Facebook Comments Scraper: every comment under a public post, video or reel

Paste post links, get the comments underneath them. One row per comment, with the author, the text, the exact time it was written, the like count and how many replies it drew.

The part worth knowing before you start: this reads posts, not Pages. A bare Page link like facebook.com/natgeo comes back rejected, because a Page is not a post and has no single comment thread. Reaction counts above roughly a thousand arrive rounded, because the rounded figure is the only one Facebook publishes.

Input	Post, video, reel, photo or group-post links, up to 100 per run
Output	One row per comment, plus one per reply if you ask for them
Ceiling	5,000 comments per post link
Account needed	None from you
Price	$0.25 per 1,000 comments, flat on every plan

🔍 What Facebook Comments Scraper does

It opens the post the way a signed-out visitor sees it, then walks the comment thread to the end or to your limit, whichever comes first. Facebook hands comments out ten at a time and quietly ignores any request for a bigger page, so a hundred comments is ten round trips. That is the pace of the source, not a setting you can turn up.

Links in the permalink.php?story_fbid=...&id=... and story.php shapes are rewritten to the equivalent /posts/ link before the run starts. Facebook serves a signed-out reader the login screen for those two shapes and the real post for the other, so the rewrite is what makes them work. Paste whichever form you have.

Two URL fields come back on every row for that reason. inputUrl is your link exactly as you pasted it, so you can join the results back onto your own list. facebookUrl is the canonical link Facebook itself uses for that post, which is often a different pfbid string.

📥 What you give it

{
  "startUrls": [
    { "url": "https://www.facebook.com/natgeo/posts/pfbid02rAK2eMyfVYDZbESKoNVuwWjcFr5BgRBfjp27nhgQToSiQD9NF3ypNdXpQuwSp7qnl" }
  ],
  "resultsLimit": 50,
  "includeNestedComments": false
}

Field	Default	What it is
startUrls	none	Public post, video, reel, photo or group-post links. First 100 per run, duplicates dropped.
resultsLimit	50	Comments per post link, 1 to 5,000. Multiply by your number of links to get the ceiling on the run and on the spend.
includeNestedComments	off	Also pull the replies under each comment. Each reply is its own row and counts towards the limit.
viewOption	All comments	How Facebook orders the thread before it is read: all comments, most relevant, or newest first.
onlyCommentsNewerThan	none	Cut-off date. 2026-01-31, a full ISO timestamp, or relative like 3 days or 2 months. Older comments are dropped before they are billed. Anything it cannot read as a date turns the filter off.
proxyUrls	none	Optional. A list such as ["http://user:pass@host:port"], one entry per server, if you want the traffic to leave through servers you already pay for. Stored as a secret.
sessionCookies	none	Optional, and most runs never touch it. Your own c_user and xs pair, so heavy runs use your account's own rate limit. Treat the values like a password.

Run it with empty input and you get one labelled sample row, free, so you can look at the shape before you spend anything.

📤 What you get back

A real row from a real run. The post text and the comment text are both cut short here, and the links are trimmed:

{
  "ok": true,
  "charged": true,
  "recordType": "comment",
  "inputUrl": "https://www.facebook.com/natgeo/posts/pfbid02rAK2eMyfVYDZbESKoNV...",
  "facebookUrl": "https://www.facebook.com/natgeo/posts/pfbid0nMcunWuEpusFmY7mj...",
  "postId": "1634927484671342",
  "postTitle": "National Geographic",
  "postDescription": "A punishing heat wave has left much of Europe in drought, causing rivers to run at record-low levels. ...",
  "pageName": "National Geographic",
  "commentUrl": "https://www.facebook.com/natgeo/posts/pfbid0nMcunWu...?comment_id=27860828630277533",
  "commentId": "27860828630277533",
  "date": "2026-08-15T06:07:03.000Z",
  "timestamp": 1786774023,
  "text": "Perfect opportunity to clean our waters!!! Bring this history to the musuem for our kids. ...",
  "likesCount": 107,
  "likesCountText": "107",
  "commentsCount": 6,
  "threadingDepth": 0,
  "replyToCommentId": null,
  "profileName": "Maggie Dalkowska",
  "profileUrl": "https://www.facebook.com/maggie.dalkowska",
  "attachments": [],
  "scrapedAt": "2026-09-21T01:45:48.038Z"
}

Field	What it is
commentId, id, feedbackId	The numeric comment id, Facebook's base64 node id, and its feedback id. Use commentId as your key.
likesCount, likesCountText	Facebook publishes reactions as a display string ("28", "1.2K"). likesCountText keeps it, likesCount is that string as a number, so 1.2K reads as 1200.
commentsCount	Replies under this comment, not the post's total.
threadingDepth, replyToCommentId	0 for a top-level comment, 1 for a reply, with the parent's id on the reply.
profileName, profileId, profileUrl, profilePicture, author	Who wrote it. profileId is a number on some accounts and an opaque pfbid string on others, and the pfbid form rotates over time.
postTitle, postDescription, pageName, postId	The post the comment sits under, repeated on every row so one row stands on its own.
attachments	{type, url} for photo, video and sticker comments. Empty on ordinary text ones.
date, timestamp, scrapedAt	When it was written, as ISO and as epoch seconds, and when this row was read.

profileUrl is null for commenters whose profile Facebook renders without a public link.

