How to Find YouTube Video Ideas in the Comments

The hardest part of running a channel isn't shooting or editing. It's the blank line in the planning doc. You sit down to pick the next video, brainstorm for twenty minutes, and end up with a list of ideas you find interesting — which is a different thing from a list of things people are actually looking for.

There is a demand-ranked list of video ideas for your niche, already written, by exactly the audience you want. It's sitting in the comments under the videos that currently rank for the searches you care about. This post covers where to look, the five comment patterns that turn into videos, how to read a comment section without losing an afternoon, and the comments that look like ideas but aren't.

Mine other people's comments, not your own

Your own comment section is the obvious place to start and the worst one. It's a small sample, it's made of people who already subscribed, and it tells you what your existing audience wants next — not what the much larger group of people searching your topic couldn't find. Videos built from your own comments grow depth with the viewers you have. They rarely bring new ones.

The comments under a competitor's ranking video are a different animal. Those people arrived from search, watched a video that was good enough to rank, and still had something left over — something they typed out afterwards. Every one of those is a viewer telling you, in their own words, what the best available answer failed to give them.

That's also why comments beat keyword tools for this particular job. A keyword tool tells you a phrase gets typed into a box. A comment tells you a specific person watched the current best answer to that phrase and walked away unsatisfied. One is volume. The other is volume with a reason attached.

Where to mine

Pick a search you want to rank for — a real one, phrased the way a viewer would type it. Search it on YouTube, take the top five results, and read the comments on all five. Not the whole channel, not a video you liked: the specific set of videos standing between you and that query.

Five is the number because the point is repetition. One person asking a question under one video is an anecdote. The same question appearing under three of the five ranking videos is a pattern — proof that the gap is in the topic's coverage rather than in one creator's script. Frequency across videos is the entire trick.

The five patterns worth mining

1. The question nobody answered

The strongest signal there is. Somebody asks something the video didn't cover, and either nobody replied or the creator replied in text because it wasn't in the video. When the same question turns up under several ranking videos, you've found a content gapwith a timestamp on it — demand that exists, supply that doesn't.

Watch for the phrasing, too. People write comments the way they'd phrase a search, which makes these questions the closest thing to free keyword research you'll get.

2. The timestamp complaint

“The actual answer starts at 8:40.” “Skip to 12:15 if you just want the settings.” These get liked into the top of the comment section constantly, and creators read them as a pacing note. They're better than that: they tell you exactly which 90 seconds of a 15-minute video the search traffic actually came for.

The video that idea becomes is the direct one — the same answer, delivered in the first thirty seconds, without the preamble the incumbent buried it under. Whole channels have been built on being the version of a video nobody has to scrub through.

3. The “what about…” qualifier

“What about on Windows?” “Does this work if you're on the free plan?” “What about for a left-hander?” The video answered the general question and the commenter has a condition it didn't cover: a platform, a budget, a skill level, a country, a piece of gear.

Each qualifier is a narrower search with less competition and a viewer who cares more. “How to dial in espresso” is a war. “How to dial in espresso on a cheap grinder” is a video with a specific, findable, underserved audience — and the comments are how you learn which qualifiers keep coming up.

4. The correction

“This menu doesn't look like this anymore.” “They changed this in the last update.” “Is this still accurate in 2026?” When a comment section has quietly turned into a support thread for an outdated video, the topic is proven — the view count says so — and the incumbent can't defend it without reshooting.

Corrections are the fastest ideas to act on and the ones with the shortest shelf life. In software, gear and anything with an interface, they reopen on a cycle.

5. The direct request

“Please make a video on X.” The most obvious pattern and the one to weight lowest, because a single request is one person's wishlist. It only becomes a signal when it repeats across videos, or when it lines up with something you saw in patterns 1–3. Treat it as confirmation, not as a brief.

Reading a comment section without losing an afternoon

A ranking video can have thousands of comments and you are not going to read them all. Two passes get you most of the signal.

Pass one: Top comments.That's the default, and YouTube's Sort by control lets you switch it. Engagement-ranked comments are where the timestamp complaints and the heavily-upvoted questions live — if fifty people liked “skip to 8:40,” fifty people had the same experience. The first screen or two is usually enough.

Pass two: Newest first.This is the pass almost everyone skips, and it's where corrections surface. An outdated video's top comments are two years old and full of praise; its newest comments are people saying the menu moved. Same distinction the YouTube Data API draws, incidentally — its commentThreads.list endpoint orders by time or relevance, and the two views answer different questions.

Keep a tally as you go, one line per recurring question, with a mark each time it reappears under a different video. The tally is the deliverable — not a list of interesting comments, but a list of questions ranked by how many of the ranking videos failed to answer them.

Doing that properly across five videos is an hour or two of scrolling, which is why most people do it once and never again. It's the job we built VidHalo to automate: give it a search term and it pulls the comments from the videos ranking for it, clusters the recurring questions, and ranks them by how often they came up and whether the transcripts of those videos ever answered them. The tally you would have kept by hand, in a couple of minutes.

Turning a comment into a video idea

A recurring question is a candidate, not a plan. Three checks before it goes on the calendar:

1. Check somebody searches it.Type the question into YouTube's search box and watch the autocomplete. If the phrase (or something close) comes back as a suggestion, real people type it. If nothing comes back, you may have found a question people ask after watching a video rather than one they search for — which is a good section of a video, not a video.
2. Check the gap is real.Comments tell you what viewers felt was missing; they can be wrong. Skim the transcripts or chapters of the ranking videos for the answer before you commit. Sometimes it's in there and simply buried — which is pattern 2, a different video.
3. Keep the viewer's words. Commenters phrase things the way searchers do. When their phrasing survives into your title, your title matches the query — and YouTube ranks search results on how well the title, description and video content match what was searched for.

Then build the video to cover the table stakes and the gap — a video that only answers the missing question, and skips everything the ranking videos all cover, tends not to rank at all. The full process is in how to outrank a YouTube video.

What to ignore

• Praise.“Great video, thanks!” tells you the video worked. It doesn't tell you what to make next.
• Production notes. Complaints about audio, music or pacing are feedback for that creator, not demand for a topic.
• The one loud request.A single “you should cover X” with no repetition anywhere else is one person. Wait for the second and third.
• Spam and engagement bait.Bot replies, self-promo and “who's watching in 2026” carry no information at all.

And one structural caveat worth holding onto: people who comment are a small, self-selected slice of the people who watched. They skew towards the frustrated, the expert and the very new. That makes comments a brilliant source of candidate questions and a poor source of proportions — which is exactly why the second check above sends you to the search box. Comments find the idea; search demand confirms it.

The short version

Stop brainstorming and go read the comments under the five videos ranking for the search you want. Look for five things: questions nobody answered, timestamp complaints, “what about…” qualifiers, corrections on stale videos, and repeated direct requests. Read Top comments for the loud signals and Newest first for the stale ones, and tally questions by how many videos they recur under. Then confirm the question against autocomplete before you shoot, because commenters are a skewed sample of the audience you're actually trying to reach.

If you'd rather not do the scrolling, VidHalo mines the comments and transcripts of the ranking videos for any search term and hands back the unanswered questions ranked by demand — or start with the free tools, no login required.