📘 Facebook Groups & Comments Scraper

Facebook Groups Scraper — Posts, Comments and Contact Data

Extract posts from any public Facebook group as clean, structured JSON — full text, author profile, a per-reaction breakdown (like, love, haha, wow, sad, angry, care), photo/video/link attachments, absolute timestamps, and optional per-post comments. The scraper also derives contact-style signals from post text — hashtags, mentions, links, emails, phones, and an engagementScore — so social listening teams, lead-gen agencies, and researchers can go straight from a group URL to a filterable dataset. Pricing is pay-per-event: you're charged per post (row_result) and, only if you opt in, per comment (comment_result).

What is Facebook Groups Posts & Comments Scraper?

Facebook Groups Posts & Comments Scraper (Actor name facebook-groups-posts-comments-scraper) reads the public feed of one or more Facebook groups and returns each post as a structured JSON row, with an optional child row per comment. It works against Facebook's own internal GraphQL feed endpoint (GroupsCometFeedRegularStoriesPaginationQuery) rather than a login-gated browser session, so no Facebook account is required for public groups. For closed or private groups you belong to, supplying cookieString (a logged-in session's c_user + xs cookies) unlocks access — this is optional, not required.

Key capabilities, all confirmed in the source:

• Group discovery by URL only — paste one or more group URLs (or numeric group IDs) into startUrls; there is no keyword-based group search.
• Four feed orderings via viewOption — chronological, recent-activity, top-posts, and a dedicated Buy/Sell listings order for marketplace-style groups.
• A full 7-way reaction breakdown per post (reactionLikeCount … reactionCareCount) plus a computed total, not just an aggregate like count.
• Derived lead-generation fields computed from post text with no AI required — hashtags, mentions, links, emails, phones, and a weighted engagementScore.
• Six real filters applied before any post is charged — keyword include/exclude, minimum likes/comments/shares, and a date range.
• Optional per-post comments and optional LLM enrichment, both off by default so the base run stays cheap.
• JSON output via the Apify dataset, exportable to CSV, Excel, XML, or RSS from the Apify Console like any Apify dataset.

What data can I extract with Facebook Groups Posts & Comments Scraper?

Every post is pushed as one dataset row of type: "post". The Actor's row-building function (normalize_post in src/extractor.py) writes 31 keys, derive_text_fields in src/enrich.py adds 7 more, and — only when aiEnhancement is on — src/ai.py adds up to 5 AI keys, for 38–43 fields per post row depending on configuration. The Apify Store's default dataset view shows only 34 of these; see the Output sample section below for the 9 fields it hides.

Field	Example value	Use case
type	"post"	Distinguish post rows from comment rows
isChild	false	true on comment rows, linking them to a parent post
parentId	null	Populated on comment rows only, points at the parent post id
facebookUrl	"https://www.facebook.com/groups/cheapmealideas/"	The group URL this row came from
inputUrl	"https://www.facebook.com/groups/cheapmealideas/"	The exact URL supplied in startUrls
url	"https://www.facebook.com/groups/cheapmealideas/permalink/..."	Canonical post permalink
id	"1234567890123456"	Facebook's internal story node ID
postId	"1204545088344463"	Facebook's post ID
legacyId	"1204545088344463"	Legacy post ID (mirrors postId)
feedbackId	"ZmVlZGJhY2s6MTIwNDU0NTA4ODM0NDQ2Mw=="	Facebook's UFI feedback object ID
timestamp	1774704000	Post creation time, Unix epoch seconds
date	"2026-03-28T14:00:00.000Z"	Post creation time, ISO 8601
scrapedAt	"2026-07-26T09:12:03.000Z"	When this row was extracted
user	{"id": "...", "name": "Jane Doe", "profileUrl": "...", "profilePicture": "...", "__typename": "User"}	Author identity
text	"Anyone have a good vegan chili recipe?"	Full post body
postType	"photo"	One of text, photo, video, link, album, reshare
reactionLikeCount … reactionCareCount	42, 11, 3, 0, 0, 0, 2	Per-type reaction counts
topReactionsCount	58	Sum of the reaction breakdown above
likesCount	61	Facebook's own total-reaction counter (may exceed topReactionsCount when Facebook doesn't surface every reaction type in the breakdown)
commentsCount	14	Total comments on the post, as reported by Facebook
sharesCount	2	Total shares
attachments	[{"type": "photo", "image": {"uri": "...", "height": 960, "width": 720}}]	Photos, videos, shared links, or albums
topComments	[{"id": "...", "text": "...", "date": "...", ...}]	Up to 2 preview comments, always present regardless of includeComments
facebookId	"106823902845"	Numeric ID of the target group
groupTitle	"Cheap Meal Ideas"	Group display name
hashtags	["mealprep", "budget"]	Hashtags parsed from text
mentions	["johnsmith"]	@mentions parsed from text
links	["https://example.com/recipe"]	URLs parsed from text
emails	["chef@example.com"]	Email addresses parsed from text
phones	["+1 555-201-4488"]	Phone-like strings parsed from text (7–15 digits)
engagementScore	109	likes + 3×comments + 5×shares — a simple weighted score, see below
postAgeDays	4	Days between date and scrape time
aiSentiment	"positive"	Only when aiEnhancement is on and enrichment succeeds
aiTopics	["recipes", "budget cooking"]	Up to 3 tags, AI-only
aiLeadSignal	"medium"	AI-only, see AI enrichment note below

