Extract YouTube comments from a video URL
Paste a YouTube video URL and it returns every public comment as a row: the text, the author, the timestamp, the like count and the reply it belongs to. Reply threads are expanded in full, and the run ends with a CSV you can sort, filter or run through a classifier.
Get started with MacWorks on Mac and WindowsAbout this workflow
Version 1 · Updated Aug 11, 2026 · Published by Incredible
What this workflow does
The comment section under a YouTube video is often the most useful thing on the page, and it is the hardest part to actually read. YouTube loads comments in batches as you scroll, hides replies behind a control you have to click thread by thread, and starts the whole section over if you change the sort order. A few hundred comments is an afternoon of scrolling. Ten thousand is not something a person can do.
This workflow takes the video URL instead. It opens the comment section, loads it to the bottom, expands every reply thread, and returns one row per comment: the text exactly as written, the author display name, their channel handle and link, the timestamp, the like count, the reply count, and the parent comment for a reply so threads can be rebuilt. It also flags the comments that are pinned, hearted by the creator, or written by the video's own channel.
What comes back is raw material rather than conclusions. The text is kept as written, emoji, line breaks and all, because whatever you do next, sorting by likes to see what resonated, searching for the question people keep asking, or running the lot through a classifier, works better on the original than on something already tidied up. The run ends with a CSV in your Downloads folder.
How it works, step by step
Every run follows the same five steps in the same order. You paste the video URLs, choose the sort order and whether you want replies:
- 1Open the video and check comments are on. Opens each YouTube video URL you supplied and confirms the video is public and its comments are enabled.
- 2Set the sort order and load every comment. Chooses top comments or newest first, then keeps scrolling until YouTube stops adding more.
- 3Expand every reply thread. Opens each collapsed reply thread and keeps clicking through until no replies are left hidden.
- 4Read each comment. Reads the comment text, the author, the timestamp, the like count and the flags for pinned, hearted and creator replies.
- 5Save and check the CSV. Writes one row per comment into a CSV in Downloads and checks the count against the number YouTube displayed.
Every run follows these steps in order. If a step cannot be completed, the run stops and tells you what happened instead of guessing.
What you need to run it
A computer with Google Chrome or Microsoft Edge and the Incredible extension installed, plus a normal browser. You do not need a YouTube or Google account to collect the comments, because the workflow reads the same public comment section any viewer sees, and no Google Cloud project or API key either. A Google account is needed only if you want the finished file checked in Google Sheets. There is nothing to code and no APIs to connect.
Why teams run this with Incredible
Incredible reads the comment section the way a person would, which is why it can expand every reply thread in full rather than the first few the Data API attaches to each comment. The run is read-only by design: it never posts, replies, likes or reports anything. Every run repeats the steps it was taught in the same order, so the same columns come back every time, and it reports how many comments it collected against how many YouTube displayed rather than quietly returning a partial file. Incredible is GDPR compliant and none of your data is stored or used for training.
Trusted by teams at
Run it on your own computer
Incredible works in your own browser, doing the steps the way a person would. Getting there takes three steps.
Featured in
Compliance
GDPR compliant. None of your data is stored or used for training.
Frequently asked questions
How do I export YouTube comments to a CSV?
Open the video, load the comment section to the bottom, expand the reply threads and write each comment to a row. This workflow does that from the video URL and saves a CSV in your Downloads folder with the comment text, author, timestamp, like count and parent comment. The text field is quoted properly, so line breaks and commas inside a comment do not break the file.
Can I download all the comments from a YouTube video?
Every public comment, yes, including the replies. The number you get will often be slightly lower than the count YouTube displays, because comments held for review, caught by the creator's blocked-words filter or removed as spam are not visible to anyone but the creator. The run reports both numbers and the gap rather than leaving you to wonder.
Does it include replies as well as top-level comments?
Yes, and in full. It opens every collapsed thread and keeps clicking through until no hidden replies remain, recording the parent comment on each reply so you can rebuild the conversation. This is the main thing the official API struggles with: it attaches only the first few replies to each comment, so complete threads mean a separate call per parent.
How do I get YouTube comments without using the API?
By reading the comment section the way a viewer does rather than calling the Data API. That means no Google Cloud project, no API key and nothing deducted from the 10,000 quota units a day, which matters because there is no self-service way to buy more once you hit the ceiling.
What happens if comments are disabled on the video?
Nothing can be collected, because there is no public comment section to read. The video is recorded in the run summary with that reason rather than returning an empty file with no explanation, so a disabled video is never mistaken for a video nobody commented on.
Why is the number of comments lower than the count YouTube shows?
YouTube's displayed count includes comments that are not publicly visible: ones held for review, caught by the creator's blocked-words filter, or removed for spam. Only the creator can see those. The run reports what it collected alongside the displayed count so the difference is visible rather than hidden.
Can I collect comments from several videos at once?
Yes. Paste as many watch URLs, share links or video IDs as you like and the workflow works through them one at a time, keeping the video ID on every row so the comments stay separable in one file. A busy comment section takes a while, since it has to be loaded in batches at a human pace and every reply thread expanded.
Does it analyse the comments or just collect them?
It collects them. There is no sentiment scoring or summarising built in, and the text is kept exactly as written rather than cleaned up, which is deliberate: you should pick the classifier rather than inherit ours. Sorting by likes, searching for a repeated question, or running the file through your own analysis all work better on the original text.