YouTube Research & Transcript Scraper

Search YouTube, export channel video lists, enrich selected videos, and collect transcripts without setting up the YouTube Data API.

This Apify Actor is built for YouTube research workflows where you do not want to scrape everything at the most expensive level. Start with broad discovery, shortlist the videos that matter, then run metadata enrichment or transcript extraction only on that smaller set.

Best for

• YouTube keyword research and topic mapping
• competitor and creator channel monitoring
• content audits for brands, agencies, and media teams
• transcript collection for LLM, RAG, summarization, and qualitative research pipelines
• building structured YouTube datasets from search results, channel pages, and known video URLs

How it works

The Actor accepts three input types. You can use one, two, or all three in the same run.

Input	Use it when	Result source
searchQueries	You want to discover videos by topic or keyword	YouTube search results
channelUrls	You want recent videos from one or more channels	Channel /videos pages with RSS fallback when possible
videoUrls	You already know the exact videos to process	Direct video metadata paths with fallback metadata extraction

Rows are deduplicated by videoId, so the same video is only pushed once even if it appears in multiple inputs.

At least one of searchQueries, channelUrls, or videoUrls must contain a value. Empty input is rejected so the Actor does not create a misleading dataset row or charge for a helper item.

Recommended workflow

1. Discover videos cheaply

Use searchQueries or channelUrls first. Keep scrapeDetails and includeTranscript off while you are still exploring.

{
  "searchQueries": ["ai workflow automation", "youtube competitor analysis"],
  "maxResults": 50
}

This gives you a clean shortlist with titles, URLs, channels, thumbnails, rough publish text, durations, view counts when available, and descriptions when present in the search result.

2. Review and shortlist

Filter the dataset outside the Actor. Pick only the videos you actually need for deeper work.

Useful shortlist signals:

• topic relevance from title and description
• creator or company from channelName
• popularity from viewCount
• freshness from publishedText or publishedAt
• video length from duration or durationSeconds

3. Enrich selected videos

Use videoUrls with scrapeDetails when you need stronger metadata for specific videos.

{
  "videoUrls": [
    "https://www.youtube.com/watch?v=XVv6mJpFOb0",
    "https://youtu.be/dQw4w9WgXcQ"
  ],
  "scrapeDetails": true
}

scrapeDetails may improve or fill:

• publishedAt
• category
• description
• viewCount

It is best used after shortlisting because it performs extra requests per video.

4. Collect transcripts only when needed

Use includeTranscript for videos where you actually need text, timestamps, or LLM-ready content.

{
  "videoUrls": [
    "https://www.youtube.com/watch?v=XVv6mJpFOb0"
  ],
  "scrapeDetails": true,
  "includeTranscript": true,
  "transcriptLanguage": "en"
}

When a transcript is available, the row includes timestamped transcript segments and a combined plain-text transcript. If YouTube does not provide captions for the video, or the captions cannot be fetched, the Actor still returns the video row without transcript fields.

Input reference

Field	Type	Default	Description
searchQueries	array of strings	empty	YouTube search keywords. Each query runs separately and can return up to maxResults videos. Best for discovery and SEO or market research.
channelUrls	array of strings	empty	YouTube channel inputs. Supports @handle, UC channel IDs, and common youtube.com channel, c, and user URLs. Returns recent public videos; RSS fallback is used when the channel page does not expose rows.
videoUrls	array of strings	empty	Exact YouTube videos to process. Supports 11-character video IDs and common watch, shorts, embed, live, and youtu.be URL formats. Best for enrichment and transcripts.
maxResults	integer	50	Maximum videos per search query or channel. It does not multiply direct videoUrls; each provided video URL is processed once.
scrapeDetails	boolean	false	Fetches richer metadata for each row. Use on shortlists or smaller runs.
includeTranscript	boolean	false	Attempts transcript extraction for each video. Use on targeted runs because this is the heaviest mode.
transcriptLanguage	string	en	Preferred transcript language code, such as en, es, fr, de, ja, or pt. If that language is unavailable, the Actor can fall back to the first available caption track.

