YouTube Comments Scraper

YouTube Comments Scraper gets public comments and replies from YouTube videos. Add one or more video URLs or video IDs, choose Top or Newest order, and set a result limit. The Actor saves one row for each comment or reply in a results table that you can download or use in other tools.

• Add up to 50 videos in one run.
• Get top-level comments, with optional replies.
• Keep comment IDs and reply links for easy matching.
• No Google API key, YouTube account, cookies, or browser setup is needed.

Independent tool: This Actor is not made, approved, or sponsored by YouTube or Google.

Quick start

1. Add at least one public YouTube video URL or 11-character video ID.
2. Choose Top comments or Newest first.
3. Set how many results you want. Turn on replies if you need them.
4. Start the Actor and open Comments and replies when it finishes.

This small example gets up to 20 top-level comments from one public video. It uses Top order and does not load replies. The Actor selects the connection and bounded safety settings automatically.

{
  "startUrls": ["https://www.youtube.com/watch?v=aqz-KE-bpKQ"],
  "sort": "TOP",
  "maxComments": 20,
  "includeReplies": false
}

Limits at a glance

The Actor accepts up to 10,000 saved results in one run. This is the total for all videos together. Top-level comments and replies both count. The number you enter is a maximum, not a promise: a video may have fewer public comments, or YouTube may stop showing more data.

Input setting	Default	Allowed maximum	What it controls
YouTube videos	—	50	Videos in one run.
Maximum total results	100	10,000	All saved comments and replies across all videos.
Top-level comments for each video	10,000	10,000	Top-level comments from one video. Replies do not count here.
Replies for each top-level comment	50	1,000	Replies from one comment when Include replies is on.
Maximum requests for each video	600	600	Managed safety ceiling for one video.
Maximum requests for the run	2,000	2,000	Managed safety ceiling across all videos.
Videos processed at once	1	4	Managed at one for predictable pagination and proxy use.
Maximum run time	1,800 seconds	1,800 seconds	Work time before the Actor saves its final report. This is 30 minutes.

The Actor also stops after 256 MiB of downloaded source data. This fixed safety limit cannot be changed in the Input form. If any safety limit is reached, the Actor keeps the rows already saved and marks the result as partial.

Choose only Maximum total results: the Actor uses direct access below 4,000 and Residential US from 4,000 through 10,000, while keeping the release-tested request, concurrency, and time ceilings. Residential traffic adds Apify platform cost. If replies are on, replies use part of the same result total.

Export YouTube comments and replies

Your results table (the Apify Dataset) contains one unique row for each saved top-level comment or reply. Main fields include:

• comment text, likes, and the public author name and channel when available;
• video, comment, thread, and parent comment IDs;
• a clear isReply value and the reply's link to its parent comment;
• the order you selected and the position seen during this run;
• public signals such as pinned, hearted, creator, and edited when the source makes them clear;
• the display time shown by YouTube, such as 2 hours ago.

The result keeps the time shown by YouTube, such as 10 days ago, in publishedTimeText. YouTube does not give this Actor the exact posting time, so publishedAt is empty instead of showing a guessed exact date.

Example result:

{
  "videoUrl": "https://www.youtube.com/watch?v=aqz-KE-bpKQ",
  "id": "ExampleComment01",
  "isReply": false,
  "text": "Helpful explanation. Thank you!",
  "likeCount": 7,
  "author": { "displayName": "Example Viewer" },
  "publishedTimeText": "2 hours ago",
  "publishedAt": null
}

Download the results table as JSON, CSV, or Excel in Apify. You can also read it through the Apify API or send it to another tool.

How charging works

You pay for each unique comment or reply that is saved. A reply is charged in the same way as a top-level comment. Duplicate results, failed saves, and results stopped before saving are not charged. Check the current price shown by Apify before you start, and use the maximum spending setting when you need a hard cost limit.

Why you may get fewer comments than expected

The number you enter is a maximum, not a promise. You may get fewer results because:

• the video has fewer public comments;
• comments are disabled, the video is unavailable, or access is restricted;
• replies are off or their separate limit is reached;
• the total result, request, time, data, or spending limit is reached;
• comments change while the Actor is working;
• YouTube blocks a request or changes the page data used by the Actor.

Open the run report (Run summary) after every run. It explains what happened for each video, why collection stopped, and whether the result is complete inside the chosen limits. Its technical record name is RUN_SUMMARY.

What “complete” means

COMPLETE_VISIBLE_SNAPSHOT means the Actor reached the end of the public comments it could see for the selected order during that run. It does not mean every comment that has ever existed.

The Actor does not promise:

• deleted, hidden, held, spam, private, or member-only comments;
• live chat or comments that need a login;
• a fixed order or an exact historical record;
• an exact posting date;
• that direct access will never be blocked;
• permanent compatibility if YouTube changes its public page data.

YouTube can add, edit, reorder, hide, or delete comments while a run is in progress. A pinned comment can also appear outside the normal Top or Newest order. Results are a current public snapshot, not a permanent record.

Input guide

Videos

startUrls accepts 1–50 values. You can use a raw 11-character video ID or a supported HTTPS YouTube watch, youtu.be, Shorts, or live video URL. Duplicate videos are processed once in the order first entered.