Engagement and reaction data

Every post carries Facebook's full reaction breakdown, not just a total: reactionLikeCount, reactionLoveCount, reactionHahaCount, reactionWowCount, reactionSadCount, reactionAngryCount, and reactionCareCount, summed into topReactionsCount. Facebook also reports a separate likesCount (its own total-reaction counter) and commentsCount / sharesCount. engagementScore is a deterministic weighted score computed in the source (compute_engagement_score in src/enrich.py) as likes + 3×comments + 5×shares — useful for ranking a batch of posts by engagement without writing your own formula. A market researcher tracking a niche community can sort the dataset by engagementScore or topReactionsCount to surface the posts driving the most reaction in a given week.

Targeting and filtering data

resultsLimit, viewOption, keywords / excludeKeywords, minLikes / minComments / minShares, and onlyPostsNewerThan / onlyPostsOlderThan are the fields you filter and segment on before a single credit is spent — all six are enforced in passes_filters() (src/enrich.py) before Actor.push_data() is called, so filtered-out posts are never charged. searchGroupKeyword and searchGroupYear add a second, client-side narrowing pass inside the feed fetch itself.

Why not build this yourself?

Facebook does not expose a public API for reading a group's post feed to ordinary third-party developers — Meta's Graph API groups permissions are restricted to a small set of approved partner use cases, not general-purpose scraping. Building this yourself against Facebook's actual feed means reverse-engineering the same things this Actor's source code does: discovering the group's internal node_id, extracting the current GraphQL doc_id for GroupsCometFeedRegularStoriesPaginationQuery (which changes whenever Facebook ships new frontend code — see extract_doc_id_from_content() in src/app.py, which tries nine regex patterns before falling back to scanning bundled JS files), pulling CSRF tokens (lsd, fb_dtsg, jazoest) out of the bootstrap HTML, handling Facebook's login-wall redirect for unauthenticated or flagged requests, paginating with an end_cursor, deduplicating overlapping post edges across pages, and rotating IPs through a proxy when Facebook rate-limits or blocks a request.

None of that is exposed through documented, stable Facebook infrastructure — it is Facebook's internal Comet frontend contract, which this Actor already tracks with fallback patterns for when Facebook changes it. Building and then maintaining that pipeline yourself — including the residential-proxy cost, since Facebook blocks most datacenter and guest IPs — is the ongoing engineering cost this Actor removes. If you already have a compliant, licensed data source for Facebook group content (for example, an approved Meta platform partnership), use that instead; if you need ad-hoc or recurring access to public group posts without maintaining scraping infrastructure, this Actor is built for that.

How to use data extracted from Facebook groups?

Lead generation and sales teams

emails, phones, mentions, and links are parsed out of every post's text automatically, and — with aiEnhancement on — aiBuyIntent and aiLeadSignal classify whether a post reads like someone wanting to buy, sell, or hire. Set keywords to your product category, minLikes to filter out low-engagement noise, and export the rows where aiLeadSignal is "medium" or "high" straight into a CRM import.

Agencies monitoring groups for clients

Point startUrls at the niche groups a client's audience lives in, schedule the Actor to run daily or weekly through Apify Scheduler, and use onlyPostsNewerThan (e.g. "1 days") so each run only returns posts since the last one. groupTitle on every row makes it trivial to split a multi-group run back out per client in a spreadsheet or BI tool.

