YouTube Pro Scraper — Videos, Channels, Comments & Transcripts

Scrape public YouTube at scale with plain HTTP: video metadata, channel profiles, playlists, top-level comments with reply threads, full transcripts with SRT, and Shorts/Live detection — in a single actor, with pay-per-event pricing.

Built for researchers, marketers, sales teams, and AI apps that need structured YouTube data without juggling three or four separate scrapers.

What it does

Five combinable modes in one run:

1. Search — keyword search with sort order (relevance, date, viewCount, rating) and country/language localization. Returns full video detail per result, not just snippets.
2. Videos — full detail from URLs or 11-char IDs: views, likes, duration, exact publish date, tags, category, description, chapters, thumbnails, channel info, Short/Live flags.
3. Channels — profile (subscribers, video count, description, avatar, banner) plus recent videos. Accepts @handle, channel URLs, or UC… IDs.
4. Playlists — expanded to full video metadata. Private/deleted entries come back as flagged partial items (free, no charge).
5. Comments — top-level comments per video with nested reply threads (author, text, likes, reply count). Charged per top-level comment.

Optional extras per video: full transcript (text + segments + SRT, manual captions preferred, auto-generated fallback), dislike estimate (via the public Return YouTube Dislike aggregate), and a markdown rendering ready to paste into an LLM prompt.

Monitor mode for scheduled runs returns only items that appeared since the previous run — track any channel, keyword, or playlist over time without re-paying for old data.

How to use

1. Add one or more inputs: searchQueries, videoUrls, channelUrls, playlistUrls.
2. Tune maxResults (per query/channel), maxComments, includeTranscript, and the hard cap maxItems (controls total spend).
3. Run. Results land in the default dataset; each video/channel/comment row maps 1:1 to a billed pay-per-event charge shown below.

Example input

{
  "searchQueries": ["lofi hip hop radio"],
  "maxResults": 10,
  "maxComments": 0,
  "includeTranscript": false,
  "maxItems": 10
}

Example output (real item)

{
  "id": "5yx6BWlEVcY",
  "url": "https://www.youtube.com/watch?v=5yx6BWlEVcY",
  "title": "Chillhop Radio - jazzy & lofi hip hop beats",
  "description": "Welcome to the Chillhop Radio, playing the best jazzy / lofi hip hop beats...",
  "viewCount": 59983304,
  "likeCount": 381000,
  "commentCount": null,
  "durationSeconds": null,
  "publishedAt": "2020-03-21T06:21:02-07:00",
  "tags": ["lofi hip hop", "nonstop lofi", "lofi radio", "study beats radio"],
  "category": "Music",
  "isShort": false,
  "isLive": true,
  "channelId": "UCOxqgCwgOqC2lMqC5PYz_Dg",
  "channelName": "Chillhop Music",
  "subscriberCount": 3300000,
  "transcript": null,
  "comments": null,
  "type": "video"
}

With maxComments: 3, includeTranscript: true, the same item additionally carries a comments array (each with replies) and a transcript object:

"comments": [
  {
    "id": "UgzAuQpdzR30bsB2KSJ4AaABAg",
    "author": "@gutluckbro9802",
    "authorChannelUrl": "https://www.youtube.com/channel/UCTStHu4wR8ZpWrHFQBb5FCQ",
    "text": "Who's watching this after IFT - 14?",
    "likeCount": 31,
    "replyCount": 1,
    "publishedTime": "7 days ago",
    "replies": [{ "author": "@user", "text": "Me!", "...": "..." }]
  }
],
"transcript": {
  "text": "full plain text...",
  "segments": [{ "start": 0, "duration": 4.8, "text": "..." }],
  "srt": "1\n00:00:00,000 --> 00:00:04,800\n...\n",
  "language": "en",
  "isAutoGenerated": true
}

Pricing (pay per event)

Event	Price	What counts
video-item	$0.002	Each video with full metadata (transcript + nested comments included in the same item)
channel-item	$0.002	Each channel profile
comment-item	$0.0004	Each top-level comment fetched (replies included free)

Examples:

• 1,000 videos ≈ $2.00
• 1,000 videos with 10 comments each ≈ $2.00 + $4.00 = $6.00
• Playlist containers and private/deleted/unavailable videos are free (pushed without charge so you can see what was skipped).

The actor respects your max-spend budget automatically: it stops requesting more results when the charge limit is reached, and maxItems caps the total dataset size no matter what.

How it compares

Most YouTube actors on the Store charge roughly $0.005 per video and ship search or comments or transcripts as separate products. This actor combines all five modes plus reply threads, full transcripts with SRT, incremental monitor mode, and LLM-ready markdown in one run, at $0.002 per video — with comments at a fraction of a cent each. Compare the price tables and feature lists before you run; the dataset schema above is exactly what you get.

Transcripts & SRT

includeTranscript: true attaches the video's captions: plain text, timestamped segments, and a ready-to-save srt string, plus the caption language and whether it is auto-generated. Manual captions in your language are preferred; auto-generated tracks are the fallback. Live streams and videos with captions disabled return transcript: null (metadata is still delivered and still counts as one video-item).

Monitor mode

Set monitorMode: true and schedule the actor (e.g. daily). State is kept in a youtube-monitor key-value store: each run returns only videos/comments not seen before and updates the state. Perfect for tracking a competitor's channel, a keyword, or a playlist.

Limits & errors

• Pure HTTP, no browser: fast and cheap. YouTube rate-limits aggressively, so the actor retries with exponential backoff (3 attempts) and supports the built-in proxy configuration — residential proxies are recommended for large runs.
• Optional youtubeApiKeys: a pool of YouTube Data API v3 keys used as automatic fallback if the public endpoints ever throttle. Quota rotates across keys.
• maxResults 1–500 per query/channel/playlist; maxItems 1–100,000 total.
• One bad URL never kills a run: per-item failures are logged and skipped.
• Exact comment timestamps are not public on YouTube; comment items carry YouTube's relative time (e.g. "7 days ago"). Video/channel dates are exact ISO timestamps.
• Dislike counts have been hidden by YouTube since 2021; dislikeEstimate is a clearly-labeled estimate from a public aggregate, only fetched when you opt in.

FAQ

Is this legal? The actor reads only public pages and endpoints — the same data any logged-out browser sees. It never touches logins, CAPTCHAs, or private/deleted content (those return empty flagged items). You are responsible for complying with YouTube's Terms of Service for your use case.

Why are some fields null? null means YouTube didn't expose it (e.g. no captions, live chat instead of comments, hidden subscriber counts). Data is never invented.

Do I need a proxy? Small runs work without one. For hundreds of videos or scheduled monitors, set a residential proxy in proxyConfiguration to avoid 429s.

Can I get more than 500 results per query? The actor paginates automatically up to maxResults (max 500 per input) and up to maxItems overall — combine queries for more.

Shorts? Yes — detected via duration/URL and flagged with isShort. Reel results from search are included automatically.

Integrations

Standard Apify dataset output: pull it via the API (/v2/acts/.../runs/.../dataset/items), connect Make / Zapier / n8n with the Apify modules, or call the actor through the Apify MCP server from Claude/ChatGPT — the dataset view and output schema are machine-readable. Combine with outputMarkdown: true to feed transcripts and comments straight into an LLM pipeline.

Support

Found a page shape that parses wrong? Open an issue with the video/channel URL and the field that came back empty — YouTube changes markup regularly and fixes ship fast.