Export TikTok Comments to CSV in 2026

TikTok's official API is invite-shaped: you apply to TikTok for Developers, wait, and hope your use case fits their program. Comment access is not something you get on a Tuesday afternoon. If you are doing research, social listening, or just trying to read what an audience said, that wait is the whole obstacle.

The TikTok Comments Scraper skips it. Paste a video URL, get the comments.

What you get

{
  "text": "the cat is the main character",
  "author": { "name": "Jane", "id": "janedoe" },
  "likeCount": 3120,
  "replyCount": 18,
  "publishedAt": "2026-07-14T09:30:00.000Z",
  "publishedAtConfidence": "exact"
}

Two details other TikTok scrapers get wrong.

The author id is the @handle, on every row. TikTok's data has two ways to identify a commenter — a numeric user id and the handle — and the two upstream sources this actor uses disagree on which they return. Because the page can be served by either source, a numeric id would change meaning halfway through your export. The handle is used throughout, so author.id means one thing from the first row to the last.

Dates are honest about the year. One source returns TikTok comment dates as month-and-day — "7-14" — with no year. Parsed naively that resolves to the year 2001, and most scrapers will hand you that date labelled as exact. Here it comes back as publishedAtConfidence: "unknown" with no timestamp, because a confidently wrong date is worse than a missing one.

Run it without code

Open the TikTok Comments Scraper on Apify.
Paste a video URL. Short vm.tiktok.com links work too — they are resolved before the fetch.
Set Maximum results, click Start, and export to CSV.

Run it from code

curl -X POST \
  "https://api.apify.com/v2/acts/scriptbase~tiktok-comments-scraper/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "url": "https://www.tiktok.com/@scout2015/video/6718335390845095173", "maxItems": 500 }'

The limits, up front

Top-level comments only — replies are not expanded, though each comment carries its replyCount. Public videos only; a private account returns a clear "not found" rather than an empty file you have to debug. One video per run. None of these are surprises you discover after paying — they are the shape of the tool, stated before you start.