always find the right business? Facebook's own on-site search requires a login, so search mode looks your query up through a public web search; if that finds nothing, it tries a page address built from the first two words of the query. For distinctive business names this generally finds the correct page, but ambiguous queries (a common name with no location) can return the wrong page or no match. For a guaranteed exact match, use url or pageId mode instead.

Does "Most Helpful" sort actually work? Rarely. The actor looks for a sort control on the reviews page, but Facebook's current review pages generally don't show one, so in practice reviews come back in the page's default order for both sortBy values. The run continues normally either way.

What is authorIdType? Facebook identifies most regular reviewers with a rotating pfbid token rather than a permanent numeric ID, meaning the same person isn't always reliably trackable as "the same reviewer" across separate runs. Business/Page accounts (e.g. a company replying to its own reviews) do get a stable numeric ID. authorIdType tells you which kind of identifier you got ("pfbid" or "numeric").

What happens if a page has no reviews or can't be reached? A record with "recordType": "error" and a description is pushed instead of the run failing or crashing.

Are review photo URLs permanent? Facebook CDN URLs can expire after some time — use the scrapedAt timestamp to track freshness. In practice most reviews are text-only or carry a recommendation tag rather than a photo, so reviewPhotos is frequently absent — that reflects real review behavior, not a gap in the scraper.

Can this scrape a business's About info (address, phone, hours, category)? Only partly. The actor doesn't open the About tab, so it never returns opening hours. page.category, page.address, page.phone and page.website are filled in only when Facebook happens to include them with the reviews data, which is often not the case.

Limitations

• 500 review cap — maxReviews and maxReviewsPerPage can't exceed 500, and without cookies far fewer reviews per page are usually available
• Public pages only — private groups, personal profiles, and Marketplace listings are not supported
• Sort control is usually missing — most_helpful falls back to the page's default order when Facebook doesn't show the control
• Search mode resolves via external web search, not Facebook's own search — most reliable for distinctive business names
• Reviewer identity isn't always a stable ID — see authorIdType in the FAQ
• Business "About" details are partial at best — see the FAQ
• Some pages are blocked more often than others — a page that still returns nothing after three attempts produces an error record