🎬 YouTube Scraper — exact counts, not rounded labels

Scrape YouTube videos, Shorts, live streams, channels, playlists, search results and comments — and get the real numbers.

Most YouTube scrapers hand you what's printed on the thumbnail grid: 18M views, 2 days ago, likes rounded to 306,000. This one opens each video and reads the actual figures:

	Typical scraper	🎯 This Actor
Views	18,000,000	18,592,498
Likes	306,000	306,738
Published	"2 days ago"	2026-09-10T07:29:55Z
Length	17:14 (off by a second)	00:17:14 · 1034 seconds
Category	—	Science & Technology
Tags	—	["iPhone Duo", "iPhone 18 Pro"]

On a single MKBHD video, the rounded figure is 344,000 views wrong. If your numbers end up in a report, a chart or a database, that gap matters.

Every row tells you how precise it is, in a dataPrecision field. Nothing is guessed, nothing is silently rounded.

---

⚡ Quick start

1. Click Try for free.
2. Type a search term, or paste a YouTube URL — a channel, a video, a playlist, a Shorts link, even a YouTube search page URL.
3. Set Maximum videos per source.
4. Hit Start ▶️

That's it. Results appear live in the Output tab and export to CSV, Excel, JSON or XML in one click.

Simplest possible input:

{
  "searchQueries": ["web scraping tutorial"],
  "maxResults": 50
}

A whole channel, videos and Shorts together:

{
  "startUrls": [{ "url": "https://www.youtube.com/@mkbhd" }],
  "maxResults": 200,
  "maxResultsShorts": 50
}

This month's long HD videos in a niche, best first:

{
  "searchQueries": ["home studio setup"],
  "maxResults": 100,
  "sortingOrder": "views",
  "dateFilter": "month",
  "lengthFilter": "over20",
  "isHD": true
}

---

📥 What you can feed it

You give it	You get back
🔍 A search term	Matching videos, with every filter YouTube offers
📺 A channel URL or @handle	Its videos, Shorts and past live streams
▶️ A video, Shorts or youtu.be link	That one video, fully detailed
📃 A playlist link	Every video in the playlist
🔗 A YouTube search-page URL	Exactly the results that page shows
💬 Any of the above + comments	Top-level comments attached to each video

Mix them freely — search terms and URLs can go in the same run. Bare video IDs (dQw4w9WgXcQ) and handles (@mkbhd) work too.

---

🏆 What makes it different

1. Exact numbers. Views, likes, publish timestamps and durations read from each video, not from the listing shelf. Verifiable in one run.

2. Fields nobody else ships. 📂 category · 🏷️ tags · 🌍 availableCountryCount (249 countries or 12?) · 💬 captionLanguages and whether they're human-written · 👨‍👩‍👧 isFamilySafe

3. Performance metrics, computed for you. 📈 engagementRate, likeRate, commentRate, viewsPerDay, daysSincePublished — plus viewsVsChannelMedian and isChannelOutlier, which answer the question a channel dump is usually for: which of these videos actually overperformed?

4. Sponsorship detection. 💰 hasSponsorSignals and sponsorSignals flag videos whose description reads like a paid placement, alongside YouTube's own isPaidContent disclosure.

5. One Actor, not four. Search, channels, Shorts, streams, playlists and comments are all here. No juggling separate tools and stitching their outputs together.