Market research and intelligence

Run the same group over time with viewOption: "CHRONOLOGICAL" to build a time-ordered archive of discussion, then use engagementScore, postAgeDays, and the reaction breakdown to chart which topics and post formats get the most traction in a community — useful for category research before a product launch or for benchmarking a competitor's community engagement.

AI agents and automated pipelines

Because the Actor is a standard Apify Actor, any agent framework that can call the Apify API can run it, poll the run status, and pull dataset items as structured JSON — no custom parser needed. Feed aiSentiment, aiTopics, and aiLeadSignal (when aiEnhancement is on) directly into a downstream classification or alerting pipeline, or use the raw text field as retrieval context for a RAG system built on community discussion data.

🔼 Input sample

All 20 input parameters, read directly from .actor/actor.json:

Parameter	Required	Type	Constraints	Description
startUrls	Yes	array	—	Public Facebook group URLs. Group numeric IDs also work.
resultsLimit	No	integer	default 20, min 1	Maximum number of top-level POSTS. Comments are additional and NOT counted against this limit.
viewOption	No	string (enum)	default CHRONOLOGICAL	Feed order: CHRONOLOGICAL (oldest→newest), RECENT_ACTIVITY, TOP_POSTS, CHRONOLOGICAL_LISTINGS (Buy/Sell).
keywords	No	array	default []	Keep only posts whose text contains at least one of these words (case-insensitive).
searchGroupKeyword	No	string	default ""	Client-side search by letter/word inside the group. Without login, single/double letters work best.
searchGroupYear	No	string	default ""	Keep only posts from a specific year, e.g. "2024". Best combined with searchGroupKeyword.
excludeKeywords	No	array	default []	Drop posts whose text contains any of these words (case-insensitive).
minLikes	No	integer	default 0, min 0	Drop posts with fewer likes/reactions than this.
minComments	No	integer	default 0, min 0	Drop posts with fewer comments than this.
minShares	No	integer	default 0, min 0	Drop posts with fewer shares than this.
onlyPostsNewerThan	No	string (date)	—	Keep posts on/after this date. Absolute (2024-01-15) or relative ("2 weeks", "30 days", "1 month").
onlyPostsOlderThan	No	string (date)	—	Keep posts on/before this date. Absolute or relative. Combine with onlyPostsNewerThan for a range.
includeComments	No	boolean	default false	Collect available comments per post, interleaved after the post row.
maxComments	No	integer	default 10, min 0	Cap comments collected per post. 0 = no per-post cap.
aiEnhancement	No	boolean	default false	Classify each post's text (sentiment, topics, buy-intent, lead signal) when on AND a key is provided.
aiModel	No	string (enum)	default claude-haiku-4-5	LLM model; provider auto-detected from the name prefix (claude-*=Anthropic, gpt-*/o1/o3=OpenAI, gemini-*=Google, grok-*=xAI, deepseek-*=DeepSeek, sonar*=Perplexity, mistral-*=Mistral). 22 models total.
aiApiKey	No	string (secret)	—	Your LLM provider API key. Falls back to ANTHROPIC_API_KEY / OPENAI_API_KEY / GEMINI_API_KEY / XAI_API_KEY / DEEPSEEK_API_KEY / PERPLEXITY_API_KEY / MISTRAL_API_KEY env vars per provider.
maxRequestRetries	No	integer	default 3, min 0	Retries per GraphQL request before giving up.
proxyConfiguration	No	object (proxy)	prefill: Apify residential	Proxy used for requests. If left unset, Apify residential proxy is used automatically — Facebook blocks most datacenter IPs.
cookieString	No	string (secret)	—	Logged-in session cookies (must include c_user + xs) for reliability or closed/private groups. Format: c_user=...; xs=...; datr=...; fr=....

Example input

{
  "startUrls": ["https://www.facebook.com/groups/cheapmealideas/"],
  "resultsLimit": 25,
  "viewOption": "RECENT_ACTIVITY",
  "excludeKeywords": ["giveaway", "spam"],
  "minLikes": 5,
  "onlyPostsNewerThan": "30 days",
  "includeComments": true,
  "maxComments": 10
}

Common pitfall

resultsLimit counts top-level posts only — comments pushed when includeComments is true are billed and returned separately and do not reduce your post count. A second, less obvious one: excludeKeywords, minLikes, minComments, minShares, onlyPostsNewerThan, and onlyPostsOlderThan are all applied before a post is charged, so a run with strict filters and a small target group may exhaust the feed (the Actor stops after 10 fetch attempts, MAX_ATTEMPTS in src/main.py) before reaching resultsLimit — this is expected behavior, not a bug, when the group simply doesn't have that many matching posts.

