YouTube Channel Scraper - Creator & Competitor Research

Build a creator research shortlist or competitor publishing report from public YouTube data. Designed for US agencies, creator partnerships teams, and content strategists: compare channel stats, recent video topics, and Shorts without a YouTube login or API key.

The same Actor also collects live streams, playlists, channel-authored community posts, and available public websites/social profiles. Use optional publication windows and bounded pagination, then inspect the coverage summary to see exactly where collection stopped.

The Actor uses bounded HTTP requests to read public YouTube pages and parses YouTube's embedded public data. It returns only fields YouTube exposes publicly and marks unavailable fields as null.

Price: $3 per 1,000 successfully saved channels on the FREE Store tier, with discounts down to $2.55 per 1,000. One channel charge includes all selected content rows—there is no separate charge for each video, Short, playlist, or post. A one-channel run is approximately $0.00305 including the minimum start event.

Best for: competitor publishing trackers, creator research tables, channel watchlists, and recurring reports. The default reads one public content page per selected tab; opt-in pagination remains bounded and is not a full historical export of every upload.

US-first describes the buyer workflow, not a source filter. Channel URLs can be global; a channel's self-reported country is not audience geography. The Actor does not expose viewer location, US audience share, watch time, retention, revenue, sponsorship rates, or private YouTube Studio analytics. It does not provide persistent growth history or automatic change reports: comparisons between snapshots happen in your own spreadsheet or reporting system.

US agency creator research

Start with the YouTube Creator Research for US Agencies task. It uses two illustrative technology channels, MKBHD and iJustine. Replace them with your client's channels or a permission-appropriate creator/competitor shortlist; inclusion is not an endorsement, sponsorship relationship, or evidence of US audience reach.

The starter requests two channel rows, up to five videos and three Shorts per channel, in Fast mode. It reads at most one public page per selected tab with ten HTTP attempts per channel, including retries. It disables detailed-video enrichment, optional extra tabs, and proxies. Availability and request budgets can produce fewer rows; this is not a promise of sixteen content rows or complete upload history.

{
  "channelUrls": [
    "https://www.youtube.com/@mkbhd",
    "https://www.youtube.com/@ijustine"
  ],
  "searchKeywords": [],
  "mode": "fast",
  "maxVideosPerChannel": 5,
  "maxDetailedVideosPerChannel": 0,
  "includeShorts": true,
  "maxShortsPerChannel": 3,
  "includeLiveStreams": false,
  "includePlaylists": false,
  "includeCommunityPosts": false,
  "maxPagesPerSection": 1,
  "maxRequestsPerChannel": 10,
  "metadataProxyFallback": false,
  "proxyConfiguration": { "useApifyProxy": false }
}

At current FREE-tier prices, two successfully saved channels cost $0.00605 in listed Actor events, including one $0.00005 start event at 256 MB. At the lowest current discounted channel price, the same events cost $0.00515. Selected content is bundled into each channel event, not billed per video or Short. These are fee calculations, not measured operating costs or a profitability guarantee. Optional proxy traffic adds costs.

This starter is a saved configuration, not newly live tested as part of this documentation release. It does not run automatically, create a schedule, or save a historical comparison. The existing MKBHD example and default Actor input are unchanged.

CSV comparison and client reporting

1. Start with direct channel URLs, Fast mode, and small limits. Read the RUN-SUMMARY coverage record before interpreting results. A successful run can still have an unavailable tab, capped collection, or uncertain dates.
2. Export the full dataset as CSV or JSON, retaining recordType, channelUrl, and scrapedAt. The Channels/Videos views select columns; they do not automatically separate record types. The existing Videos view omits channelUrl, so use the full dataset or an explicit field selection containing it for a reliable join.
3. Separate recordType = channel and recordType = video in your reporting tool. Join video rows to channel rows by channelUrl; do not join on display name. Deduplicate content by videoId, and separate contentType = video, short, and live_stream.
4. Compare public subscribers, sampled video topics, views, and durations. Label results as a bounded sample, retaining requested limits, coverage, and observation time with the report. Missing values stay unknown, not zero.
5. If you choose to run again later, retain the earlier export and compare the same channel IDs/URLs and video IDs. A new snapshot is not automatically a growth report. Repeated collection can incur charges; no schedule is created by this example.