6. Honest about its limits. When YouTube withholds something, the row says so instead of inventing a value.

---

📤 Example output

One row per video (trimmed):

{
  "id": "Od6M0AXpcxQ",
  "url": "https://www.youtube.com/watch?v=Od6M0AXpcxQ",
  "title": "iPhone 18 Pro/Duo Impressions: Mogged",
  "type": "video",
  "dataPrecision": "exact",

  "viewCount": 18592498,
  "likes": 306738,
  "commentsCount": 21043,
  "date": "2026-09-10T07:29:55Z",
  "duration": "00:17:14",
  "durationSeconds": 1034,

  "category": "Science & Technology",
  "tags": ["iPhone Duo", "iPhone 18 Pro", "MKBHD"],
  "hashtags": ["#iPhone18"],
  "captionLanguages": ["en", "ja", "pt", "ru", "tr", "vi"],
  "hasManualCaptions": true,
  "availableCountryCount": 249,
  "isFamilySafe": true,
  "isPaidContent": false,
  "hasSponsorSignals": true,
  "sponsorSignals": ["check out"],

  "likeRate": 1.6498,
  "commentRate": 0.1129,
  "engagementRate": 1.7627,
  "viewsPerDay": 8810621,
  "daysSincePublished": 2.1,
  "channelMedianViews": 4884885,
  "viewsVsChannelMedian": 3.806,
  "isChannelOutlier": true,

  "channelName": "Marques Brownlee",
  "channelUrl": "https://www.youtube.com/@mkbhd",
  "numberOfSubscribers": 21200000,
  "channelTotalViews": 5609670895,
  "channelTotalVideos": 1849,
  "channelLocation": "United States",
  "isChannelVerified": true,
  "channelExternalLinks": ["http://twitter.com/MKBHD", "http://instagram.com/MKBHD"],
  "channelXUrl": "http://twitter.com/MKBHD",
  "channelInstagramUrl": "http://instagram.com/MKBHD",

  "text": "Full video description…",
  "descriptionLinks": [{ "url": "https://ridge.com/MKBHD", "text": "https://ridge.com/MKBHD" }]
}

Up to 89 fields per video. The Output tab groups them into three ready-made views — 🎬 Videos, 📈 Performance, 📺 Channel & links.

---

🎯 How precise are the numbers?

Every row carries a dataPrecision field, so nothing has to be taken on trust:

Value	What it means
✅ exact	Everything: exact view and like counts, full publish timestamp, category, tags, subtitle languages, country availability.
🟡 partial	The counts are still exact — views, likes, publish date, duration and every metric derived from them. YouTube declined to serve the classification details for this particular video, so category, tags and captions are absent rather than guessed.
⚪ approximate	YouTube's rounded listing labels only (18M views, 2 days ago). Produced by the Fast preset.

On Balanced and Detailed, every row comes back exact or partial — both with exact counts. A typical 60-video run lands around three quarters exact.

Presets

Preset	What it does	Use it when
🚀 Fast	Reads the listing only. Rounded counts, relative dates.	You just want an inventory of what exists.
⚖️ Balanced (default)	Opens each video for the real figures.	Almost always.
🐢 Detailed	Same data, gentler pace.	Very large runs.

---

🔎 Search filters

Every filter YouTube's own search page offers — and unlike that page, they combine:

• ↕️ Sort by — relevance · upload date · view count · rating
• 📅 Uploaded — last hour · today · this week · this month · this year
• ⏱️ Length — under 3 min · 3–20 min · over 20 min
• 📦 Type — videos · Shorts · channels · playlists · movies
• ⚙️ Features — HD · 4K · HDR · subtitles/CC · Creative Commons · live · 360° · 3D · VR180

So "HD videos over 20 minutes, uploaded this month, sorted by views" is one run, not four.

📆 There's also a separate date range (Published after / Published before) accepting 2026-01-31 or a relative form like 7 days. It works on channels and playlists too — not just search — and on Balanced/Detailed it's applied to the exact publish timestamp.

---

💬 Comments

Switch on Also scrape comments and set a maximum per video. Each comment carries the author, their channel ID and avatar, whether they're verified or the channel owner, like and reply counts, and whether the creator ❤️ hearted it.

Enabling comments also upgrades commentsCount on the video row from YouTube's rounded label to the exact total.

---

📱 Shorts

Shorts come back as ordinary rows with type: "shorts" and their /shorts/ URL. Set Maximum shorts per channel above zero to include them, or set Result type to Shorts to search them specifically.

