YouTube Comments Scraper: comment text, authors and like counts from any public video

Point it at watch links, youtu.be links, Shorts links or bare video ids and you get the comments as flat rows: the text, the author handle and channel id, the like and reply counts, the verified badge and the avatar. No API key to request and no Google quota to keep an eye on.

Read the like count before you build anything on it. YouTube abbreviates it on the page, so a comment with 315,000 likes arrives as the string "315K". Small numbers come through exact, large ones do not, and the exact figure is not recoverable from the text.

Input	Watch, youtu.be or Shorts links, or bare 11 character video ids
Output	One row per comment
Ceiling	10,000 comments per run
Account needed	None from you
Price	$0.40 per 1,000 comments, flat on every plan

💬 What YouTube Comments Scraper does

It reads each video's comment list in YouTube's own default order, which is roughly top comments rather than newest first, and pages through it until the budget runs out.

maxComments is the budget for the whole run, divided evenly across your videos. 500 over five videos is about 100 from each, not 500 from the first. The division rounds down and the remainder is not handed back, so 100 across three videos returns at most 99. Ask for fewer comments than you have videos and each one gets exactly one.

Pass videoUrls and videoIds together if that is what you have. They are merged and deduplicated before anything is fetched.

📥 What you give it

{
  "videoUrls": [
    "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "https://youtu.be/9bZkp7q19f0"
  ],
  "maxComments": 500
}

Field	Default	What it is
videoUrls	none	Watch links, youtu.be short links and /shorts/ links all parse.
videoIds	none	Or bare 11 character ids, if that is what your pipeline holds.
maxComments	100	Total across every video in the run, 1 to 10,000. This is also your spend cap.
notionConnector, notionParentId	none	Optional. Write each comment into a Notion database as well as the dataset.
proxyConfiguration	Apify default	Leave it alone unless you have your own servers you want the run to go out through.

Send at least one video. With nothing usable in either field the run fails rather than returning an explanatory row.

📤 What you get back

A real row from a real run, with the avatar link cut short:

{
  "ok": true,
  "type": "comment",
  "id": "EhpVZ3pnZTM0MGRCZ0I3NWhXQm01NEFhQUJBZyAoKAE%3D",
  "videoId": "dQw4w9WgXcQ",
  "text": "can confirm: he never gave us up",
  "publishedTime": "1 year ago",
  "likeCount": "315K",
  "replyCount": "963",
  "isReply": false,
  "author": "@YouTube",
  "authorChannelId": "UCBR8-60-B28hp2BmDPdntcQ",
  "authorIsVerified": true,
  "authorAvatar": "https://yt3.ggpht.com/3s6evpqAiDU9tQR4sC2siJippbH2RWVPnwHgyl4V0th2iuQz0VDQZbUhQBGmsxLYo-..."
}

Field	What it is
text	The comment as written, newlines and all.
author, authorChannelId	The @handle as displayed, and the stable UC id. Join on authorChannelId, because handles change.
likeCount, replyCount	Display strings, abbreviated by YouTube: "315K", "963". Not numbers.
publishedTime	Relative text such as "1 year ago". There is no absolute timestamp in this payload.
isReply	true when the row sits under another comment rather than at the top level.
id	The comment's entity key from the page. Unique within a video, and stable enough to dedupe on.
authorIsVerified, authorAvatar	The verified tick, and a link to the avatar image.

🧾 Reading the output

Two kinds of row land in your dataset.

Row	How to spot it	Billed
A comment	ok: true and type: "comment"	yes
A diagnostic	ok: false and an errorCode	no

Filter on ok == true and you have your data. A video whose comment list could not be read gets one uncharged NO_RESULTS row naming the videoId, and the run moves on.

Read that row as "the list did not come back", not as "this video has no comments." Comments being switched off is the usual reason, and it is the reason the row text gives, but a request that came back without the list looks the same from here. If a video you know has comments returns NO_RESULTS, run it again on its own before believing it.

The Overview table shows the comment columns only, so a diagnostic reads as a blank line there. Open the row or download the JSON for the errorCode.

▶️ How to run it

1. Open YouTube Comments Scraper and click Try for free.
2. Paste links into Video URLs, or ids into Video IDs.
3. Set Max comments. Start at 100 while you look at the shape.
4. Click Start.
5. Download the dataset as JSON, CSV or Excel, or read it over the Apify API.

💰 How much does it cost?

$0.40 per 1,000 comments. Flat on every Apify plan, no volume tiers.

You pay per comment row delivered. Diagnostic rows and duplicates dropped inside a run are never billed, so a video with comments switched off delivers no charged rows. maxComments is the ceiling on both the rows and the spend.

💡 What people use it for

• Reading a launch video's reaction without scrolling for an hour. The upvoted comments come first, which is where the complaints sit as well as the praise.
• Pulling a few hundred comments off videos in your market and seeing which two or three questions keep coming back. That is what your own page should answer.
• Finding repeat commenters: group by authorChannelId and the accounts on every video are either your regulars or bots.
• Collecting quotes for research, with the handle and channel id attached for attribution.

🚧 What it does not do

• Reply threads are not expanded. Every row carries replyCount, so you know how much conversation sits under a comment, and replies YouTube returns inline are marked isReply: true. It does not open each "view 963 replies" thread. If you need complete reply trees, this is not it.
• No video-level data. Every row is about the comment. The title, view count and description are not in the payload, so use a video scraper for those.
• Top comments order only. There is no newest-first or sort option, because the page does not offer one without signing in.
• Like and reply counts are abbreviated strings. Sort on them and "1.2K" orders as text.
• No real timestamps. publishedTime is relative to when the run read the page.
• The budget divides evenly, so one video in a long list cannot take more than its share.

🧭 Which YouTube scraper do you need?

If you want	Use
Comments under videos you already have links for	This one
Videos to run this on, by keyword or channel	YouTube Scraper
Every upload from one channel	YouTube Channel Scraper
What is actually said in the video	YouTube Transcript Scraper
Shorts, with exact view and like counts	YouTube Shorts Scraper

❓ Questions people ask

Do I need a YouTube or Google account? No. Nothing here is signed in, and there is no quota.

Is maxComments per video? No, per run, split evenly across the videos you pass.

Can I get every comment on a big video? Up to 10,000 in one run, in YouTube's top-comments order. There is no resume, so a second run starts at the top again.

Why are the like counts text? Because that is what the page prints. "315K" is YouTube's own rounding, and the exact number is not on the page to read.

Can I filter by keyword or date? Not in the input. Pull the comments and filter text and publishedTime yourself.

Is scraping YouTube legal? This reads public comments on public videos. They are personal data, which GDPR and similar laws cover, so have a reason for collecting them and be careful about republishing handles. Apify's write-up on the legality of web scraping is a good starting point, and we are not lawyers.

🆘 If something breaks

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