TikTok Comments Scraper: the thread under the video

One URL field, one switch behind it

There is only one way in: the URL of a post. Everything else about a run is decided by whether replies are being collected and where the ceiling sits.

• startUrls

  $0.0003per comment

  No per-post fee

  A video or photo post link. Profile and hashtag URLs are not accepted here, those are other actors.

  https://www.tiktok.com/@user/video/7050551461734042926

• Multiplies the bill without changing the price

  includeReplies

  same ratemore rows

  Rows, not a new fee

  Off by default. On, the run walks the reply threads under each top-level comment as well.

  includeReplies: true

• maxItems

  ÷ $0.0003budget to cap

  Counts replies too

  The ceiling across the whole run, shared between every post in it. Comments and replies both count against it.

  5000

• customMapFunction

  freeno charge

  Transform, never filter

  Reshapes each row on the way out. For transformation only: accounts that use it to filter get blocked.

  (o) => ({ user: o.user.username, text: o.text })

Nothing is billed per post, so a run over one video and a run over fifty cost the same for the same number of comments. The variable that moves the total is the switch.

Top level only, the full thread, or a qualifying pass

Three runs worth setting up differently. The first two differ by one boolean; the third changes what you do with the output rather than how you ask for it.

{
  "startUrls": [
    "https://www.tiktok.com/@user/video/7050551461734042926"
  ],
  "includeReplies": false,
  "maxItems": 1000
}

{
  "id": "7277992603752203013",
  "text": "TE AMO MI NIÑA💚💚💚💚💚💚🙌🌷",
  "createdAt": "2023-09-12T17:28:42.000Z",
  "likeCount": 0,
  "replyCount": 0,
  "commentLanguage": "es",
  "isAuthorLiked": false,
  "awemeId": "7050551461734042926"
}

{
  "startUrls": [
    "https://www.tiktok.com/@user/video/7050551461734042926"
  ],
  "includeReplies": true,
  "maxItems": 4000
}

{
  "parentId": "42517992603752202618",
  "id": "7277992603752203013",
  "replyCount": 0,
  "isAuthorLiked": false
}

{
  "startUrls": [
    "https://www.tiktok.com/@user/video/7050551461734042926"
  ],
  "maxItems": 5000
}

{
  "user": {
    "username": "21billie_eilish",
    "displayName": "¡★!",
    "url": "https://www.tiktok.com/@21billie_eilish",
    "bio": "Billiestan for ever¡★!",
    "region": "CO",
    "language": "es",
    "hasEmail": false,
    "hasPhone": false
  }
}

A comment, and the person who left it

Two blocks per row. The comment side is compact; the commenter side is not, and it is what separates this output from a generic comment dump.

• 5

  The comment

  Always present

  • id
  • text
  • createdAt, ISO 8601
  • commentLanguage
  • awemeId, the post it sits under

• 3

  Engagement

  Small on purpose

  • likeCount
  • replyCount
  • isAuthorLiked, the creator hearted it

• 1

  Thread position

  parentId

  • Absent or empty on a top-level comment
  • Names the comment a reply answers

• 11

  The commenter

  user, on every row

  • username, displayName, id
  • url
  • bio
  • region, language
  • hasEmail, hasPhone
  • avatarUrl, coverImage

Two language fields sit side by side and are not the same thing: commentLanguage is what this comment was written in, user.language is what the account is set to. A mismatch is often the most interesting row in a sentiment pull.

What a run needs to go well

Four conditions for a clean run, plus one known gap in how replies come back.

• 01

  Each URL should return at least ten comments

  It is a usage rule rather than advice: very low-engagement posts create a negative margin and may be blocked. Pick targets by comment count, not by views.

• 02

  Post URLs only

  startUrls takes video and photo post links. A profile URL or a hashtag page belongs to a different actor in the family and will not produce comments here.

• 03

  Replies are not guaranteed

  Replies sometimes fail to come through when includeReplies is on. Verify against your own use case and measure before you budget.

