The 10 Best Prompts to Find YouTube Video Ideas When Your Brain Gives Out

You open YouTube Studio.
You look at your editorial calendar.
Nothing.

The real problem is not that you lack talent. It is that you ask your brain to produce a clear, original, shootable, SEO-optimized idea, adapted to your audience, compatible with your production level, and strong enough to survive the algorithm. All of this while staring at a blank page.

The prompts to find YouTube video ideas do not serve to replace your creativity. They serve to remove the fog. Well used, they become a miniature writers' room, a demanding producer, an audience analyst and a sparring partner who pushes you to get out of soft concepts.

Understanding why a good YouTube idea is never just an idea

A usable YouTube idea must carry a tension. With no tension, you get an informative but flat video. "How to choose a camera" is an idea. "I tested a 400-euro camera to shoot an ad like a pro team" is a promise. The difference is enormous, because the second contains a conflict, a stake, a transformation and an immediate curiosity.

When I work with beginners, I often see the same mistake: they ask the AI "give me 20 YouTube video ideas about cinema". Predictable result: a lukewarm soup. The AI answers with titles that could come from a 2017 blog. Normal. The prompt gives no angle, no audience, no level, no production constraint, no emotion to provoke.

Here's the thing: generative AI is excellent when you give it a precise playground. It becomes bad when you ask it to be creative in the void. To find YouTube video ideas, you must feed it with your universe, your limits, your audience, your format, your tone, and above all the real frustrations of the people you want to help.

A good YouTube idea must answer four simple questions. Who suffers? From what exactly? What transformation does the video promise? Why watch now rather than later? If your idea does not answer that, it is probably going to die in the published-content tab with 137 views and three crypto-bot comments.

The method I use on AI Studio looks more like an audiovisual concept development than a school brainstorming. We start from a human problem. We turn it into an angle. We test it as a title. We check the visual promise. We imagine the thumbnail. Then only after that do we ask the AI to decline it.

To deepen this visual-promise logic, also read our guide on how to create more cinematic AI images. You will quickly understand that a good video often starts with a sharp mental image.

The 10 best prompts to find YouTube video ideas

These prompts are not magic formulas. They are creative-direction frames. Copy them, but above all adapt them. Replace the variables, add your tone, impose your constraints, and always ask for a bolder second pass.

The first prompt is my favorite to get out of the void: "Act as a YouTube producer specialized in [niche]. List 30 concrete frustrations that [precise audience] encounters when it tries to [goal]. For each frustration, propose a video idea with a conflictual angle, a clear promise, a provisional title and a visual opening scene." This prompt forces the AI to start from the pain, not the theme.

The second prompt serves to build a series: "From my channel that helps [audience] get [result], propose 10 recurring formats of YouTube videos. Each format must have a repeatable mechanic, a title example, a recommended duration, a production level, and a reason the audience would come back." Watch what happens when you do this: you no longer look for isolated videos. You build a grid.

The third prompt is SEO: "Find 25 YouTube video ideas around the keyword [main keyword]. Sort them by search intent: learn, compare, solve a problem, buy, get inspired. For each idea, give a natural title, 5 probable PAA questions, and a value promise in one sentence." There, you do not only fight for the recommendation algorithm. You also build an archive findable by search.

The fourth prompt turns a flat concept into narration: "Take this idea: [idea]. Turn it into 5 different narrative angles: field test, personal mistake, before/after transformation, investigation, timed challenge. For each angle, propose a 20-second opening, a central conflict and a satisfying ending." It is often there that beginners feel a click.

The fifth prompt is brutal but necessary: "Here are 10 video ideas. Critique them like a demanding editorial director. For each idea, say why it risks being ignored, which angle is missing, how to make it more specific, and propose a more clickable version with no falling into clickbait." The AI becomes useful when you ask it to contradict you.

💡 Frank's Cut: always ask the AI "which idea seems the most banal?" then "how to make it more risky but still useful?". The best videos are often born there, in the tension between pedagogy and editorial danger.

The trench workflow to go from 0 ideas to 30 shootable concepts

The first step consists of writing a context note. Not a novel. One page maximum. You put in it your audience, your production level, your gear, your available time, your tone, your forbidden subjects, and three videos you admire. With no it, you ask the AI to guess your studio. Bad idea.

