YouTube Comments Scraper – Replies, Likes, Shorts, No API Key

YouTube Comments Scraper that pulls every public comment from YouTube videos and Shorts – text, author, channel, likes, reply count, pinned and hearted flags, dates – and, if you want them, the replies threaded to their parent comment. Paste video URLs, get flat rows you can export to CSV or Excel, load into a database, or feed straight into sentiment analysis and LLM pipelines.

• Comments and replies in one dataset. Every reply is its own row with type: "reply" and parentCommentId, so threads rebuild with one join.
• No API key, no quota. Unlike the YouTube Data API v3 (10,000 units a day, OAuth project, key management), this YouTube comment scraper reads the same public comment feed the YouTube website loads. Nothing to set up.
• Pay only for delivered comments. A video with comments turned off returns one explanatory row that costs nothing.

---

What you get

Every row the YouTube Comments Scraper writes has the same flat shape:

Field	Description
type	comment (top-level), reply, or error (a video that returned nothing, with the reason)
commentId	YouTube's comment id (replies look like parentId.replyId)
videoId, videoUrl, videoTitle	The video the comment belongs to
text	Full comment text, emojis and links included
author	Display name / handle, e.g. @YouTube
authorChannelId, authorChannelUrl	Commenter's channel id and link
authorAvatar	Profile picture URL
authorIsVerified	Verified badge
authorIsChannelOwner	true when the video's own creator wrote it
likes	Like count as an integer ("1.2K" becomes 1200)
replyCount	Number of replies under a top-level comment
publishedTimeText	YouTube's own label, e.g. 3 days ago
publishedAtApprox	ISO 8601 UTC timestamp derived from that label – approximate by nature (YouTube only shows relative times)
isEdited	The comment was edited
isPinned	Pinned by the creator
isHearted	Hearted by the creator
parentCommentId	For replies: the comment they answer (null for top-level comments)
position	Order within the video (comments) or within the thread (replies)
commentUrl	Direct link that opens the video with this comment highlighted
videoCommentCount	Total comment count YouTube shows for the video
scrapedAt	ISO 8601 UTC timestamp of the run
error	Only on type: "error" rows – why a video returned nothing

---

Input

The YouTube Comments Scraper takes one field for all videos. Watch URLs, youtu.be short links, Shorts URLs (youtube.com/shorts/...), live URLs and bare 11-character ids all work, in any mix. With no input at all the actor scrapes 100 top comments of a well-known video, so a test or scheduled run is never empty.

Top 500 comments of two videos

{
  "videoUrls": [
    "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "https://youtu.be/jNQXAC9IVRw"
  ],
  "maxCommentsPerVideo": 500
}

Newest comments with replies – videos and Shorts together

{
  "videoUrls": [
    "https://www.youtube.com/shorts/5mU6SRS2Bxo",
    "dQw4w9WgXcQ"
  ],
  "sortBy": "newest",
  "maxCommentsPerVideo": 200,
  "includeReplies": true,
  "maxRepliesPerComment": 50
}

Option	Default	What it does
videoUrls	one demo video	Video, Shorts, live or youtu.be URLs, or bare ids. Duplicates are skipped. Channel and playlist URLs are not expanded
maxCommentsPerVideo	100	Cap on top-level comments per video (up to 100,000)
sortBy	top	top (YouTube's "Top comments") or newest ("Newest first"). A pinned comment is always listed first
includeReplies	false	Also collect replies, one row each
maxRepliesPerComment	20	Cap on replies per comment when replies are on
concurrency	3	Videos scraped in parallel (1–10)
proxyConfiguration	off	Not needed; Apify datacenter proxy can be switched on for very large volumes

---

Output example

A real top-level comment as the YouTube Comments Scraper returns it:

{
  "type": "comment",
  "commentId": "Ugzge340dBgB75hWBm54AaABAg",
  "videoId": "dQw4w9WgXcQ",
  "videoUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
  "videoTitle": "Rick Astley - Never Gonna Give You Up (Official Video) (4K Remaster)",
  "text": "can confirm: he never gave us up",
  "author": "@YouTube",
  "authorChannelId": "UCBR8-60-B28hp2BmDPdntcQ",
  "authorChannelUrl": "https://www.youtube.com/@YouTube",
  "authorIsVerified": true,
  "authorIsChannelOwner": false,
  "likes": 321000,
  "replyCount": 962,
  "publishedTimeText": "1 year ago",
  "publishedAtApprox": "2025-09-30T12:20:33Z",
  "isEdited": false,
  "isPinned": true,
  "isHearted": true,
  "parentCommentId": null,
  "position": 1,
  "commentUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ&lc=Ugzge340dBgB75hWBm54AaABAg",
  "videoCommentCount": 2457943,
  "scrapedAt": "2026-09-30T18:20:33Z"
}

A reply row has the same fields with type: "reply", replyCount: 0 and parentCommentId set. A video with comments turned off gives one row like {"type": "error", "videoId": "...", "error": "Comments are turned off for this video"}.

