Steps
- Copy the YouTube video URL from the address bar — anything like
https://www.youtube.com/watch?v=abc123XYZworks. - Paste it into SubExtract's video comments tool.
- Click Extract. The tool pulls the comment thread and renders it in a sortable table.
- Review and filter — sort by likes to surface top comments, or filter by keyword to narrow the dataset.
- Click Download CSV. The file lands in your downloads folder, ready to open in Excel, Google Sheets, or any analysis tool.
The whole flow takes under a minute for a typical video. Larger comment threads (thousands of comments) take longer to fetch but still complete in a single click.
What's in the CSV
Each row is one comment. The columns are:
- author — display name of the commenter as shown on YouTube.
- comment_text — the full text of the comment, including emoji and line breaks (preserved as
\ninside the cell). - like_count — number of likes/thumbs-up the comment received.
- reply_count — number of replies posted under the comment. Top-level comments only; replies have
0here. - parent_comment_id — empty for top-level comments; populated with the parent's ID when the row is itself a reply. Use this to reconstruct threads.
- timestamp — ISO 8601 timestamp of when the comment was posted.
The structure is flat — replies sit alongside top-level comments, linked by parent_comment_id. Sort or pivot on that column to rebuild the thread hierarchy when you need it.
Use cases
Sentiment analysis. Drop the CSV into a spreadsheet, classify each row as positive/neutral/negative with a quick LLM pass, then chart sentiment over time using the timestamp column. Useful for product launches, controversial videos, or measuring audience reaction to a creator's pivot.
Audience research. Filter for comments with high like counts — those are the ones the audience already endorsed. Reading the top 50 most-liked comments on a competitor's video tells you more about what their audience actually cares about than any keyword tool.
Content ideas. Question-style comments are gold. Filter for rows containing ? and you'll surface the exact things viewers want answered next. That's the next video, the next blog post, the next email.
Customer feedback mining. For brands and SaaS founders running YouTube ads or product demos, comments are unfiltered customer voice. Export, tag for objections / feature requests / praise, and feed the patterns into your roadmap or messaging.
Competitive teardowns. Pair comment exports with the channel videos extractor — pull the top 20 videos on a competitor's channel, export comments for each, and run a single big analysis to see what their audience consistently asks for, complains about, or praises.
Frequently asked questions
Is there a limit per video? The free tier caps at a few hundred comments per video — enough for sentiment sampling on most videos. Paid tiers raise the ceiling significantly, and the upper plans pull the full comment thread on high-engagement videos with tens of thousands of comments. If a video has more comments than your tier allows, the tool returns the top comments first (by relevance, the same default YouTube uses).
Are replies included?
Yes — replies are returned as their own rows with parent_comment_id set to the top-level comment's ID. To work with thread structure in Excel or Sheets, sort by parent_comment_id and you'll see each thread grouped together. Some tools omit replies by default to save quota; SubExtract includes them in the standard export.
What happens with deleted or hidden comments? Deleted comments don't appear in the export — YouTube's API doesn't expose them. Comments held for review by the creator's moderation settings also won't be returned. Hidden users (channels the creator has blocked) are filtered out at the platform level. What you get is what's publicly visible on the video at the moment of extraction.
Will the CSV open cleanly in Excel and Google Sheets? Yes. The file is UTF-8 encoded with quoted fields, so emoji, non-Latin characters, commas inside comment text, and newlines all parse correctly. In Excel, use Data → From Text/CSV and select UTF-8 if you see garbled characters from a direct double-click open. Google Sheets handles it natively via File → Import.
Does the format work with NVivo or other qualitative analysis tools?
NVivo, MAXQDA, and Atlas.ti all import CSV with column headers — point them at the file and map comment_text as the document/source column. The author and timestamp columns become metadata attributes for filtering and coding. For thread-aware analysis, pre-process the CSV to concatenate replies under each parent before importing.
Can I export comments from a live stream or premiere? Once the live stream ends and the chat is archived as comments, yes. Active live chat during a stream uses a different YouTube system that the comments extractor doesn't read. Wait until the broadcast finishes and the video is published as a regular VOD — at that point comments work normally.