Research question	Existing fields	Interpretation
Which channel is this?	channelId, channelUrl, handle	Use stable identity, not a mutable display name.
How large is its public subscriber count?	subscriberCount, subscriberCountNumber	Abbreviated labels produce approximate parsed numbers, not private exact counts.
What did this bounded sample cover?	recordType, contentType, videoId, RUN-SUMMARY	Fewer rows do not prove fewer uploads; tabs and requests may be capped or unavailable.
What topics and formats appear?	videoTitle, durationSeconds, isShorts	Compare regular videos and Shorts separately; keep missing durations unknown.
How many public views were shown?	viewCount, viewCountNumber, scrapedAt	Counts are observations at collection time, not views earned during your reporting period.
When was content published?	sourceDateText, publishedAtPrecision, publishedAtEarliest, publishedAtLatest, publicationWindowMatch	Relative labels and uncertain window matches do not establish exact posting time or cadence.
Is detailed contact/engagement evidence available?	websiteLinks, socialProfiles, likeCountNumber, commentCountNumber	Requires Detailed mode where supported; not present in this Fast starter and not guaranteed.

For an external snapshot comparison, calculate subscriber or same-video view deltas only when both observations have non-null counts and matching identities. Keep the original displayed labels and timestamps: rounded counts can conceal small changes, video removals and count corrections can produce negative deltas, and two channels' small recent samples are not directly comparable lifetime performance. Do not calculate engagement or sponsorship value from missing likes/comments.

Use the creator comparison checklist for a client handoff. CSV exports and coverage evidence support your own reporting workflow; the Actor does not generate an automatic scored shortlist, contact verification, or a rendered client report.

Track a competitor channel

This low-cost first run creates one channel row plus up to five recent videos, five Shorts, and three community posts:

{
  "channelUrls": [
    "https://www.youtube.com/@mkbhd"
  ],
  "mode": "fast",
  "maxVideosPerChannel": 5,
  "includeShorts": true,
  "maxShortsPerChannel": 5,
  "includeCommunityPosts": true,
  "maxCommunityPostsPerChannel": 3
}

Schedule the same input daily or weekly, then compare subscriberCountNumber, totalVideoCountNumber, recent titles, views, and publishing dates in your spreadsheet or dashboard. Add more direct channel URLs to build a watchlist.

Enrich a creator profile

Switch to detailed mode when you also need public websites, classified social profiles, About-page details, and exact engagement fields for selected recent videos:

{
  "channelUrls": ["https://www.youtube.com/@mkbhd"],
  "mode": "detailed",
  "maxVideosPerChannel": 5,
  "maxDetailedVideosPerChannel": 2,
  "includeShorts": true,
  "maxShortsPerChannel": 5,
  "includeLiveStreams": true,
  "maxLiveStreamsPerChannel": 5,
  "includePlaylists": true,
  "maxPlaylistsPerChannel": 5,
  "includeCommunityPosts": true,
  "maxCommunityPostsPerChannel": 5
}

Export the results as JSON, CSV, Excel, XML, or HTML, or consume them through the Apify API, schedules, webhooks, Make, Zapier, n8n, and other integrations.

If direct detailed metadata is withheld by a confirmed YouTube automation check, you can explicitly set "metadataProxyFallback": true. This uses Apify Residential only for compact player metadata, not full pages or video/audio files. It requires Residential access and adds proxy usage costs. It is disabled by default and must not be combined with full-page proxyConfiguration. Availability is still source-dependent; inspect RUN-SUMMARY.metadataFallback and detailed failures rather than assuming every field was returned.

What it extracts

Channel rows

• Channel URL, channel ID, name, and handle
• Direct public URLs for the Videos, Shorts, Live Streams, Playlists, and Community tabs
• Subscriber count as displayed and as a parsed number
• Total video count as displayed and as a parsed number
• Public channel description with contact details redacted
• Avatar and banner image URLs when available
• Verified-channel flag
• Extraction timestamp
• In detailed mode: total channel views, join date, country, named website links, and classified social/community profiles, including Facebook, Instagram, LinkedIn, X, YouTube, TikTok, Reddit, Twitch, Threads, and Discord
• The backward-compatible socialLinks array still contains all accepted public external URLs; email addresses and email links are excluded