🔽 Output sample

Output is typed, normalized JSON pushed to the Apify dataset — export as JSON, CSV, Excel, or XML from the Apify Console or API like any Apify dataset. Two row shapes exist: post rows (type: "post") and, only when includeComments is on, comment rows (type: "comment") interleaved directly after their parent.

The Apify Store's default dataset view shows 34 fields; the actual post row carries up to 43. Fields present on every row but excluded from the default view: facebookUrl, inputUrl, postId, legacyId, timestamp, topComments, and facebookId. Two AI-only fields are also written to every row but missing from the view: aiLanguage and aiBuyIntent — both are computed and charged for like the other AI fields when aiEnhancement is on, they simply aren't columns in the default table. Switch the dataset view or pull raw items via the API to see them.

Full post row example (post-type fields, values illustrative):

{
  "type": "post",
  "isChild": false,
  "parentId": null,
  "facebookUrl": "https://www.facebook.com/groups/cheapmealideas/",
  "inputUrl": "https://www.facebook.com/groups/cheapmealideas/",
  "url": "https://www.facebook.com/groups/cheapmealideas/permalink/1204545088344463/",
  "id": "Uzpmc3RvcnkuUzoxOjEyMDQ1NDUwODgzNDQ0NjM",
  "postId": "1204545088344463",
  "legacyId": "1204545088344463",
  "feedbackId": "ZmVlZGJhY2s6MTIwNDU0NTA4ODM0NDQ2Mw==",
  "timestamp": 1774702800,
  "date": "2026-03-28T14:00:00.000Z",
  "scrapedAt": "2026-07-26T09:12:03.000Z",
  "user": {
    "id": "100063669491743",
    "name": "Jane Doe",
    "profileUrl": "https://www.facebook.com/jane.doe",
    "profilePicture": "https://scontent.fbcdn.net/...",
    "__typename": "User"
  },
  "text": "Anyone have a good vegan chili recipe? Email me at chef@example.com if you want to swap notes! #mealprep",
  "postType": "photo",
  "reactionLikeCount": 42,
  "reactionLoveCount": 11,
  "reactionHahaCount": 3,
  "reactionWowCount": 0,
  "reactionSadCount": 0,
  "reactionAngryCount": 0,
  "reactionCareCount": 2,
  "topReactionsCount": 58,
  "likesCount": 61,
  "commentsCount": 14,
  "sharesCount": 2,
  "attachments": [
    {"type": "photo", "__typename": "Photo", "id": "...", "thumbnail": "...", "image": {"uri": "...", "height": 960, "width": 720}, "ocrText": null}
  ],
  "topComments": [
    {"id": "...", "legacyId": "...", "commentUrl": "...", "date": "2026-03-28T14:20:00.000Z", "text": "This looks amazing!", "profileId": "...", "profileName": "...", "profileUrl": "...", "profilePicture": "...", "likesCount": 3, "threadingDepth": 0}
  ],
  "facebookId": "106823902845",
  "groupTitle": "Cheap Meal Ideas",
  "hashtags": ["mealprep"],
  "mentions": [],
  "links": [],
  "emails": ["chef@example.com"],
  "phones": [],
  "engagementScore": 109,
  "postAgeDays": 120,
  "aiSentiment": "positive",
  "aiLanguage": "en",
  "aiTopics": ["recipes", "meal prep"],
  "aiBuyIntent": false,
  "aiLeadSignal": "low"
}

Comment row example (only present when includeComments is true, and only carries these 14 fields — separate shape from the post row):

{
  "type": "comment",
  "isChild": true,
  "isReply": false,
  "parentId": "1204545088344463",
  "parentUrl": "https://www.facebook.com/groups/cheapmealideas/permalink/1204545088344463/",
  "facebookUrl": "https://www.facebook.com/groups/cheapmealideas/",
  "id": "Y29tbWVudDoxMjA0NTQ1MDg4MzQ0NDYzXzk4NzY1NDMy",
  "legacyId": "9876543210",
  "url": "https://www.facebook.com/groups/cheapmealideas/permalink/1204545088344463/?comment_id=9876543210",
  "date": "2026-03-28T14:22:00.000Z",
  "text": "Try adding smoked paprika, game changer.",
  "user": {"id": "...", "name": "John Smith", "profileUrl": "...", "profilePicture": "..."},
  "likesCount": 5,
  "threadingDepth": 0
}

Comment rows are also mirrored to a per-run child dataset named comments-<runId> for teams who want posts and comments in separate datasets.