---

Use cases

Who uses a YouTube Comments Scraper, and for what:

• Sentiment analysis and brand monitoring – marketing teams score thousands of comments on a launch video, an ad or a competitor's review and track the mood over time.
• Creators and social media managers – export YouTube comments to CSV or Excel, find the most-liked questions to answer, collect FAQ ideas for the next video, spot unanswered replies.
• AI and LLM builders – flat JSON rows are ready for RAG, topic clustering, or a "summarise what viewers think" prompt in ChatGPT or Claude; likes gives a natural relevance weight.
• Market and academic researchers – build comment corpora for discourse, misinformation or audience studies, with thread structure intact via parentCommentId.
• Product teams – mine tech-review and unboxing videos for feature requests and bug reports.
• Agencies and influencer marketers – measure real engagement on sponsored videos and Shorts, not just view counts.
• Community moderators – pull the newest comments on a schedule and flag spam or abuse downstream.

---

Pricing

The YouTube Comments Scraper uses pay per event: you are charged once per delivered comment or reply row. There is no charge for error rows (comments turned off, private or deleted videos, bad URLs), and turning replies off means you only pay for top-level comments. See the Pricing tab for the current per-1,000 price. Platform usage is modest: about 2,400 comments and replies were collected in under a minute in testing.

---

Integrations

Python

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("nourishing_courier/youtube-comments-scraper").call(run_input={
    "videoUrls": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],
    "maxCommentsPerVideo": 300,
    "includeReplies": True,
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(row["type"], row["likes"], row["text"][:80])

JavaScript

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_APIFY_TOKEN' });
const run = await client.actor('nourishing_courier/youtube-comments-scraper').call({
    videoUrls: ['https://www.youtube.com/shorts/5mU6SRS2Bxo'],
    sortBy: 'newest',
    maxCommentsPerVideo: 200,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items.length, 'comments');

curl

curl -X POST "https://api.apify.com/v2/acts/nourishing_courier~youtube-comments-scraper/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN&format=csv" \
  -H "Content-Type: application/json" \
  -d '{"videoUrls":["dQw4w9WgXcQ"],"maxCommentsPerVideo":100}'

n8n, Make, Zapier – add the Apify node, pick the YouTube Comments Scraper, paste an input above and send the rows to Google Sheets, Airtable, Notion, Slack or an OpenAI/Claude node for sentiment analysis. MCP – through the Apify MCP server an AI agent can call it as a tool ("get the top 200 comments on this video and tell me the main complaints"). Datasets download as JSON, CSV, Excel, XML or RSS.

Pairs well with: YouTube Transcript Scraper for what the video says, and YouTube Scraper for video and channel metadata – combine them for a full picture of a video and its audience.

---

Limits and fair use

The YouTube Comments Scraper reads only public comments – the same ones any logged-out visitor sees. Comments on videos with comments turned off (including most "made for kids" videos), private videos and deleted videos cannot be read; each gets an error row explaining why. Age-restricted videos work: their comments load without signing in. YouTube shows only relative dates ("2 weeks ago"), so publishedAtApprox is an estimate whose precision matches the label; like counts above 1,000 are rounded the way YouTube displays them ("1.2K"). No proxy is needed at normal volumes; requests back off automatically on rate limits. Comments are personal data of their authors – use them in line with YouTube's Terms of Service and privacy law such as GDPR, and avoid re-identifying or profiling individuals.

---

FAQ

Can I scrape YouTube comments without an API key?

Yes. This YouTube Comments Scraper needs no Google Cloud project, no API key and no OAuth. It is a practical YouTube comments API alternative with no daily quota.

How do I export YouTube comments to CSV or Excel?

Run the actor, open the Output tab and choose CSV or Excel – or use the format=csv API link shown above. Each comment is one flat row, so it opens cleanly in Excel or Google Sheets.

Does it scrape replies to comments?

Yes. Turn on includeReplies and the YouTube Comments Scraper fetches each thread; every reply becomes its own row with parentCommentId, capped by maxRepliesPerComment.

Can I scrape comments from YouTube Shorts?

Yes. Paste the Shorts URL (youtube.com/shorts/...) exactly as it appears; Shorts comments come out in the same format as regular video comments.

How many comments can I get per video?

The YouTube Comments Scraper collects up to 100,000 top-level comments per video (maxCommentsPerVideo), plus replies if enabled. For very popular videos YouTube keeps loading older comments page by page, and the actor follows until your cap.

Can I get the newest comments instead of top comments?

Yes – set sortBy to newest. It uses YouTube's own "Newest first" order, handy for monitoring a video on a schedule.

Is it good for sentiment analysis with an LLM?

That is a main use case for this YouTube Comments Scraper. Rows are flat and clean, with likes for weighting and isHearted/authorIsChannelOwner for creator signals. Send the dataset to an LLM in n8n, Make or your own code.

Does it work with n8n, Make and MCP?

Yes. The YouTube Comments Scraper is a standard Apify actor. Use the Apify integration in n8n or Make, or the Apify MCP server to give an AI agent the actor as a tool.

Why did a video return an error row?

Comments are turned off, the video is private or deleted, or the URL was not a video (channel and playlist URLs are not expanded). The error text says which, and those rows are free.

Are dates exact?

YouTube only exposes relative times such as "5 months ago", so publishedAtApprox is an approximation derived from that label; the original label is kept in publishedTimeText.