Video, Shorts, and live-stream rows

• Channel URL and channel name
• Video URL and title
• View count as displayed and as a parsed number
• Duration in seconds and formatted text
• Relative published date shown by YouTube
• sourceDateText, publishedAt, publishedAtPrecision, earliest/latest publication bounds, and publicationWindowMatch. Relative ages are approximate intervals, never exact timestamps; missing dates remain unknown.
• Detailed mode can retain a video-ID-verified calendar date from the main watch page when the player timestamp is unavailable. This stays day-level precision; it is not an invented upload time.
• Thumbnail URL
• Content classification as video, short, or live_stream
• Live status when YouTube exposes it
• Extraction timestamp
• In detailed mode for the selected latest videos: exact public views, likes, description, tags, category, exact publish date, and public comment count when YouTube exposes a number

Playlist rows

• Playlist URL, ID, title, thumbnail, and public video count
• Channel URL and channel name
• Extraction timestamp

The Playlists tab may also contain show-style collections. These are saved as playlist rows only when their public destination confirms the same playlist ID; unrelated show or recommendation cards are excluded.

Community-post rows

• Post URL, ID, channel-authored public text, thumbnail or attachment URL
• Published-date text and public like/comment counts when YouTube exposes them
• The same publication precision and window-match fields as video rows
• Channel URL and channel name
• Extraction timestamp

Fast mode collects selected content grids efficiently and leaves detailed-only fields as null, empty arrays, or an empty social-profile object. Detailed mode enriches the About page and a bounded number of normal video pages. If one optional tab or video page cannot be read, the Actor still saves the available records instead of fabricating data.

Output dataset

One run can write four record types to the default dataset:

• The Channels view shows channel-level records.
• The Videos view shows normal videos, Shorts, and live-stream records.
• The Playlists view shows playlist records.
• The Community posts view shows channel-authored post records.
• Every row has an explicit recordType field for reliable filtering.

RUN-SUMMARY in the default key-value store reports each requested tab's inspected pages, rows seen/selected/saved, excluded rows, date uncertainty, remaining-page signal and stop reason. Failed detailed requests carry fixed stage/category codes and, when available, a recognized public playerStatus such as LOGIN_REQUIRED; never raw source errors, response bodies, anonymous visitor values, API keys or proxy URLs. This report is separate from the dataset and adds no channel charge. Disabled tabs are omitted. complete means a recognized feed ended without a cap, parsing/fetch failure or uncertain date eligibility; it is not a guarantee that YouTube exposed all content or every optional field. A successful run can still have incomplete coverage.

Bounded publication-window collection

{
  "channelUrls": ["https://www.youtube.com/@mkbhd"],
  "mode": "fast",
  "maxVideosPerChannel": 25,
  "maxPagesPerSection": 3,
  "maxRequestsPerChannel": 15,
  "publishedAfter": "2026-09-01",
  "publishedBefore": "2026-09-30"
}

Date-only inputs include the entire UTC day. Timestamp inputs must include a timezone. Videos, Shorts, live streams and community posts are excluded only when their known date or approximate interval is wholly outside the window. Unknown dates and boundary overlaps stay in the output with publicationWindowMatch: "uncertain"; filter these explicitly if your workflow requires certain matches. Upcoming streams may have unknown publication dates. Playlists have no reliable public publication date here and are not date-filtered.

Date filtering does not assume chronological ordering or stop at the first old card: pinned and reordered items can appear later. It does not guarantee a target number of matching rows. Detailed-mode dates may exclude a previously uncertain selected row; the scraper does not perform unbounded refill requests. The existing channel charge applies when channel metadata is saved, even if no content matches your window.

Verified channel sample

This shortened sample comes from a successful public Actor run:

{
  "channelUrl": "https://www.youtube.com/@mkbhd",
  "channelName": "Marques Brownlee",
  "handle": "@mkbhd",
  "subscriberCount": "21M subscribers",
  "subscriberCountNumber": 21000000,
  "totalVideoCount": "1.8K videos",
  "totalVideoCountNumber": 1800,
  "isVerified": true,
  "scrapedAt": "2026-06-22T07:56:07.305Z"
}

Verified video sample

{
  "channelUrl": "https://www.youtube.com/@mkbhd",
  "channelName": "Marques Brownlee",
  "videoUrl": "https://www.youtube.com/watch?v=WOzcFkld6_g",
  "videoTitle": "The Most Interesting Displays In The World!",
  "viewCount": "2.3M views",
  "viewCountNumber": 2300000,
  "durationSeconds": 957,
  "durationFormatted": "15:57",
  "publishedDate": "5 days ago",
  "isShorts": false,
  "scrapedAt": "2026-06-22T07:56:11.022Z"
}

Counts, titles, thumbnails, and relative dates can change when YouTube updates the page.

