Export Filtered YouTube Comments and Selected Fields

1. Start with a saved extraction

Sign in, submit a public YouTube video or Shorts URL, review the extraction limit, and start the job. You need at least 500 available credits to start any extraction. Each successfully saved comment or reply costs 1 credit; unused reserved credits return when the job settles. The 500-credit minimum is not a flat extraction fee.

Open the results after extraction completes. Record the video URL, extraction date, saved row count and limit if you plan to use the dataset for research. “All comments” in an export means all saved results for that extraction, subject to the chosen extraction limit and source availability.

2. Choose filtered rows or all saved comments

Apply the results filters that suit your question, then check the matching count. For example, use a text search to find comments mentioning a product or filter to replies for a conversation study. Inspect several matching rows before exporting so that an overly narrow filter does not hide useful responses.

Open the export dialog. When filters are active, choose “Filtered comments” to download only matching rows. Choose “All comments” to export the complete saved extraction instead, even while filters remain active on screen. Pagination in the results table does not restrict an export to the current page.

For a reproducible workflow, keep one full export as your source snapshot and a second filtered export as your working dataset. Write down the filter settings outside the file rather than assuming another researcher can reconstruct them from the rows alone.

3. Select columns for Excel, CSV or JSON

Under “Export fields”, keep “All fields (including comment IDs)” for the full schema or choose “Selected fields” and check the fields you need. “Use visible table fields” copies the current table column selection into the export selection. At least one field must remain checked.

For a compact research spreadsheet, select Comment ID, Parent Comment ID, Comment, Published and Type. Add Likes when comparing engagement, or Author Channel ID when you need a stable author identifier. Keep the two comment IDs whenever you may need to connect replies to threads later.

Choose Excel (XLSX), CSV or JSON and generate the download. Selected fields apply to all three formats. JSON keeps its export metadata while limiting the fields inside comment records; selecting fewer comment fields does not remove the metadata wrapper.

4. Check the downloaded dataset

Compare the exported row count with the matching count for a filtered export, or with the saved total for an all-comments export. Count data rows rather than including the header. Confirm the expected columns are present and spot-check a few IDs and comment texts against the results.

Open XLSX directly in Excel. Import CSV as UTF-8 and preserve identifiers as text; an ID is not a quantity to round or calculate. For JSON, count records in the comments array rather than counting lines in the file. A reply’s parent can be absent from a filtered subset without indicating a broken export.

Generating another format, changing the fields, exporting all rows, and downloading again cost 0 credits. You do not need to start a new extraction just to change an export. A new extraction collects a new snapshot and uses credits again.