YouTube Video Comment & Reply Thread Scraper

Give this actor a list of YouTube video links and it pulls their comment sections — top-level comments with their nested reply threads attached, author identity, like counts, and the pinned/hearted/verified flags that signal which comments the creator actually engaged with.

What this actor does

• Comments and replies: top-level threads with nested replies
• Sorting: top comments or newest first
• Engagement: like counts, reply counts, pinned and hearted flags
• Author data: name, channel URL, and channel ID
• Empty fields are omitted — records only contain values that are actually populated.

Output

Each dataset row is a top-level comment (replyDepth: 0). When includeReplies is enabled and a comment has replies, they're embedded in that row's replies field — an array of reply objects (replyDepth: 1, same shape as a comment row, plus parentCommentId) — rather than being pushed as separate top-level rows of their own.

Field	Type	Description
commentId	string	Unique comment identifier
text	string	Comment text content
authorName	string	Author's display name
authorChannelId	string	Author's YouTube channel ID
authorChannelUrl	string	Author's channel URL
authorProfileImageUrl	string	Author's profile image URL
authorIsChannelOwner	boolean	Whether the author is the video's channel owner
likeCount	integer	Number of likes on the comment
replyCount	integer	Number of replies (always 0 on reply objects themselves; only meaningful on top-level comments)
replyCountText	string	Reply count as text, e.g. "963" (present on top-level comments only; a bare number, not a formatted phrase)
publishedTimeText	string	Relative publish time (e.g., "2 days ago")
commentUrl	string	Direct link to the comment (https://www.youtube.com/watch?v=VIDEO_ID&lc=COMMENT_ID)
isHearted	boolean	Whether the creator hearted this comment
isPinned	boolean	Whether this comment is pinned
isVerifiedAuthor	boolean	Whether the author's channel is verified
isCreator	boolean	Alias for authorIsChannelOwner
isArtist	boolean	Whether the author is a YouTube Music artist
isReply	boolean	Whether this is a reply (true) or top-level comment (false)
parentCommentId	string	Parent comment ID (present only on reply objects, absent on top-level comments)
replyDepth	integer	0 for top-level comments, 1 for replies
videoId	string	YouTube video ID
videoUrl	string	Full video URL
videoTitle	string	Video title
videoChannelName	string	Video channel name
videoChannelId	string	Video channel ID
videoViewCount	integer	Video's total view count
videoPublishedDate	string	Video's absolute publish date (e.g., "Oct 24, 2009")
videoDescription	string	Video's full description text
totalCommentCount	integer	Video's total comment count (present when available)
scrapedAt	string	ISO 8601 timestamp of when the data was scraped
replies	array	Nested reply objects (present only on top-level comment rows, only when includeReplies=true and the comment has replies)

Example Output

{
    "commentId": "UgxB...",
    "text": "Great video!",
    "authorName": "John Doe",
    "authorChannelId": "UCxxx...",
    "authorChannelUrl": "https://www.youtube.com/@johndoe",
    "authorProfileImageUrl": "https://yt3.ggpht.com/...",
    "authorIsChannelOwner": false,
    "likeCount": 42,
    "replyCount": 3,
    "publishedTimeText": "2 days ago",
    "isHearted": false,
    "isPinned": false,
    "isReply": false,
    "replyDepth": 0,
    "videoId": "dQw4w9WgXcQ",
    "videoUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "videoTitle": "Rick Astley - Never Gonna Give You Up",
    "videoChannelName": "Rick Astley",
    "videoChannelId": "UCuAXFkgsw1L7xaCfnd5JJOw",
    "videoViewCount": 1811547602,
    "videoPublishedDate": "Oct 24, 2009",
    "videoDescription": "The official video for \"Never Gonna Give You Up\" by Rick Astley...",
    "totalCommentCount": 2400000,
    "scrapedAt": "2026-02-11T12:00:00.000000+00:00",
    "replies": [
        {
            "commentId": "UgxB....AbCdEfGhIj",
            "text": "I agree!",
            "authorName": "Jane Smith",
            "likeCount": 5,
            "publishedTimeText": "1 day ago",
            "isReply": true,
            "parentCommentId": "UgxB...",
            "replyDepth": 1,
            "commentUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ&lc=UgxB....AbCdEfGhIj"
        }
    ]
}

Each reply object actually carries the full same field set as a top-level comment row (all author fields — authorChannelId, authorChannelUrl, authorProfileImageUrl, authorIsChannelOwner, isVerifiedAuthor, isCreator, isArtist, isHearted, isPinned — plus the full video* context block), not just the trimmed set shown above; the example above only highlights the fields that differ meaningfully from a top-level comment. The one exception is totalCommentCount, which is present on top-level comments but never on reply objects.