Input

Field	Type	Default	Description
channelUrls	array	One sample channel	Up to 50 full YouTube channel URLs or @handles
searchKeywords	array	Empty	Up to 10 optional keywords used to discover channels
mode	string	fast	fast for low-request monitoring or detailed for About and selected video-page fields
maxChannels	integer	1	Maximum channels scraped per search keyword, from 1 to 50
maxVideosPerChannel	integer	1	Maximum selected rows from inspected public Videos pages, from 1 to 100
maxPagesPerSection	integer	1	Initial page plus public continuations, 1–5 pages per selected content tab
maxRequestsPerChannel	integer	30	Shared source HTTP-attempt cap, 1–50, including retries, redirects, initial/About/tab/detail/player and continuation requests
publishedAfter / publishedBefore	string	Disabled	Inclusive UTC dates or timezone-qualified timestamps; uncertain dates stay visible
maxDetailedVideosPerChannel	integer	1	Detailed mode only: enrich the first 0 to 5 saved video rows per channel
includeShorts	boolean	false	Read the public Shorts tab and save Shorts as separate video records
maxShortsPerChannel	integer	10	Maximum Shorts saved per channel, from 1 to 50
includeLiveStreams	boolean	false	Read the public Live tab and save past, upcoming, or active streams
maxLiveStreamsPerChannel	integer	10	Maximum live-stream rows saved per channel, from 1 to 50
includePlaylists	boolean	false	Read the public Playlists tab and save playlist records
maxPlaylistsPerChannel	integer	10	Maximum playlists saved per channel, from 1 to 50
includeCommunityPosts	boolean	false	Read the public Posts tab and save channel-authored community posts
maxCommunityPostsPerChannel	integer	10	Maximum community posts saved per channel, from 1 to 50
proxyConfiguration	object	Direct	Optional Apify Proxy, country, custom-proxy, or direct settings

Provide at least one channel URL, handle, or search keyword. Direct channel inputs are more predictable than keyword search. Search results vary by region and ranking. Fast mode handles at most 50 unique channels per run. Detailed mode handles at most 10 channels and enriches at most 5 video rows per channel.

Common workflows

Monitor selected channels

Use direct channel URLs, schedule repeated runs, and compare subscriber counts, video counts, and latest-video rows over time.

Build a creator research table

Use detailed mode to collect public channel size, total views, country, websites, classified social profiles, descriptions, and recent video engagement for a defined set of channels.

Track competitor publishing

Compare latest titles, view counts, durations, and relative publish times across competing channels in the same niche.

Create recurring reports

Send dataset rows to a spreadsheet, warehouse, dashboard, or workflow tool through Apify integrations.

Pricing

This Actor uses Pay Per Event pricing.

Event	Price
Actor start	$0.00005 per GB of memory
Each successfully saved channel-scraped channel	$0.003, with Store-tier discounts down to $0.00255

The Actor defaults to 256 MB of memory and can be raised to 1 GB for larger batches. Actor-start billing uses a minimum of one event, so the startup charge remains approximately $0.00005 per run at the default memory. All selected content rows are included in the channel charge—there is no extra per-video, per-Short, per-playlist, or per-post event fee. A one-channel run on the FREE Store tier is therefore approximately $0.00305 before any applicable account-level charges.

Failed channel extractions and duplicate channel aliases are not charged as channel-scraped events. When a maximum-cost limit is reached, the Actor completes the bounded available content bundle for a successfully saved current channel, then skips queued channel work. A rejected channel save does not trigger its content requests. This event-charge limit is not a hard cap on platform execution, proxy or build costs.

