Comments

The Comments page at podhood.com/studio/<slug>/insights shows what your audience is saying across every episode — mirrored read-only from YouTube, analyzed comment by comment, and aggregated with the receipts one click away. PodHood never replies, moderates, or writes anything back: every comment deep-links to YouTube, where the conversation actually lives.

Availability

• YouTube-sourced Channels only. The page mirrors the YouTube comment section; an RSS-sourced Channel has no comment source and never shows it.
• Pro and Studio. The capability follows the Owner's Plan. On Free the page describes the capability, and none of the Channel's comment data is pulled or analyzed.
• No Credits. Comment mirroring and analysis never draw from the Owner's Credit pool — Credits stay dedicated to indexing. See Plans & credits.

How comments arrive

• On first contact with an episode, PodHood imports its most-liked comments (with their like counts as captured) plus the newest page — so an established back catalog shows its best comments on day one.
• From then on, a daily pass picks up each episode's new comments.
• Refresh now runs the same pull on demand — useful during a launch-day spike.
• Like counts are captured at import and not refreshed afterwards; recency, not likes, keeps the stream current.

What each comment gets

Every comment is analyzed in whatever language it was written:

The Intent classes:

• Question — asks you or other listeners something. The Needs reply chip is this slice.
• Suggestion — asks for something: topics, guests, format changes.
• Praise — compliments the episode, the show, or a guest.
• Complaint — criticizes the content, the production, or a guest.
• Experience — adds the commenter's own account, evidence, or counterpoint.
• Noise — empty chatter, emoji-only, spam. Excluded from every statistic, still visible (dimmed) in the stream.

Sentiment and Intent are independent: a suggestion can be enthusiastic, an experience can be a negative counterexample. On indexed episodes a timestamped comment also names the chapter it points into — "the moment at 18:42" becomes the moment in a named section of the episode.

Read the page

Two columns: the comment stream on the left, insight cards on the right. The page has one hard rule — every number opens its evidence. Click any card, chip, or bar and the stream filters to exactly the comments behind it, with matching totals.

• Sentiment — the positive/neutral/negative split, plus the trend across recent episodes.
• What they're saying — comment counts by Intent.
• Content ideas — recurring asks in the audience's own words ("more technical guests", "fix the quiet audio"), mined from suggestions and complaints once at least three comments voice the same thing. Ideas persist across refreshes, so their counts grow instead of churning.
• By guest — how the audience responds to each guest's episodes. Guests come from indexed episodes' Briefs, so this fills in as you index.
• By topic — reception and asks per topic, from the same indexed structure.

The stream itself filters by episode, sentiment, intent, and timestamped-only, and sorts by Newest or Popular (likes as captured). Two things stay out of every statistic but remain visible in the stream:

• Noise renders dimmed and never moves a number.
• Your own comments carry a Host badge — they are not audience voice.

Act on it

PodHood is read-only toward YouTube by design. Each comment carries Reply on YouTube — the deep link into that exact thread. Reply, pin, heart, or moderate there; the next pass mirrors whatever the audience does in response.