The short answer
Most advice about comments is about replying quickly to boost engagement. That is fine, but it treats the comment section as a metric to farm rather than as the most direct research channel you will ever have.
The three comments worth acting on
1. The question you already answered
Someone asks something the video covered. The instinct is mild irritation. The correct reading is that your explanation failed — and if one person typed it, many more thought it and scrolled on.
These are the single highest-value comments you get. They point at the exact sentence that did not land, and they are usually worth a whole follow-up video, because the question is demonstrably one real people have.
2. The "what about..." question
"This is great but what about X?" That is your audience handing you the next episode with demand pre-verified. If three people ask the same "what about", stop planning and make that.
3. The one using different words than you
The most underrated. When viewers describe your topic in words you would not have chosen, that is the vocabulary they would also type into a search bar. Search is now a major discovery surface on TikTok in particular, so their phrasing is worth more than yours.
Write those phrases down verbatim and use them in captions and on-screen text. We go into why in TikTok is a search engine now.
What to ignore
- Pure praise. Pleasant, useless. It tells you nothing you can act on.
- Pure abuse. Same, in reverse.
- Requests from people who will never watch. "Do a video about crypto" from an account with no interest in your subject is not demand, it is noise.
- One person's strong opinion about your format. Act on patterns of three or more, never on a single confident stranger.
A 20-minute weekly routine
- Open your last two weeks of posts across every platform, in one sitting.
- Copy out only questions into one list. Ignore everything else on this pass.
- Cluster them. Anything asked twice or more goes into a "make this" pile.
- Note the vocabulary. Any phrasing that differs from yours goes into a separate list for captions.
- Pick one for next week. Not five — one, actually made.
Twenty minutes a week produces a permanently full content calendar sourced entirely from verified demand. It is also the most reliable input for a recurring format, since the supply never runs out — see people follow formats.
Replying is still worth it
Separately from research: early replies matter. Engagement in the first hour helps distribution on most platforms, and replying is the cheapest way to generate it. Practically, that argues for scheduling posts at times you are actually awake rather than theoretically optimal times you will sleep through — an argument we make at more length in why best-time charts are lying to you.
Questions people actually ask
How do I learn what my audience wants?
Read your comments as data rather than as feedback. Collect questions only, cluster anything asked more than once, and make the most common one. That is demand you have verified rather than guessed.
Which comments actually matter?
Questions you thought you already answered (your explanation failed), "what about X" questions (your next video, pre-validated), and comments using different vocabulary than yours (the words people search with).
Should I act on every comment request?
No. Commenters are a small, unrepresentative slice of your viewers. Treat comments as a source of hypotheses, not as a vote, and act on patterns of three or more rather than one confident opinion.
Does replying to comments help reach?
Early engagement helps distribution on most platforms, and replying is a cheap way to generate it. It also makes people more likely to comment next time.