What is YouTube Comments?
Scavio's YouTube comments endpoint returns the top comments on a video sorted the same way YouTube sorts them by default, which is a mix of recency and engagement. Each comment includes the author handle, channel link, avatar, the full text, like count, reply count, timestamp, and a flag for whether the comment is pinned or authored by the channel owner. You can request more pages of comments using a cursor, and you can optionally expand replies. Comments are an underused signal for sentiment, product feedback, and content gaps, especially for how-to and review videos where creators often correct themselves or clarify in the comments rather than by editing the video.
Example Response
{
"video_id": "dQw4w9WgXcQ",
"total_comments": 482,
"comments": [
{
"author": "@mlrockstar",
"author_channel": "https://youtube.com/@mlrockstar",
"text": "The part at 3:45 finally made checkpoints click for me. Huge.",
"likes": 214,
"replies": 3,
"published": "2 days ago",
"published_iso": "2026-04-14T09:12:00Z",
"pinned": false,
"channel_owner": false
},
{
"author": "@LangChain",
"author_channel": "https://youtube.com/@LangChain",
"text": "Pinned: source code link is in the description. Reply here with questions!",
"likes": 842,
"replies": 42,
"published": "3 weeks ago",
"published_iso": "2026-03-26T12:00:00Z",
"pinned": true,
"channel_owner": true
}
],
"next_cursor": "Ei0SC2RRdzR3OVdnWGNRGAYyFSIRIgtkUXc0dzlXZ1hjUTAAeAJCEGNvbW1lbnRzLXNlY3Rpb24%3D"
}Use Cases
- Product feedback mining on review and unboxing videos
- Sentiment analysis on creator and brand content
- Surfacing pinned creator corrections and clarifications
- Spam and moderation analysis for MCN agencies
- Identifying super-fans and high-engagement viewers
Why YouTube Comments Matters
Comments are the most candid signal a creator has about how content actually landed, and they often contain product requests, corrections, and competitor mentions that do not appear in transcripts. Scavio returns them with stable authorship metadata so sentiment and clustering pipelines can track contributors over time. For brands monitoring review videos, this is the only reliable way to see purchase intent and objections surfacing in real time.
LangChain Example
Drop youtube comments data into your LangChain agent in a few lines:
from langchain_scavio import ScavioYouTubeCommentsTool
from langchain_anthropic import ChatAnthropic
tool = ScavioYouTubeCommentsTool(api_key="your_scavio_api_key")
llm = ChatAnthropic(model="claude-opus-4-6")
comments = tool.invoke({"url": "https://youtube.com/watch?v=dQw4w9WgXcQ"})
joined = "\n".join(c["text"] for c in comments["comments"][:50])
report = llm.invoke(f"Summarize top complaints and praise:\n\n{joined}")
print(report.content)