YouTube Search Scraper

Type a keyword, get the search results as rows — with the numbers YouTube's own search page never shows you.

Exact view counts instead of "1.7M". The real publish date instead of "9 months ago". Like counts, channel subscriber counts, and the subtitle track if you want it. Videos, Shorts and live streams, from one keyword or a hundred.

What a row looks like

{
  "id": "kYfNvmF0Bqw",
  "url": "https://www.youtube.com/watch?v=kYfNvmF0Bqw",
  "title": "The Best Smartphone Camera 2025",
  "channelName": "Marques Brownlee",
  "channelUrl": "https://www.youtube.com/channel/UCBJycsmduvYEL83R_U4JriQ",
  "channelSubscriberCount": 20100000,
  "viewCount": 4821904,
  "likes": 79605,
  "lengthSeconds": 1043,
  "publishedDate": "2026-01-31T00:00:00.000Z",
  "publishedDateEstimated": false,
  "publishedTimeText": "7 months ago",
  "isShort": false,
  "keywords": ["smartphone", "camera test", "blind test"],
  "description": "Full blind camera test results...",
  "thumbnailUrl": "https://i.ytimg.com/vi/kYfNvmF0Bqw/hqdefault.jpg",
  "transcript": null,
  "captionTracks": [],
  "subtitlesLanguage": null
}

Two fields are worth calling out, because most scrapers quietly get them wrong:

• viewCount is exact. A listing card says 1.7M; this returns 1,748,203.
• publishedDate is the real date. Cards only say "9 months ago", and reversing that lands you on the wrong day. This actor reads the exact date from YouTube's own player data, and when it genuinely cannot, it says so with publishedDateEstimated: true instead of pretending.

Input

Field	Meaning
searchQueries	One or more keywords. Each is searched separately and the results are merged, de-duplicated
maxResults	Regular videos per keyword
maxResultsShorts	Shorts per keyword (0 to skip)
maxResultStreams	Live streams per keyword (0 to skip)
dateFilter	any, hour, today, week, month, year
lengthFilter	any, short (under 4 min), long (over 20 min)
sortingOrder	relevance, date, views, rating
hl / gl	Interface language and country — ko/KR returns what a Korean viewer sees
downloadSubtitles	Fetch the transcript and available caption tracks
proxyConfiguration	Apify Proxy; datacenter is enough for search

{
  "searchQueries": ["home workout", "meal prep"],
  "maxResults": 200,
  "maxResultsShorts": 0,
  "maxResultStreams": 0,
  "dateFilter": "month",
  "lengthFilter": "any",
  "sortingOrder": "relevance",
  "hl": "en",
  "gl": "US",
  "downloadSubtitles": false,
  "proxyConfiguration": { "useApifyProxy": true }
}

What it is good for

• Content research — what is ranking for a keyword right now, and how well it actually performs once you have exact views and likes.
• Creator and influencer shortlists — every result carries its channel and subscriber count, so a keyword sweep becomes a candidate list.
• Trend tracking — run the same keywords on a schedule and diff the view counts; real publish dates make the time series honest.
• Transcript mining — turn on subtitles and search what was actually said.

Honest limits

• YouTube search is capped, not infinite. Roughly a thousand results per keyword is where the result pages stop. Ask for more keywords rather than more depth on one.
• Like counts come from the watch data and can be hidden by the uploader. A hidden count comes back empty, never as 0.
• Members-only and age-restricted videos return the listing fields but not the full detail record.
• YouTube's terms restrict automated collection. The use you make of the data is yours.

More actors from this author

• YouTube Channel Search — type a topic, get matching channels with subscriber counts. $1 per 1,000 channels
• YouTube Channel Scraper — one channel to profile, every video and the comments
• YouTube Video Scraper — every video on a channel with tags, chapters and captions
• YouTube Comments Scraper — comments and reply threads from any video URL
• Encar Scraper — Korean used-car listings with accident history

Questions and problems

Open an issue on the Issues tab of this actor. Include the run ID — it is the fastest way to get a fix.