Error Handling

If a video URL can't be parsed, doesn't resolve to a real video, has comments disabled, or fails to fetch, the actor pushes a single marker row instead of comment rows — so failures are visible in the dataset instead of silently producing nothing. Marker rows use success (boolean) and error (a human-readable message) instead of comment fields. inputUrl is the original raw input string only when the URL couldn't be parsed at all; for every other case (video not found, comments disabled, fetch failure, zero matches) it's the normalized/canonical video URL, since parsing already succeeded by that point:

Case	success	commentsDisabled	Other fields
URL couldn't be parsed into a video ID	false	false	No videoId/video metadata (never resolved far enough to have any)
Video ID doesn't resolve to a real page (deleted, never existed)	false	false	videoId/videoUrl present, but videoTitle and other video metadata are empty
Video exists, comments are disabled	false	true	Full video metadata (videoTitle, videoViewCount, etc.) is present
Video page/comments fetch failed (bot detection, network error)	false	false	videoId/videoUrl present; video metadata depends on how far the fetch got
Comments enabled but the video has zero (or zero matching your filters)	true	false	commentCount: 0, full video metadata present

Input

Field	Type	Default	Description
videoUrls	array	–	YouTube video URLs, short links (youtu.be), shorts URLs, or plain video IDs. Required.
maxComments	integer	100	Max top-level comments per video (1–50,000). Replies don't count toward this limit
includeReplies	boolean	true	Whether to fetch reply threads for each comment
maxRepliesPerComment	integer	5	Max replies to fetch per comment (0–100)
sortBy	string	"top"	Sort order: "top" (most relevant) or "newest" (chronological)
commentKeywordFilter	string	(none)	Only return comments/replies whose text contains this word or phrase (case-insensitive)
minLikeCount	integer	0	Only return comments/replies with at least this many likes

Supported URL Formats

• https://www.youtube.com/watch?v=dQw4w9WgXcQ
• https://youtu.be/dQw4w9WgXcQ
• https://www.youtube.com/shorts/dQw4w9WgXcQ
• https://www.youtube.com/embed/dQw4w9WgXcQ
• https://www.youtube.com/live/dQw4w9WgXcQ
• https://www.youtube.com/v/dQw4w9WgXcQ
• dQw4w9WgXcQ (bare video ID)

Extra query parameters (&t=30s, &si=..., etc.) and domain casing are both ignored — only the video ID is used. If the same video appears more than once (even in a different URL format), only the first occurrence is scraped; later duplicates are skipped. Within a single video, YouTube can also resurface the same comment more than once during pagination (most commonly a pinned comment, which appears both in its pinned slot and again in its normal chronological position) — the actor tracks comment IDs already returned and skips re-pushing them, so the dataset never contains duplicate comment rows for the same video.

Example Input

{
    "videoUrls": [
        "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
    ],
    "maxComments": 100,
    "includeReplies": true,
    "maxRepliesPerComment": 5,
    "sortBy": "top"
}

Examples

Example: top comments

{
  "videoUrls": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],
  "maxComments": 100,
  "sortBy": "top"
}

Example: newest with replies

{
  "videoUrls": ["https://www.youtube.com/watch?v=9bZkp7q19f0"],
  "maxComments": 50,
  "sortBy": "newest",
  "includeReplies": true,
  "maxRepliesPerComment": 5
}

Use cases

• Brand sentiment tracking — pull comments on a product review or launch video to gauge audience reaction.
• Creator engagement audits — compare like counts and reply volume on top comments across competing channels.
• Pinned-comment monitoring — catch a creator's pinned announcement, discount code, or correction as soon as it's posted.
• Moderation and toxicity research — assemble comment and reply datasets for training or evaluating content-moderation models.
• Audience Q&A mining — scan high-traffic videos for recurring viewer questions worth addressing in future content.
• Discourse analysis — reconstruct full reply chains under a comment to study how a conversation developed.

FAQ

Why are some comments missing? YouTube loads comments page by page, and on very large videos it may stop serving further pages before every comment has been returned. The count is also limited by maxComments, and the keyword and like filters drop non-matching comments.

Are replies included by default? Yes, when includeReplies=true (the default). You can set includeReplies: false or reduce maxRepliesPerComment to skip or cap reply fetching.

Does this require a YouTube API key or cookies? No. The actor reads publicly available comment data and does not require any authentication, API keys, or cookies.

What happens if a video URL is invalid or the video doesn't exist? The actor still produces one dataset row for that entry instead of silently skipping it, marked with success: false and an error message explaining why — see Error Handling above.

What's the difference between top and newest sort? top returns comments in YouTube's own "Top comments" order. newest returns them in YouTube's "Newest first" order.

Can I scrape comments for private or age-gated videos? No. Only public videos are supported. Age-gated and private videos require a logged-in session, which this actor does not use.

How do the keyword filter and minimum like count work? Both are applied locally to the comments the actor already fetched — they don't change what YouTube returns, only what gets saved to the dataset. Both are checked against a top-level comment before deciding whether to fetch its replies: if a comment fails either filter, its replies are skipped entirely (not fetched at all, to save time) — even if some of those replies would individually have passed. Each reply that does get fetched is then checked against both filters again on its own merits.

Is this affiliated with YouTube? No. This is an independent tool and is not affiliated with, endorsed by, or sponsored by YouTube or Google.