Order and result limits

• sort: TOP or NEWEST.
• maxComments: maximum total saved comments and replies across all videos. Allowed range: 1–10,000. Default: 100.
• maxTopLevelCommentsPerVideo: maximum top-level comments for each video. Allowed range: 1–10,000. Default: 10,000.
• includeReplies: turn reply collection on or off.
• maxRepliesPerComment: maximum replies for each saved top-level comment. Allowed range: 0–1,000. Default: 50.

Replies count toward maxComments and your charged results. They do not count toward maxTopLevelCommentsPerVideo.

Automatically managed safety limits

• maxRequestsPerVideo: page loads and retries for one video. Range: 1–600. Default: 600.
• maxTotalRequests: page loads and retries across the run. Range: 1–2,000. Default: 2,000.
• concurrency: videos processed at the same time. Range: 1–4. Default: 1.
• maxRunSeconds: Actor work time. Range: 30–1,800 seconds. Default: 1,800.

These bounded fields remain in the JSON API for compatibility, but the visual Input form hides them and keeps their stable defaults. More allowed work still may not return more comments. The run can stop earlier because of its source-data, result, reply, or maximum spending limit.

Connection selection is also automatic when proxyConfiguration is omitted: direct below 4,000 requested results, and Residential US from 4,000 through 10,000. A validation run reached exactly 10,000 unique NEWEST rows with the Residential setup; corrected Direct access was blocked around 4,780 rows on the same control, so the threshold includes a safety buffer. Residential traffic adds platform cost. A Residential session stays sticky for one video, is isolated to one Actor run, and rotates after a block or transient transport failure. An explicitly supplied legacy proxy object is still validated; enabled empty groups are repaired to Residential US.

Invalid values, unsupported URLs, custom proxy URLs, and unknown input fields are rejected before a YouTube request is made.

Empty or partial results

First open Run summary and find the result code for the video:

• COMMENTS_DISABLED: the source clearly says comments are off.
• ZERO_COMMENTS: the source clearly returned no visible comments.
• VIDEO_UNAVAILABLE or PRIVATE_OR_RESTRICTED: the video cannot be read in the supported public mode.
• AMBIGUOUS_EMPTY: the Actor received no comments but could not prove why.
• PARTIAL_LIMIT: a result, request, time, data, reply, or spending limit stopped collection.
• SOURCE_CHANGED or SOURCE_UNCLASSIFIED: the source data was not in a form the Actor could safely understand.

An empty results table alone does not prove that a video has no comments.

Privacy and allowed use

Public comment text, names, and channel details can still be personal data and user-created content. Collect only data you are allowed to use. Keep it secure, keep it only as long as needed, and handle valid correction or deletion requests.

Do not use this Actor for harassment, stalking, doxxing, surveillance, mass profiling, sensitive-trait guesses, spam, automated outreach, targeting minors, or bypassing private, member-only, login, location, age, or moderation controls. The Actor does not give you a license to republish comments.

See docs/PRIVACY-TERMS-AND-ACCEPTABLE-USE.md and docs/ABUSE-AND-OPERATIONS.md. These documents give practical risk guidance, not legal advice or a promise of legal compliance.

Troubleshooting

The run saved no rows

Open Run summary, check the code for each video, and confirm that the video and its comments are public in a signed-out browser. Do not assume that an empty table means zero comments.

The run stopped early

Check the stop reason and the result, request, time, data, reply, and maximum spending limits. Raise only the relevant user setting and stay within its shown maximum. Some data limits are fixed by the Actor and cannot be changed in the Input form.

The source failed or changed

Do not repeat a large batch. Try one small public video. If it fails again, send the safe support details below.

Support

For normal technical support, use Apify Issues for Cliqto Media (CliqtoMedia). No response-time promise is offered.

Send only:

• the Actor run ID and build ID;
• the UTC start time;
• the number of videos, selected order, limits, and reply setting;
• result codes, stop reasons, counters, and the safe error hash from the run summary.

Never send comment text, author data, screenshots with personal data, cookies, API keys, authorization headers, raw responses, request bodies, page tokens, session data, or proxy URLs. Do not place personal data in a public Issue. Use the publisher's private contact for privacy or security reports.

FAQ

Is this an official YouTube or Google tool?

No. It is an independent Actor. It is not made, approved, or sponsored by YouTube or Google.

Does it use the official YouTube Data API?

No. It reads the public comments data used by YouTube's website. The technical source name is InnerTube. This source is not a public, supported API.

Does this YouTube comment scraper need an API key?

No. The Actor does not accept your Google API key, account, OAuth token, or cookies.

Can I export YouTube comments from multiple videos?

Yes. Add up to 50 supported public video URLs or IDs in one run. Duplicate videos are processed once.

Can I download YouTube comments as CSV or Excel?

Yes. Open the Apify results table and download it as JSON, CSV, or Excel. Each comment or reply is one row.

Can it get every comment or recover deleted comments?

No. It gets the public comments visible during the run and inside your limits. It cannot recover deleted, hidden, held, private, or member-only comments.

Are replies included?

Only when includeReplies is on. Each parent comment has its own reply limit, and every saved reply counts toward the total and charged results.

Why is publishedAt empty?

YouTube gives this Actor display text such as 2 hours ago or 10 days ago. That text is rounded and changes with time. Turning it into an exact date would add accuracy that YouTube did not provide. The Actor keeps the original text in publishedTimeText and leaves publishedAt empty.

Can I use YouTube comments for analysis?

You can export the rows for research or analysis when you have the right to use the data. The Actor does not run sentiment analysis, profile people, or add data from other sources.