How to Export YouTube Comments for Sentiment Research
Reading a comment section tells you an impression. Structured data tells you a distribution. Here's how to get the second one.
Updated 26 September 2026 · 7 min readIf you're studying audience reaction to a video — a product launch, a political clip, a viral moment — the comment section is often the richest signal on the page. But a sentiment-analysis workflow needs rows, not a scrolling feed. It needs an author field, a text field, a like count and a timestamp it can sort, filter and feed into whatever scores positive, negative or neutral.
This guide covers why manually reading or copying comments breaks down past a handful of them, what fields a proper export gives you, which file format suits which next step, and the honest limits of a tool that exports data rather than analyzes it.
Why reading comments doesn't scale
YouTube's comment panel is built for scrolling and replying, not for research. There's no way to select a comment section as a table, no way to sort by likes across the whole thread, and nested replies collapse until you click into each one individually. Copying comments by hand into a spreadsheet works for the first twenty; past a few hundred it stops being a method and starts being a part-time job.
Sentiment analysis needs consistent fields, not a feed
Getting comments out as structured data
Paste the video URL
Let it pull the thread
Choose a format for your next step
Download and load it into your workflow
Which format fits your next step
The right export format depends entirely on what scores the sentiment afterward.
| Format | Best for | Why |
|---|---|---|
| CSV / Excel | Spreadsheet sentiment scoring, no-code tools | Opens directly as rows and columns — sort, filter, or run a formula against the text column. |
| JSON | Python/NLP scripts, LLM pipelines | Nested reply structure survives intact, and it parses straight into a list of dicts or objects. |
| Markdown / HTML | Sharing a thread in a report or write-up | Readable formatting for a human reviewer, not built for programmatic scoring. |
| XML | Feeding into an existing XML-based pipeline | Structured and tagged, useful if your toolchain already expects XML input. |
What researchers actually do with this
- Manual tagging passes
Export to CSV, add a sentiment column, and go row by row for a small, high-stakes sample where automated scoring alone isn't trusted.
- Spreadsheet formula scoring
Run a keyword-based or lexicon-based formula against the comment text column to get a first-pass positive/negative/neutral split.
- Feeding an LLM or NLP script
Load the JSON export into a script or paste batches of comment text into a prompt for classification at scale.
- Engagement-weighted analysis
Use the like count alongside sentiment to see which opinions the audience actually amplified, not just which ones were posted.
What the export doesn't do
- One video at a time. There's no channel-wide or multi-video batch mode — research spanning several videos means one export per video.
- No built-in sentiment scoring. This is a data export tool, not an analysis tool — positive/negative/neutral tagging happens after the export, in whatever you send the data to next.
- Free tier returns a portion of the thread. A free account can try the export at 1 credit per run, but only a paid plan pulls the full comment set on a heavily-commented video.
How MyFreshTools Helps
What the tool does
The Comment Exporter turns one video's comment section into a structured file — CSV, Excel, JSON, Markdown, HTML, XML or a ZIP of several formats — with author, comment text, like count, timestamp and nested replies preserved. Connect YouTube with read-only access to also read your own private or unlisted video's comments.
Who it's for
- Researchers and students studying audience reaction to a specific video.
- Analysts building a sentiment-scoring pipeline that needs clean input rows.
- Anyone doing a manual tagging pass who needs the comment text in a spreadsheet first.
How to use it
- Paste the video's watch link or Shorts link and run the export.
- Pick the format your next tool expects — CSV/Excel for spreadsheets, JSON for code.
- Feed the resulting rows into your own scoring method — formula, script or prompt.
Requirements and limits
- A free MyFreshTools account. No card, free credits on signup.
- 1 credit per export; the free tier returns a portion of the comment thread, not the full set.
- One video at a time — no channel or multi-video batch export.
- No sentiment scoring is built in — this exports the data, it doesn't classify it.
- Connect YouTube (read-only) is available for your own private/unlisted videos.
Frequently asked questions
No. The Comment Exporter gives you the raw structured data — author, comment text, likes and timestamp — as clean rows. Scoring positive, negative or neutral sentiment happens in whatever you feed that data into next: a spreadsheet formula, a Python script, or an LLM prompt.
No, the tool works on one video at a time. If your research spans several videos, run the export once per video and combine the resulting files before analysis.
CSV or Excel if you're scoring sentiment with spreadsheet formulas or a no-code tool; JSON if you're feeding the data into a Python script or an LLM prompt. Markdown, HTML and XML suit sharing or archiving a comment thread rather than processing it programmatically.
Nested replies are included and structured under their parent comment, so a reply thread stays attached to the comment it responds to rather than getting flattened into an unordered list.
No — a free MyFreshTools account with free credits is enough to try it, at 1 credit per export. The free tier returns a portion of the comment set rather than the full thread; a paid plan is only necessary once you need every comment on a heavily-commented video.
The takeaway
Sentiment research on a YouTube comment section starts with getting the comments out as consistent rows — author, text, likes, timestamp — not with reading the thread yourself. Once that export exists, any sentiment method you already trust can run against it.