Two honest notes: the Shorts shelf carries no length, so durationSeconds appears only when the video's own page answered; and it carries no publish label either, so a Short's date always comes from the video page.

📊 Want Shorts-specific depth? For posting cadence, audience signals, comments and creator avatars as their own datasets, use the dedicated YouTube Shorts Scraper — Growth & Audience Insights.

---

🔗 Creator links

Every row carries the creator's own links exactly as the channel lists them — the raw list, the labels they chose, and recognised profiles broken out into their own fields:

🐦 channelXUrl · 📸 channelInstagramUrl · 🎵 channelTiktokUrl · 👥 channelFacebookUrl · 💼 channelLinkedinUrl · 🎮 channelDiscordUrl · 🟣 channelTwitchUrl · 💻 channelGithubUrl · 🧡 channelPatreonUrl · ✈️ channelTelegramUrl

…plus channelExternalLinks and channelLinkTitles for everything else, including the creator's own website. This costs nothing extra — the channel is read anyway for the subscriber and view totals — so links are on every row in every mode.

📧 Looking for creator email addresses? That's a different job: finding a business address, confirming the domain actually accepts mail, and scoring how reachable a creator is. It has its own Actor — YouTube Channel Email & Sponsor Lead Scraper — which does it properly. This one deliberately stays out of it and gives you the links instead.

---

🍳 Recipes

📈 Find this month's breakout videos in a niche Search term + sort by view count + uploaded this month. Then sort the dataset by viewsPerDay.

🔬 Audit a channel's whole catalogue Channel URL with a high Maximum videos per source. viewsVsChannelMedian and isChannelOutlier show which videos beat that creator's own average — the ones worth studying.

💰 Build a sponsorship prospect list Search your niche, filter for hasSponsorSignals: true to find creators already running paid placements, then reach them through their own links. For verified business emails, pair with YouTube Channel Email & Sponsor Lead Scraper.

🌐 Check subtitle coverage before localising captionLanguages and hasManualCaptions show which videos already have human-written subtitles, and in which languages.

🏁 Track competitors Several channel URLs in one run, then schedule it weekly. viewsPerDay and engagementRate make the comparison fair across videos of different ages.

📱 Pull a channel's Shorts only Use the /shorts channel URL, set Maximum videos per source to 0 and Maximum shorts per channel to what you need.

---

❓ FAQ

Why do my view counts differ slightly from YouTube? YouTube counts continuously — a number read a minute later will be a little higher. What's guaranteed is that the figure is the exact one YouTube reported at the moment it was read, not a rounded label.

Do I need a YouTube account or cookies? No. 🔓 Everything scraped here is public.

What does it cost? You pay per video returned, and it's all-in — no separate charge for the run itself, no add-ons. Set a maximum charge on the run if you want a hard ceiling.

Can I get transcripts? The Actor tells you which subtitle languages exist for each video (captionLanguages, captionTracks), so you know what's available before pulling any.

What if a video can't be opened? Nothing is dropped silently. The row is still returned with whatever YouTube did provide, marked partial or approximate, and the run summary reports how many rows landed in each state.

What if I make a mistake in the input? You get a plain-English explanation on the run's status line — "'https://vimeo.com/1' is not a YouTube URL" — not a stack trace. The run stops in a few seconds without burning your budget.

Can I schedule it? Yes ⏰ — use Apify's scheduler. The Actor keeps no state between runs, so every run is independent and reproducible.

Can I call it from my own code? Yes. It works through the Apify API and the JavaScript/Python clients like any other Actor, and the dataset is available as JSON, CSV, Excel or XML.

---

🧰 More YouTube tools

Actor	Use it for
📧 YouTube Channel Email & Sponsor Lead Scraper	Creator business emails, deliverability checks and sponsor-readiness scoring.
📊 YouTube Shorts Scraper — Growth & Audience Insights	Shorts-only growth analysis, audience signals, comments and avatars.