How to Find YouTube Video Ideas Worth Making

01 /Start with the viewer, not the topic

“Sleep,” “fitness,” and “AI” are subjects, not yet video ideas. An idea becomes useful when it names a person, a question, and a reason to click now. “Why do I wake up exhausted after eight hours?” is closer to a video than “sleep science.”

Write a viewer sentence before brainstorming titles: “Someone who ___ wants to ___ but keeps getting stuck because ___.” This forces you to describe the problem in everyday language rather than creator jargon. It also gives you a filter for rejecting attractive topics that will not serve your audience.

• Audience: Who is this for, specifically?
• Tension: What do they misunderstand, fear, want, or keep failing to do?
• Payoff: What will be clearer or easier after watching?

02 /Build a small, deliberate reference set

Choose five channels close to your topic, including smaller creators. Collect around 15–20 relevant videos: their titles, thumbnails, duration, upload dates, view counts, and the question each one promises to answer. You are collecting examples, not copying scripts.

Keep a separate column for your interpretation: what did the creator make interesting? Was it an unexpected comparison, a personal challenge, a myth, a visual demonstration, or a clear practical outcome? The interpretation column is where research becomes creative work.

• Search the audience question in the words viewers would use.
• Include both successful and ordinary uploads from the same channel.
• Record when the numbers were observed, because views change.

03 /Look for unusual performance, with context

A video doing much better than a channel's usual uploads can signal a promising angle. An informal starting point is its views relative to the channel's median recent video. But the raw ratio is not a verdict: video age, audience size, topic seasonality, paid promotion, and external traffic can all distort the comparison.

Instead of asking “Which title got the most views?”, ask “Which framing performed unusually well for this creator, and what other evidence supports that interest?” Look for repeats across different creators, recurring comments, and search suggestions. One viral video is a clue; several independent clues make a better hypothesis.

• Compare videos of roughly similar age when possible.
• Never assume view count alone proves a thumbnail or title caused success.
• Write down what you cannot know from public data.

04 /Give the idea a unique point of view

Research tells you what deserves attention. Your own experience, evidence, access, taste, or visual language tells you why your version should exist. If others made “Top 10 camera mistakes,” you might demonstrate one mistake with a side-by-side experiment or build an inexpensive fix on camera.

Write a one-sentence promise: “By the end, you will understand ___ because I will show ___.” If the second blank is generic, keep working. A better angle is usually a clearer promise, not a louder adjective.

• Use an original test, dataset, demonstration, or story.
• Decide what a viewer should remember one day later.
• Remove any claim you cannot support in the video.

05 /Score the shortlist before filming

Choose your best five candidate ideas. Rate each from 1 to 5 for audience relevance, curiosity, your unique contribution, and production feasibility. These are editorial judgments, not a predictive growth model. A slightly less exciting idea you can finish this week often beats the perfect idea that needs three months.

For the top candidate, draft three titles, one thumbnail concept, a 20-second opening, and a short outline. If the opening cannot deliver the title's promise, revisit the idea before investing in production.

• Can I state the value without clickbait?
• Can I show the payoff, rather than only describe it?
• Can I finish a credible version with the resources I have?

06 /Publish, learn, and make the next one better

Once the video is live, separate packaging signals from content signals. Impressions and click-through rate can inform title and thumbnail questions; audience retention and qualitative comments can help you examine whether the video delivered. Each metric is contextual, especially for small samples.

Log one specific lesson you will change next time. A 100-video practice compounds when the research process remembers what the last upload taught you.