Real example: a creator wants to launch a channel on AI video. He owns a MacBook, CapCut, Runway, a correct voice, and two hours per evening. His target audience: creative freelancers who want to sell local ads with AI. If his prompt does not contain these constraints, the AI is going to propose a "40-minute documentary on the history of AI in cinema". Beautiful on paper. Dead to produce.

The good starting prompt would be: "I am launching a YouTube channel for beginner creative freelancers who want to use generative AI to produce local advertising videos. My tone is direct, field mentor, anti-generic content. I can produce 8 to 12-minute videos with screen capture, voiceover, visual examples and 2 hours of editing. Propose 30 solo-shootable ideas." There, the AI breathes.

Then, I ask for a classification by potential. Not by feeling. I want four scores: beginner interest, SEO potential, thumbnail potential, production difficulty. An idea scored 9 in SEO but 2 in thumbnail can become an article, not necessarily a video. An idea scored 9 in thumbnail but 3 in educational value can get the click then disappoint. You have to arbitrate.

Here is a scoring prompt: "Rate these ideas out of 10 according to: audience pain, curiosity, SEO potential, thumbnail potential, ease of production, originality. Add a recommendation: produce now, keep for later, turn into a Short, or abandon." It is simple. It is cold. It is very useful.

To feed this system, also use the public data. Google Trends can spot a rise in interest. The official documentation of YouTube Creator Academy recalls the basics on audience, packaging and regularity. And if you want to understand how to structure your ideas into a series, reread our method on the best AI video tools for creators.

Scenario 1: beginner channel on AI cinema

Let's imagine you want to create a channel on AI cinema. Bad prompt: "Give me video ideas about AI cinema." Too broad. You are going to receive titles like "The future of cinema with artificial intelligence". No one clicks, except maybe your cousin out of pity.

Good prompt: "I want to help beginner videographers create cinematic scenes with Runway, Pika, Kling and Flux without falling into the plastic render. Give me 20 YouTube video ideas based on concrete mistakes: inconsistent characters, artificial light, impossible camera movements, too-vague prompts, edit with no rhythm. Each idea must include a before/after example."

There, the AI can produce: "I fixed 5 AI shots that looked fake in 20 minutes", "Why your AI videos look like a failed perfume ad", "The setting that makes your Runway shots less artificial". These ideas have a bite. They start from a lived problem. They promise a visible improvement.

I would then push with: "For the 5 best ideas, write a 15-second opening with a punchy sentence, an image to show immediately, and an educational promise." You then get shootable beginnings, not only titles. A YouTube video is often won in the first 20 seconds.

Scenario 2: creative business channel

For a business channel, the trap is to produce too-abstract videos: "How to make money with AI". Everyone has already done it. Your angle must be more precise, dirtier, closer to the field. Example: "I proposed an AI ad to a local restaurant: here is the quote, the brief, the render and the objections."

Prompt to use: "Propose 15 YouTube concepts to help creative freelancers sell AI video services to local businesses. Each concept must include a realistic client case, a commercial objection, a concrete deliverable and a beginner mistake to avoid." There, you get content that smells of the real.

This type of prompt generates videos with a real business promise: quote, brief, client revisions, delivery, usage rights. You are no longer in "AI is going to change the world". You show how to send an email, frame a project, avoid promising a Pixar film for 300 euros, and deliver something clean.

Scenario 3: general-public educational channel

For an educational channel, clarity wins. Your prompt must force the AI to simplify without infantilizing. Example: "I want to explain generative AI to non-technical people, with cinema, cooking and creative-workshop analogies. Propose 20 videos where each complex subject becomes a simple visual experience."

This prompt works because it imposes a pedagogy. You can ask: "For each idea, give a visual demonstration I can film on a desk." A good educational video is not only an explanation. It is a scene. A sheet, three cards, a lamp, a screen, a failed example, then a corrected example.

The logical follow-up: ask for titles in three styles. Direct style: "Understanding prompts in 10 minutes". Conflict style: "Why your prompts give mediocre results". Transformation style: "I turn a bad prompt into a cinematic scene". You will quickly see which style fits your channel.

To connect these ideas to a broader editorial strategy, our article on writing SEO-optimized YouTube descriptions complements this method well.

What beginners break with YouTube prompts

The first break is the too-vague prompt. It produces a too-general list, then the beginner accuses the AI of being bad. No. The AI answered a soft question. If you want sharp ideas, you must give a sharp frame: audience, problem, tone, format, production level, duration, goal.

