YouTube comments can reveal video ideas through repeated questions, follow-up requests, confusion, objections and comparisons. They are a useful audience signal, but they come from a sample of viewers, not your whole audience, so they show possible demand rather than guaranteed demand. Validate the promising topics before you spend time producing them.
The rest of this guide shows how to do it by hand, how to avoid being misled by a few loud commenters, and where our free AI YouTube Video Idea Generator can save time.
Why YouTube comments contain useful content ideas
Keyword tools show what people type into a search bar. Comments show what someone said right after watching, in their own words. A viewer who searches for "budget camera for youtube" is looking for a product. The same viewer commenting "wish you had shown this on a cheaper setup" is telling you which video is missing.
The downside is noise. Most comments are compliments, jokes, spam or arguments. The skill is in picking out the few that describe a real gap, and in being honest about how much one comment can tell you.
Seven comment patterns worth looking for
1. Direct video requests
"Can you do a video on..." or "please cover..." are the clearest signals. They name the topic for you, but one request is only a starting point.
2. Repeated questions
When different people ask the same thing in different words, the video probably left a gap. Look for the same underlying question, not the same wording.
3. Confusion about something shown in the video
Comments like "what did you click at 4:10?" point to a step you moved through too fast. That can become a short explainer or a slower walkthrough.
4. Requests for comparisons
"Is this better than X?" means viewers are choosing between options and want help deciding. Check them against YouTube search suggestions.
5. Objections or complaints
"This would never work for a small team" is not an attack. It describes a viewer whose situation you did not cover.
6. Requests for an advanced or beginner version
"Great, but I need the basics first" and "now show the advanced version" both suggest a different level of the same topic.
7. Follow-up questions the video did not answer
Viewers often ask what happens next. "How do I keep this up after the first month?" is a follow-up video that your existing viewers have already asked for.
A manual step-by-step comment research workflow
You can do this with a spreadsheet and about an hour. Here is a simple version.
- Choose a relevant video. Pick one that matches the topics you want to grow, not just your most popular upload.
- Review comments and replies. Replies often hold the best detail, because people explain what they actually need.
- Search for question and request phrases. Try "how do", "can you", "what about", "next video", "versus" and "why". On a computer, YouTube Studio's comments page has a search box and filters for things like questions and response status, so you can narrow your own channel's comments without scrolling. YouTube's help pages note that this search tool is only available on a computer, so check the current YouTube Help page on reviewing comments for what it covers today.
- Record candidate topics. Write each idea in one line, with the comment pattern it came from and a link to the comment.
- Group comments with the same underlying intent. "How do I start?" and "I'm a total beginner, where do I begin?" are one need.
- Separate repeated demand from duplicates by one person. One viewer commenting three times is one data point.
- Validate the strongest ideas. Use the checks in the section below before you commit.
How to group similar requests without overcounting
Grouping is where most comment research goes wrong. A few habits help.
- Count people, not comments. Note the commenter name once per idea.
- Group by need, not by keyword. Two people can ask for the same thing in different words.
- Keep a "single mention" list. Ideas mentioned once are not useless, but they are not a pattern yet.
- Do not treat a like count as a vote. It shows interest in the comment, not that people want a whole video.
Here is a reusable table. The examples are hypothetical and are only there to show how to fill it in. They are not real viewer comments or research.
| Comment pattern | Underlying viewer need | Possible video angle | Validation needed |
|---|---|---|---|
| Direct request (hypothetical: "Can you cover invoicing for freelancers?") | Wants a topic explained from scratch | Beginner guide to the topic | Check search suggestions and whether other viewers ask the same thing |
| Repeated question (hypothetical: several people asking how to set up a tool) | Setup steps were unclear | Slow, step-by-step setup video | Count distinct people and look for existing videos that already answer it |
| Confusion (hypothetical: "What did you click there?") | A step moved too fast | Short explainer on that one step | Check whether the confusion is common or a one-off |
| Comparison request (hypothetical: "Is A better than B?") | Choosing between two options | Side-by-side comparison | Check search demand and competitor comparisons |
| Objection (hypothetical: "This will not work for small teams") | Their situation was not covered | Version for the viewer who objected | Check whether the objection is shared by others |
| Level request (hypothetical: "Now show the advanced version") | Wants more depth | Advanced follow-up or series | Check watch time on the original video |
| Unanswered follow-up (hypothetical: "What do I do after week one?") | Wants the next step | Follow-up video | Check whether you have the expertise and material to answer it well |
How to validate an idea before making the video
Comment research should support the rest of your research, not replace it. Before you commit to a topic, check it against:
- YouTube search suggestions. Type the topic in and see whether it autocompletes the way you expect.
