How to Use YouTube Comments With AI for Content Ideas
How to Find Content Ideas in YouTube Comments Using AI
YouTube comments are one of the most honest content research sources available. People aren’t performing there the way they do on Twitter or LinkedIn. They’re asking real questions, expressing real confusion, and telling you exactly what the video didn’t cover.
The problem is volume. A channel with any traction has thousands of comments and no obvious way to turn them into a content plan. You end up skimming, pattern-matching by gut feel, and making the same type of content you’ve always made.
These seven prompts change that. Copy a batch of comments, paste them in, and you’ll have a prioritised content angle list inside ten minutes.
Prompt 1: The raw comment mine
What it does: Scans a raw dump of YouTube comments and extracts every distinct content signal: questions asked, frustrations expressed, topics mentioned, and things people say they wish the video had covered.
When to use it: Before anything else. This is the starting point for the whole sequence.
You are a content strategist analysing YouTube comments to find content opportunities. Below is a batch of comments from a YouTube video about [VIDEO TOPIC]. Read every comment carefully. Extract and group the following: (1) questions viewers asked that weren’t answered in the video, (2) frustrations or problems mentioned, (3) topics or subtopics viewers brought up themselves, (4) things people said they wanted to see next or wished had been included. Present each category as a plain list. Do not summarise or editorially comment on the patterns yet. Just extract. [PASTE COMMENTS HERE]
How to use it:
Go to any YouTube video relevant to your niche. Doesn’t have to be your own channel.
Copy 50 to 100 comments. Sort by “Top comments” for the highest-signal batch.
Paste them in place of [PASTE COMMENTS HERE].
Fill [VIDEO TOPIC] with a plain description, e.g. “AI writing tools for content creators.”
Example input: Video topic: AI writing tools for content creators. Comments: [100 comments pasted from a popular video on the topic]
What you’ll get: Four clean lists that turn an unstructured comment section into readable signal. This is your raw material for every prompt that follows.
Advanced note: 100 comments is a good working size. Under 30 and the patterns are too thin. Over 200 and you start getting repetition that dilutes the output. If the video has thousands of comments, run the prompt twice with different batches and compare.
That prompt turns a wall of unstructured comments into four clean lists you can actually work from.
The next six prompts take those signals through to a finished content plan.
Prompt 2 — The pattern spotter: Groups the extracted signals by theme and ranks them by frequency, so you can see which questions and frustrations kept coming up rather than guessing at it.
Prompt 3 — The content angle generator: Turns the top recurring themes into specific publishable angles, each with a working title written the way a reader would search for it, not the way a content marketer would pitch it.
Prompt 4 — The gap finder: Compares your chosen angle against what already exists on YouTube for that topic, and identifies the specific thing that’s missing before you commit to making it.
Prompt 5 — The hook builder: Uses the actual language, phrases, and complaints from the comments to write opening lines that feel like you’ve had the exact same experience as your reader.
Prompt 6 — The title test: Generates eight title options drawn from real comment language, picks the three strongest, and explains in one sentence why each works.
Prompt 7 — The multi-video synthesis: Runs the analysis across multiple videos and surfaces only the themes that appear consistently across different comment sections, which is where your safest content bets are.
Plus: the comment collection shortcut for pulling batches without copying manually.