Note on group metadata: groupTitle on every row is the group's display name only. Aggregate group metadata collected once per run — member count, privacy setting, description — is written to the run's key-value store under the RUN_SUMMARY key, not as a per-row dataset field.

How do you filter and target specific posts?

This is the Actor's most useful skill: narrowing a group's entire feed down to the handful of posts you actually need, without paying for the rest. Four axes matter.

Free-text keywords vs. structural search. keywords and excludeKeywords are case-insensitive substring matches against post text, applied client-side after extraction — accurate but literal (no stemming or fuzzy match). searchGroupKeyword runs a second, earlier client-side pass inside the feed fetch itself; per the schema, when scraping without cookieString, single or double-letter search terms return more complete results than full words, because Facebook's own unauthenticated search behaves the same way.

Scope precision via date range. onlyPostsNewerThan and onlyPostsOlderThan accept either an absolute date (2024-01-15) or a relative expression ("2 weeks", "3 months", "1 year"). Combine both to bound a window; combine either with searchGroupYear for coarse year-level scoping before applying a tighter date filter.

Quality thresholds. minLikes, minComments, and minShares drop low-engagement posts before they're charged — useful for cutting straight to the posts an audience actually reacted to, rather than paying to review every post in a busy group.

Volume and feed order. resultsLimit caps top-level posts (not comments); viewOption decides which posts you see first — CHRONOLOGICAL for a complete time-ordered archive, RECENT_ACTIVITY for what's currently being discussed, TOP_POSTS for the highest-engagement content, or CHRONOLOGICAL_LISTINGS for Buy/Sell-style groups where the feed is organized around listings rather than discussion.

Three real examples:

{ "startUrls": ["https://www.facebook.com/groups/cheapmealideas/"], "viewOption": "TOP_POSTS", "minLikes": 20, "resultsLimit": 15 }

{ "startUrls": ["https://www.facebook.com/groups/localmarketplace/"], "viewOption": "CHRONOLOGICAL_LISTINGS", "resultsLimit": 50 }

{ "startUrls": ["https://www.facebook.com/groups/cheapmealideas/"], "keywords": ["recipe"], "excludeKeywords": ["giveaway"], "onlyPostsNewerThan": "2024-01-01", "onlyPostsOlderThan": "2024-12-31" }

▶️ Want to try other Facebook scrapers?

Scraper	What it extracts
🔎 Facebook Groups Search Scraper Plus	Searches and filters Facebook groups (SimpleAPI account)
📘 Facebook Group Posts Scraper Pro	Public Facebook group posts and comments, a separate implementation from this Actor (SimpleAPI account)
📸 Facebook Photos Scraper	Photos from Facebook profiles and Pages (SimpleAPI account)
👥 Instagram Followers Scraper	Follower lists and profile data from Instagram accounts (SimpleAPI account)
📇 Instagram Profile Contact Enricher	Public contact details (email/phone) from Instagram profiles (SimpleAPI account)
🔎 LinkedIn Profile Contact Lookup	Public contact details from LinkedIn profile URLs (SimpleAPI account)

How to extract Facebook group data programmatically

This Actor runs on Apify like any other Actor — start it from the Apify Console, via a schedule, or programmatically through apify-client, and read results back as JSON from the run's dataset.

Python example

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_API_TOKEN>")

run = client.actor("facebook-groups-posts-comments-scraper").call(run_input={
    "startUrls": ["https://www.facebook.com/groups/cheapmealideas/"],
    "resultsLimit": 25,
    "minLikes": 5,
    "onlyPostsNewerThan": "30 days",
})

for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["groupTitle"], item["text"][:80], item["engagementScore"])

Export to spreadsheets or CRM

From the Apify Console, export the dataset directly as CSV or Excel — text, emails, phones, engagementScore, and aiLeadSignal (when AI enrichment is on) map cleanly to spreadsheet or CRM columns for lead review, with url as the record's canonical link back to the source post.

Is it legal to scrape Facebook groups?