Limits and reliability

• YouTube changes its page structure regularly. Select fields may temporarily become unavailable.
• Subscriber counts can be hidden or abbreviated.
• Search results depend on region and YouTube ranking.
• Fast runs are capped at 50 unique channels. Detailed runs are capped at 10 channels and 5 enriched video pages per channel. Requests are sequential and use bounded retries.
• Optional Shorts, Live, Playlists and Posts share one per-channel attempt budget. They use one page by default and follow at most five pages when explicitly requested. A channel may not expose every tab; a missing/unsupported tab is incomplete, not a confirmed empty feed.
• Pagination uses public browse configuration and scoped channel-feed continuation tokens. It stops at row/page/request limits, repeated tokens, no progress, unavailable data or ambiguous response structure. It never follows recommendations, comments or engagement-panel continuations.
• Each channel's source collection has a 240-second deadline; each request is limited to 30 seconds or the remaining deadline, and each decoded response to 8 MiB. These are defensive bounds, not a guaranteed run-time, memory or spending cap. Searches have a separate four-attempt/120-second budget per keyword.
• Additional pages and detailed fields require more source requests. Channel event prices are unchanged; choose modest row/page/request limits for recurring watchlists. Optional proxies and account-level charges may add costs.
• Shorts detection prefers YouTube's explicit /shorts/ route. Duration is only a fallback because Shorts can now be up to three minutes and ordinary videos can be shorter than one minute.
• If a channel page succeeds but its Videos tab is unavailable, the Actor saves and charges the channel metadata row without fabricating video rows.
• Public like and comment totals are not present in every YouTube page payload. When YouTube exposes a label without a number, count fields remain null.
• Community output contains only posts authored by the selected public channel. It does not collect commenter identities or comment text.
• Detailed fields are public page data, not private analytics, and can be hidden or changed by YouTube or the channel owner.
• An anonymous cloud player request may return LOGIN_REQUIRED even when the public watch page is readable. The Actor retains verified watch-page fields and reports missing details; it does not use account login cookies, invent category/tags, or automatically enable a paid proxy. A proxy is optional and does not guarantee access.
• A confirmed "not a bot" player response is reported as source-automation-check. Further player API calls in that channel session stop, while available watch-page fields and other bounded public content collection remain eligible. This does not mark missing details as complete. A new channel budget is independent; generic sign-in or video-specific restrictions are not treated as a session-wide automation block.
• The opt-in metadata fallback allows one Residential attempt per selected detailed video, at most five per channel. It stops paid attempts for that channel after a fallback failure. Compact responses are bounded to 32 KiB each and 64 KiB of successfully read decoded bodies per channel, with no redirects. Network overhead, provider setup and bytes received before an oversized response is cut off mean these are not guaranteed traffic or spending caps. A proxy failure never enables another provider, full-page proxying, login cookies or extra retries.
• Requests within one channel collection reuse an anonymous browser-header session and, when a proxy was explicitly selected, the same proxy session ID. Proxy setup is deadline-bounded; a failed selected proxy does not silently switch to a direct request. Provider session expiry or interruption can still change the connection.
• External-link classification recognizes Facebook, Instagram, LinkedIn, X/Twitter, YouTube, TikTok, Reddit, Twitch, Threads, and Discord; other accepted HTTP(S) links are returned as websites.
• Email addresses and mailto: links are not collected. Email addresses and phone numbers found in public descriptions are redacted.
• The Actor reads public pages only and does not access YouTube Studio, private analytics, account data, or private videos.

API example

curl -X POST "https://api.apify.com/v2/acts/fascinating_lentil~youtube-channel-scraper/runs?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "channelUrls": ["https://www.youtube.com/@mkbhd"],
    "searchKeywords": [],
    "mode": "detailed",
    "maxChannels": 1,
    "maxVideosPerChannel": 3,
    "maxDetailedVideosPerChannel": 1,
    "includeShorts": true,
    "maxShortsPerChannel": 3,
    "includeLiveStreams": true,
    "maxLiveStreamsPerChannel": 3,
    "includePlaylists": true,
    "maxPlaylistsPerChannel": 3,
    "includeCommunityPosts": true,
    "maxCommunityPostsPerChannel": 3,
    "proxyConfiguration": {"useApifyProxy": false}
  }'

Responsible use

Use this Actor only for lawful collection of publicly available information. You are responsible for complying with YouTube's terms, copyright rules, privacy laws, and regulations that apply to your use case.

Do not use the output for spam, harassment, profiling, or unlawful collection of personal data. This Actor is an independent tool and is not affiliated with, endorsed by, or sponsored by YouTube or Google.

License

Apache-2.0.