The second break is the obsession with the title before the angle. A title can be catchy and empty. "I tested ChatGPT for 30 days" says nothing if we do not know why, for whom, with what stake. Before the title, work the promise: "can a beginner write 10 sellable Shorts scripts with ChatGPT in a week?"

The third break is asking for "viral ideas". The word viral attracts the most generic answers. Ask instead for "ideas with strong visual curiosity, clear conflict and measurable transformation". It is less sexy, but much more productive. YouTube does not only reward the subject. It rewards the packaging, the retention, the satisfaction.

The fourth break is copying competitors too literally. Use them as a map, not as a mold. Useful prompt: "Analyze these 5 competitor titles and extract the angle mechanics without copying the subjects. Propose 15 original ideas that use the same psychological mechanics." You learn the grammar of the market without becoming a clone.

The fifth break is ignoring the thumbnail. A YouTube idea must be able to condense into an image. If you cannot imagine a thumbnail in 5 seconds, the idea maybe lacks contrast. Ask the AI: "Describe a simple thumbnail for each idea, with a main subject, a facial emotion, a visual object, a short text of 3 words maximum."

The sixth break is not checking the search intent. Use Google Trends and the YouTube search before recording. If no one searches the subject, it is not necessarily serious, but you must know it. A video can be made for search, recommendation, community or authority. Not always the four.

FAQ: prompts to find YouTube video ideas

What is the best prompt to find YouTube ideas?

The best prompt is the one that starts from a precise audience and a concrete frustration. Write: "Act as a YouTube producer for [audience]. List 30 concrete frustrations linked to [goal], then turn each one into a video idea with angle, title, promise and opening scene." This frame gives the AI enough material to produce useful concepts instead of a generic list. The more you add your production constraints, the more the ideas become shootable.

How many ideas to generate before choosing?

I advise generating at least 30 ideas, then keeping only 5. The first ten are often obvious, the next ten better, and the last ten sometimes more surprising. The sorting is more important than the generation. Score each idea according to audience pain, curiosity, SEO potential, thumbnail and ease of production. An average idea well produced can work, but a weak idea with a beautiful edit stays weak.

Can ChatGPT replace a real YouTube strategy?

No, and that is very good. ChatGPT can speed up the brainstorming, reformulate angles, analyze titles and propose formats. But the strategy comes from your understanding of the audience, your positioning, your publishing pace and your ability to keep a promise. The AI is an editorial assistant. The judgment must stay with you, especially if you want to build a lasting channel rather than pile up interchangeable videos.

How to avoid too-generic ideas?

Add constraints. An idea becomes specific when you specify the audience, the situation, the level, the result and the obstacle. "Video ideas about AI" becomes "videos for freelancers who want to sell an AI ad to a local restaurant with less than 500 euros of budget". It is no longer the same world. Also ask the AI to critique its own proposals and eliminate the angles already seen everywhere.

Should you create videos according to SEO or according to trends?

Both, but not at the same time. SEO builds a lasting base: tutorials, comparisons, guides, frequent mistakes. The trends bring attention spikes, especially if you react fast. A healthy channel mixes the two. If you are a beginner, mostly produce evergreen videos with a clear search intent. Then add more reactive formats when your workflow is stable.

Can you use these prompts for Shorts?

Yes, but you have to change the requested output. For a Short, ask for an idea with a 2-second hook, an immediate visual demonstration, a single lesson, and a clear punchline. Shorts do not have the patience of a long video. Useful prompt: "Turn these ideas into 35-second Shorts with hook, visual action, key sentence and final loop." Keep a single promise per Short.

How to know whether an idea deserves a long video?

An idea deserves a long video if it contains several steps, frequent mistakes, visual examples and a measurable transformation. If you can explain everything in 45 seconds, make a Short. If the subject requires comparison, demonstration, before/after, a real case and nuance, make a long video. The AI can help you test that: ask it to develop an 8-minute plan. If the plan seems hollow, the idea is not deep enough.

Which AI tool to use to generate these ideas?

ChatGPT, Claude, Gemini or Perplexity can all work. What counts is not only the tool, but the quality of the context. Perplexity can help to search for recent trends and sources. ChatGPT is solid to decline angles. Claude is often very good to structure long editorial concepts. Test the same prompt in two tools, compare, then keep the most specific ideas.