🧾 Reading the output

Three kinds of row can land in your dataset.

Row	How to spot it	Billed
A comment or a reply	ok: true and charged: true	yes
The sample row	_sample: true	no
A diagnostic	ok: false, _diagnostic: true and a code	no

Filter on ok == true and you have your data. The charged flag is what separates real rows from the sample and the diagnostics, not a receipt.

code	What happened
BAD_INPUT	That link is not a post link. The row says which part failed.
NOT_FOUND	The post is private, deleted, or has no comment thread to read.
NO_COMMENTS	The post is readable but has no comments, or none newer than your cut-off.
BLOCKED	Facebook would not serve that post on this run. Re-running usually clears it.
RATE_LIMITED	Facebook asked for a slower pace on that post.
TARGET_ERROR	Facebook answered with something the run could not read.
NETWORK	A request did not finish.
DEAD_BATCH	Fifteen links in a row came back empty before anything had been delivered, so the run stopped instead of working through a list of dead links. The rest were not read and not charged.
PROXY_INPUT_ADJUSTED	A proxyConfiguration setting this Actor does not use was dropped. The run carried on.
CHARGE_ERROR	A billing event could not be recorded. Told to you rather than hidden.

▶️ How to run it

1. Open Facebook Comments Scraper and click Try for free.
2. Paste your links into Facebook post URLs, one per line.
3. Set Max comments per post. Start at 50 while you look at the shape of the output.
4. Click Start.
5. Download the dataset as JSON, CSV or Excel, or pull it from the Apify API.

💰 How much does it cost?

$0.25 per 1,000 comments. The same rate on every Apify plan, no volume tiers.

A reply costs the same as a top-level comment. You pay for comments that come back with data, so the sample row, every diagnostic row, comments dropped by your date filter and duplicates inside a run are not charged. Nor is a post that turns out to have no comments.

💡 What people use it for

• Reading what people actually said under a campaign post, instead of the reaction total.
• Watching a competitor's announcement thread for complaints and feature requests.
• Pulling the commenters on a giveaway post to check who entered.
• Tracking a thread over days by re-running the same link and keying on commentId.

🚧 What it does not do

• Posts, not Pages. There is no Page-wide crawl here. Give it the links to the posts you care about, or use the Posts scraper below to find them first.
• Public content only. A private group, a friends-only post or a Page restricted by age or country comes back as an uncharged diagnostic row saying so.
• No reactions breakdown. You get the total, not how many were love, haha or angry. The public page does not carry it.
• No commenter email or phone. Not published, not inferable, not returned.
• One level of replies. Facebook collapses replies-to-replies into that same level, so that is as deep as the public thread goes.
• Reply counts can exceed the rows you get. A comment may say six replies and return four when some were written by accounts that have since gone.
• Counts above about a thousand are rounded, because Facebook rounds the display string the number is read from.
• 5,000 comments per link. Past that Facebook starts repeating pages rather than continuing, and repeats are dropped rather than charged.

🧭 Which Facebook scraper do you need?

If you want	Use
The comments under posts you already have links for	This one
Recent posts from a Page, with their links	Facebook Posts Scraper
Posts from a public group	Facebook Groups Scraper
Reels from a Page	Facebook Reels Scraper
The reviews and recommendations on a Page	Facebook Reviews Scraper

❓ Questions people ask

Do I need a Facebook account? No. Public posts read fine without one. sessionCookies exists only for people running heavily who would rather use their own account's rate limit.

Is the limit per post or per run? Per post link. Ten links at 500 each is up to 5,000 rows, so keep an eye on it when you paste a long list.

Why did I get fewer comments than the post shows? Either your cut-off date dropped them, or Facebook served fewer than the counter claims, which happens when commenters delete or deactivate.

Can I get replies to replies? No, and neither can anyone else from the public thread. Facebook folds them into the one reply level that includeNestedComments returns.

Is scraping Facebook legal? This reads public posts only, never private content or messages. Comments are personal data, which GDPR and similar laws cover, so have a reason for collecting them. Apify's write-up on the legality of web scraping is a good starting point, and we are not lawyers.

Can I run it on a schedule? Yes, on Apify's scheduler, or start it from the API and read the dataset when it finishes.

🆘 If something breaks

Open the Issues tab on the actor page. Paste the link you used and the run ID, and the log is enough for us to see what happened.