Input examples

Search by keyword

{
  "searchQueries": ["supply chain automation"],
  "maxResults": 25
}

Export latest channel videos

{
  "channelUrls": ["https://www.youtube.com/@freecodecamp"],
  "maxResults": 100
}

Process known videos

{
  "videoUrls": [
    "https://www.youtube.com/watch?v=XVv6mJpFOb0",
    "https://youtu.be/PXMJ6FS7llk"
  ],
  "scrapeDetails": true
}

Transcript run for a shortlist

{
  "videoUrls": [
    "https://www.youtube.com/watch?v=XVv6mJpFOb0"
  ],
  "includeTranscript": true,
  "transcriptLanguage": "en"
}

Mixed discovery run

{
  "searchQueries": ["ai sales outreach"],
  "channelUrls": ["https://www.youtube.com/@HubSpot"],
  "maxResults": 30
}

Output fields

Each dataset item is one YouTube video row. The Actor does not write helper rows for empty input.

Core fields

Field	Type	Description
recordVersion	string	Output contract version, currently 1.0.
enrichmentLevel	string	base, detail, or transcript. Shows how far the row was enriched.
videoId	string	YouTube video ID.
title	string	Video title.
url	string	Canonical YouTube watch URL.
channelName	string	Channel or author name when available.
channelId	string	YouTube channel ID when available.
channelUrl	string	Channel URL when available.
viewCount	integer	View count when available. May be 0 when the source does not expose it.
duration	string	Human-readable duration from listing pages when available.
durationSeconds	integer	Duration in seconds when available.
publishedText	string	Relative publish text from listing pages, such as 2 weeks ago, when available.
publishedAt	string	Publish date when available. Detail mode can improve this field.
description	string	Search snippet, RSS description, or fuller video description depending on source and enrichment.
thumbnailUrl	string	Video thumbnail URL.
category	string	Video category when detail metadata is available.
isLive	boolean	Whether the source marks the video as live content.
source	string	Source path used for the row: search, channel, or detail.
scrapedAt	string	ISO timestamp when the row was created.

Transcript fields

Transcript fields appear only when includeTranscript is true and captions are successfully returned.

Field	Type	Description
transcript	array	Timestamped caption segments. Each segment has text, start, and duration.
transcriptLanguage	string	Language code of the transcript actually returned.
transcriptText	string	Full transcript joined into one plain-text string.

Example transcript segment:

{
  "text": "Welcome back to the channel.",
  "start": 12.4,
  "duration": 3.2
}

Standby API

The Actor includes a Standby API for small interactive requests. The same validation rules apply as normal runs.

Endpoint	Method	Use
/	GET	Readiness check
/search?query=python%20automation&maxResults=10	GET	Search videos
/channel?url=https://www.youtube.com/@freecodecamp&maxResults=10	GET	List recent channel videos
/video?url=XVv6mJpFOb0	GET	Fetch one direct video
/run	POST	Run the normal Actor input JSON through Standby

Limits and practical notes

• Transcripts are not guaranteed. They depend on whether YouTube exposes captions for the video and whether those captions can be fetched.
• includeTranscript can still return a valid video row without transcript fields.
• Search and channel rows may have lighter metadata than direct detail rows.
• maxResults applies per search query and per channel URL.
• Channel scraping works best with public channels and common YouTube URL formats.
• Very large transcript runs are slower and more expensive than discovery runs. Shortlist first when possible.
• YouTube page structure and availability can change. If a source path fails for a specific video or channel, try the most direct input type, especially videoUrls for known videos.

Pricing model

The Actor is designed for tiered pay-per-event use:

1. base video rows for search and channel discovery
2. detailed video rows when scrapeDetails is enabled
3. transcript-ready rows when transcript extraction succeeds

This is why the recommended workflow is to discover broadly first, then run detail or transcript modes only on a shortlist.

Check the Apify Store pricing panel for the current event prices before running large jobs.

Why use this Actor

This Actor is focused on research, not just bulk scraping. It separates discovery, detail enrichment, and transcript extraction so you can control speed, dataset size, and cost.

Use it when you need structured YouTube data for market research, creator research, competitor monitoring, content strategy, or LLM-ready transcript workflows without maintaining your own YouTube scraping stack.