Scraping publicly accessible Facebook group posts is generally lawful — public posts are, by definition, visible to anyone without logging in, and courts in the US (notably hiQ Labs v. LinkedIn) have held that scraping publicly available web data does not violate the Computer Fraud and Abuse Act. That said, post rows in this dataset carry personal data — author names, profile URLs, profile pictures, and comment text from identifiable individuals — so GDPR (if you or your data subjects are in the EU/UK) and CCPA (if they're in California) govern how you may store, process, and reuse it, not just Facebook's own Terms of Service. You need a lawful basis for processing personal data at scale, and Facebook's Terms separately restrict automated data collection regardless of the data's legal status. Consult legal counsel for commercial applications involving bulk storage of personal data.

❓ FAQ

What happens if a post or comment is deleted after I scrape it?

The Actor returns data exactly as it existed on the group's public feed at request time. It does not track a post after scraping, so a deleted or edited post afterward will not update your dataset — re-run the Actor to get current state.

Can I get comments along with the posts?

Yes — set includeComments to true and comments are interleaved right after their parent post, capped by maxComments (default 10, 0 = no cap). One caveat worth knowing upfront: comments come from what Facebook's own feed response already includes for that post (get_comments_from_payload in src/extractor.py), not from a separate paginated comment-thread request — so maxComments can only return up to what Facebook bundled with the post, and threaded replies beneath a top-level comment are not fetched (every comment row is written with isReply: false; there is no reply-nesting logic in the source). Every post row also always carries up to 2 preview comments in its topComments field regardless of the includeComments setting.

How accurate is the contact data extracted from posts?

emails, phones, mentions, and links are extracted directly from each post's public text with regular expressions — they are exactly what the poster wrote, not verified or deduplicated against an external directory. Treat extracted phone numbers and emails as leads to validate, not confirmed contact records.

How many posts can I get per run?

There is no maximum enforced by the input schema — resultsLimit has a minimum of 1 but no stated upper bound. In practice, the run stops itself after 10 fetch attempts (MAX_ATTEMPTS in src/main.py) even if resultsLimit hasn't been reached, which happens when the group's feed is exhausted or strict filters (excludeKeywords, minimum thresholds, date range) leave nothing left to match.

What's the most useful filter for narrowing down a group's feed?

viewOption: "CHRONOLOGICAL_LISTINGS" is the one most competitors don't expose — it switches the feed to Facebook's Buy/Sell listing order for marketplace-style groups, instead of forcing you to filter a discussion-ordered feed for listing posts after the fact.

Does this Actor work with Claude, ChatGPT, and AI agent frameworks?

Yes, as a standard HTTP-callable Apify Actor — any agent framework that can call the Apify API (start a run, poll status, read dataset items) can drive it. The Actor also has its own optional aiEnhancement input that classifies post text with your choice of Anthropic, OpenAI, Google, xAI, DeepSeek, Perplexity, or Mistral models via your own API key, independent of whichever agent framework is orchestrating the run.

How does this compare to other Facebook group scrapers?

As observed on the Apify Store on 2026-07-26: unseenuser/fb-groups returns posts only (no group-search, no group metadata, no private-group access by design) at $4 per 1,000 results with a 50-result free-plan cap, and sells full comment threads and replies as separate billed add-ons. whoareyouanas/facebook-group-scraper covers both groups and Pages from one input and claims cookie-based access to private groups plus a full 6-reaction breakdown and group/Page metadata capture. alien_force/facebook-posts-comments-scraper takes individual post URLs rather than group URLs, so it scrapes one known post at a time instead of paginating a group's feed, and is priced as a flat monthly subscription rather than pay-per-event. This Actor differs by pairing full group-feed pagination and a 7-reaction breakdown with derived contact-signal fields (emails, phones, mentions) and optional multi-provider AI classification (sentiment, topics, buy-intent, lead signal) computed on your own key, on pay-per-event pricing.

Can I use this without a Facebook API key or developer account?

Yes for public groups — no Facebook account, API key, or developer app is required; the Actor works from a public group URL alone. cookieString (your own logged-in session cookies) is only needed to improve reliability against login walls or to access closed/private groups you already belong to.

Conclusion

Facebook Groups Posts & Comments Scraper turns a public group URL into a structured, filterable dataset of posts and — on request — their comments, with a full reaction breakdown, derived contact signals, and optional AI classification layered on top. It's built for teams who need a repeatable, schema-stable way to pull group discussion data without maintaining Facebook's internal GraphQL contract themselves, priced per result so a small monitoring run and a large research pull scale the same way. Start it from the Apify Console with a group URL and resultsLimit, or drive it from apify-client as shown above.