- Keyword research. Comments tell you what viewers said. Keyword research tells you what many people search for. Each answers a different question, and neither wins in every case.
- Competitor performance. See whether videos on the topic already get views and whether the existing ones leave a gap.
- Your channel analytics. Look at which of your videos hold attention on similar topics.
- Your own expertise and production capability. A good idea you cannot deliver well is still a weak video.
Also ask whether the topic is big enough for a full video. A request can be popular in the comments and still be too narrow to carry ten minutes.
Using comments from competitor videos responsibly
Your own comments only reflect people who already watch you. Public comments on a competitor's video in your niche can show what a wider group is asking for, including gaps that creator has not filled. Used carefully, that is useful research.
A few ground rules. Use comments to understand patterns, not to copy a creator's video. Do not quote or call out individual commenters. And remember that another channel's audience is not yours, so check that an idea fits.
How the MBI YouTube Video Idea Generator works
If you want a faster first pass, our free AI YouTube Video Idea Generator does a limited version of this workflow. Here is exactly what it does, so you can judge how far to trust it.
- You paste a public YouTube video URL. No account is needed.
- It fetches up to 100 comments from YouTube, in YouTube's relevance order. YouTube's API offers two orderings, time and relevance, and returns at most 100 comments per request.
- It sends the first 80 of those comments to an AI model.
- It returns five suggested video ideas, each with a reason.
- Each idea gets an AI-estimated demand label of high, medium or low. There is no numeric demand score.
- It also lists trending topics, viewer questions, content gaps and a sentiment overview.
It does not read every comment on a video, and it does not rank ideas using a verified count of how often each one was mentioned. The labels are estimates, and the results do not guarantee views. We do not store the comments or the generated results. We store the submitted video URL and a timestamp, and temporary usage limits apply.
Use it as a starting list, then check each idea against the validation steps above.
Limitations and sampling bias
- Commenters are not your whole audience. Most viewers never comment.
- Relevance ordering is not a complete dataset. It surfaces the comments YouTube ranks highly, which may differ from the full set.
- One loud commenter is not broad demand. Look for several different people with the same need.
- Sarcasm and context can be misread, by people and by AI.
- A topic can be popular in the comments and too narrow for a video.
- Sentiment analysis is directional, not definitive. Treat a "70% positive" reading as a rough mood, not a measurement.
A repeatable comment-to-video checklist
- Pick a video and gather its comments and replies.
- Search for question and request phrases, and note each candidate idea with its source pattern.
- Group by underlying need and count distinct people.
- Check search suggestions, keyword research, competitor videos and your analytics.
- Decide whether you can deliver it well.
- Make the strongest one or two ideas, then read the new comments and repeat.
Frequently Asked Questions
Can YouTube comments give you good video ideas?
Yes, they often can. Repeated questions, follow-up requests and points of confusion all suggest topics. They are a sample of your audience, so treat them as a lead to validate, not proof of demand.
How many comments should you analyze?
There is no fixed number. A few dozen well-read comments on one video can show a pattern, and more helps you tell a pattern from a one-off. Looking across several videos is usually more reliable than going deep on one. The MBI tool fetches up to 100 and analyzes the first 80.
Can you analyze comments on a competitor's video?
Yes, public comments can be read and analyzed. Use them to spot gaps and patterns, not to copy the video.
How do you tell whether a request represents real demand?
Look for several different people asking for the same thing, then check it against search suggestions, keyword research and your analytics. One request is a hint. Several requests plus outside evidence is much stronger.
Can AI accurately understand every YouTube comment?
No. AI can miss sarcasm, slang, inside jokes and missing context. It is useful for finding themes quickly, but you should read the comments behind any idea you plan to act on.
Does the MBI tool analyze every comment?
No. It fetches up to 100 comments in YouTube's relevance order and analyzes the first 80.
Do comment-based ideas guarantee more views?
No. They can make your topic choices better informed, but views also depend on the title, thumbnail, topic demand, competition and how well the video is made.
Conclusion
Comments will not hand you a perfect content plan, but they are a cheap, specific source of ideas, and the manual workflow above works without any tool. The main discipline is to count people, group by need and validate before you produce anything.
If you want a quicker starting list, try the free AI YouTube Video Idea Generator on a public video, then check each idea with the steps in this guide. If you want help connecting your YouTube content to your wider marketing, see our digital marketing and social media services, browse our portfolio, or get in touch.