• 04

  The ceiling is shared and counts replies

  maxItems limits the run as a whole across every URL, and a reply consumes one of those rows exactly as a comment does. Turning the switch on halves your effective comment depth unless you raise it.

• 05

  Large batches can outrun the log

  If output stops partway, look for a "logging is too fast, some lines were skipped" message. The fix is smaller batches, not a retry.

Free accounts are in demo mode: 5 runs a month, 10 items each, no charge.

What the switch costs you

The rate never moves: every row is $0.0003, whether it is a comment or a reply. So the only question that matters for a budget is how many rows a thread produces, and that is the number worth measuring on a small run first.

The last row is the point: a hundred thousand rows costs $30 regardless of how they split between comments and replies. The switch changes how fast you get there, not the price of arriving.

Who reads comments for a living?

Brand monitoring, sentiment work, lead generation and competitor research are the usual jobs. The commenter block is what decides which of those this actor is actually best at, a bio and a region on every row is a qualifying tool, not a mood ring.

• Lead generation

  Top level, then filter on the user block

  Pull the commenters on a niche video and keep the ones whose bio and contact flags say they are running a business

  $1.50 per 5,000 comments

• Brand monitoring

  Top level, across a campaign’s posts

  Read the reaction to a launch with isAuthorLiked showing which replies the brand itself endorsed

  $0.30 per 1,000

• Sentiment analysis

  Full threads

  Build labelled corpora where commentLanguage lets you split the set before it reaches a model

  $1.20 per 1,000 with 3 replies each

• Competitor research

  Top level, on a rival’s posts

  Find the complaints an audience will make in public but never in a survey

  $0.03 per 100-comment probe

• Community teams

  Full threads, scheduled

  Catch unanswered questions by looking for top-level comments with a replyCount of zero

  $9.00 per 30,000 rows

Wiring it up

The actor is apidojo/tiktok-comments-scraper. The one thing worth writing carefully is the reply check: threads may come back short, so measure the ratio on a small run and size the ceiling from what you actually got.

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
ACTOR = "apidojo/tiktok-comments-scraper"

def comments(urls, with_replies=False, cap=1000):
    run = client.actor(ACTOR).call(run_input={
        "startUrls": urls,
        "includeReplies": with_replies,
        "maxItems": cap,
    })
    return list(client.dataset(run.default_dataset_id).iterate_items())

rows = comments(["https://www.tiktok.com/@user/video/7050551461734042926"],
                with_replies=True, cap=2000)

# Replies may not always arrive. Check, do not assume.
replies = [r for r in rows if r.get("parentId")]
print(f"{len(replies)} replies in {len(rows)} rows")

# Contact flags are booleans: use them to narrow, not to contact.
leads = [r for r in rows if r["user"]["hasEmail"] and r["user"]["bio"]]
for r in leads:
    print(r["user"]["username"], r["user"]["region"], r["user"]["bio"][:40])

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });

const run = await client.actor('apidojo/tiktok-comments-scraper').call({
  startUrls: ['https://www.tiktok.com/@user/video/7050551461734042926'],
  includeReplies: true,
  maxItems: 4000,
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();

// parentId is the whole threading model: absent means top level.
const threads = Object.groupBy(items.filter((c) => c.parentId), (c) => c.parentId);
const unanswered = items.filter((c) => !c.parentId && c.replyCount === 0);

console.log(Object.keys(threads).length, 'threads,', unanswered.length, 'unanswered');

curl -X POST \
  "https://api.apify.com/v2/acts/apidojo~tiktok-comments-scraper/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"startUrls":["https://www.tiktok.com/@user/video/7050551461734042926"],"includeReplies":false,"maxItems":1000}'

JSON, CSV or Excel from the Storage tab, or read straight off the API. The Console preview shows a subset of fields, the full commenter block is in the download.

TikTok Comments Scraper FAQ

Where TikTok Comments Scraper fits