YouTube Comments Scraper
Pricing
from $0.29 / 1,000 comments
YouTube Comments Scraper
Export public YouTube comments and replies to CSV, Excel or JSON. Collect top or newest comments, author details, likes and parent IDs from videos and Shorts. No API key. Streaming output, reply caps and transparent limits.
Pricing
from $0.29 / 1,000 comments
Rating
0.0
(0)
Developer
Inus Grobler
Maintained by CommunityActor stats
1
Bookmarked
20
Total users
2
Monthly active users
10 days ago
Last modified
Categories
Features
Share
Export public YouTube comments and replies to CSV, Excel or JSON. Paste video links, choose a comment limit, and collect full comment text, author details, likes, public reply totals and direct comment links. No YouTube API key or login is required.
Use it for audience research, product feedback, comment archives and datasets for your own analysis. Standard videos, Shorts URLs, youtu.be links and video IDs are accepted.
Who uses it?
- Creators and community teams: export audience questions and examine reply conversations.
- Brands and product teams: collect public feedback under selected reviews and launch videos.
- Researchers and analysts: build source-linked comment datasets for archiving or analysis in their own tools.
Quick start and input examples
- Open the Input tab.
- Add your YouTube video links or IDs and set Comments per video.
- Enable Include replies if you need conversations.
- Start the run. Open the dataset to preview rows or export them to a spreadsheet.
The ready-to-run example requests five real comments:
{"videoUrls": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],"maxCommentsPerVideo": 5}
For a larger sample with replies:
{"videoUrls": ["https://youtu.be/dQw4w9WgXcQ"],"maxCommentsPerVideo": 100,"includeReplies": true,"maxRepliesPerThread": 3,"commentSort": "top"}
What you get
- One comment or reply per row. Video details are included on comment rows; separate metadata rows do not consume your comment limit.
- Top or newest order. Select YouTube's top-comment ordering or newest-first feed.
- Controlled reply collection. Limit replies per thread so a busy conversation does not consume your whole sample.
- Early dataset delivery. Rows are saved in batches of up to 20 while collection continues. Saved batches remain available if a later source request fails.
- Multiple videos and duplicate protection. Repeated links to the same video are collected once. Saved comment IDs are checked when the same run restarts.
- Honest limits. The run summary shows each video's delivered count, source requests and stopping reason.
Input options
| Input | Default | Purpose |
|---|---|---|
videoUrls | One example video | Array of public video links or 11-character video IDs. |
maxCommentsPerVideo | 5 | Maximum rows per video, including replies. 0 skips collection. |
includeReplies | false | Include available public replies as individual rows. |
commentSort | "top" | "top" or "new" for newest first. |
maxRepliesPerThread | Omitted | Optional cap on replies to each top-level comment. 0 returns only top-level comments. |
useResidentialProxy | false | Allow bounded Apify residential recovery after direct access fails. |
The last three options are in Advanced comment options. Existing API clients can also use maxCommentPagesPerVideo to impose a smaller source-request cap. A low cap can stop before the requested row count, especially with replies.
Direct collection can be blocked from cloud IP addresses. The Actor tries another public YouTube comment endpoint after an empty direct-access failure. If all direct routes are blocked and the summary reports YOUTUBE_ACCESS_BLOCKED, retry later or explicitly enable useResidentialProxy. Recovery permits up to two residential sessions shared across the run and respects the remaining page and time limits. Residential routing adds platform usage cost to the account responsible for usage; the Pricing tab determines your customer charges. Recovery cannot guarantee access to every video.
Comment output
| Fields | Meaning |
|---|---|
comment_id, url, text | Native comment ID, direct YouTube comment link and full public text. |
video_id, video_url, video_title | Video identity and title when available. |
author_name, author_channel_id, author_url, author_handle | Public author details. |
like_count, reply_count | Source-provided counts. Public reply-button totals are retained when present. |
parent_comment_id, depth | Parent ID and known nesting depth; top-level comments have a null parent. |
published_text, published_at | Original age/date label; an absolute timestamp only when the source supplies one. |
is_pinned, is_hearted, author_is_channel_owner, verified | Source-provided indicators. |
channel_id, channel_name, channel_url, view_count | Available video context. |
scraped_at, dataInputSource, sourceVerified | Collection time and provenance. |
Illustrative output shape:
{"record_type": "comment","video_id": "dQw4w9WgXcQ","video_url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ","comment_id": "example-comment-id","url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ&lc=example-comment-id","text": "How was this video made?","author_name": "Example author","like_count": 0,"reply_count": 3,"parent_comment_id": null,"depth": 0,"published_text": "2 months ago","published_at": null,"dataInputSource": "web","sourceVerified": true}
Unknown values remain null. Known zero counts and false flags are preserved. Relative labels such as “2 months ago” are not converted into invented exact dates. Deleted authors, missing badges and short or emoji-only comments are handled without discarding otherwise valid text.
Pricing and coverage
$0.29 per 1,000 delivered comments or replies on every Apify plan, plus the existing $0.00005 Actor start event per GB of memory, with a minimum of one start event. At the default 256 MB, 100 comments cost $0.02905 and 1,000 cost $0.29005. See Pricing for the current rates. A reply costs the same as a top-level comment. There is no separate video-processing event. Set a maximum run charge in Console to bound event spending.
Collection is capped at 10,000 rows across a run and 15 minutes of source work, or the platform timeout minus shutdown headroom when shorter. Each remaining video receives a share of the remaining time. Source-page caps and spending limits can stop it sooner. The input accepts a higher per-video maximum for compatibility; the run-wide cap still applies.
YouTube's public feed can omit comments or stop before the video's displayed total. Top and newest feeds can differ, and replies depend on public thread availability. Results are a bounded collection of reachable comments, not a completeness guarantee. Restart protection uses saved IDs; a changing feed can still affect which additional comments are reachable.
A source failure after some rows were saved produces a partial summary. If every video returns no usable comments, the run fails clearly and keeps the dataset empty. Start charges can still apply. Private videos, restricted comments, active live chat, playlists and channel-wide discovery are outside this Actor's scope. Media files are not downloaded.
Python API example
import osfrom decimal import Decimalfrom apify_client import ApifyClientclient = ApifyClient(os.environ["APIFY_TOKEN"])run = client.actor("thescrapelab/Apify-YouTube-Comment-Scraper-2-0").call(run_input={"videoUrls": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],"maxCommentsPerVideo": 100,"includeReplies": True,"maxRepliesPerThread": 3,},max_total_charge_usd=Decimal("0.10"),)assert run is not Nonefor comment in client.dataset(run.default_dataset_id).iterate_items():print(comment["comment_id"], comment.get("parent_comment_id"), comment["text"])
Open Run summary for per-video counts and errors. Existing integrations retain analysisQuery, the separate EVIDENCE_ANALYSIS output, and the hidden providedRecords / providedDatasetId import interfaces. Imported records remain labeled sourceVerified: false; they do not establish live collection.
For support, open Issues with your run ID, a public video link and the expected result. Keep credentials out of issue reports.
Independently developed by TheScrapeLab. Not affiliated with YouTube or Google.