Ideas with real demand, honest titles, retention-first scripts, descriptions and chapters. Below are 7 copy-ready prompts. Fill in the [BRACKETS], copy, and paste into ChatGPT, Claude, Gemini or any capable assistant.
Most YouTube advice is about the algorithm. The algorithm follows viewers, so the useful questions are simpler: will someone click this, and will they stay? These prompts work on both – packaging and retention – and they diagnose a video that underperformed in the order that finds the real problem fastest.
The 7 prompts
Find video ideas your audience is already searching for
Generate a list of YouTube video ideas grounded in real viewer demand, not guesses.
You are a YouTube strategist. Help me find video ideas people are actively looking for. MY CHANNEL: [CHANNEL TOPIC AND WHO IT IS FOR] WHAT HAS WORKED BEFORE: [TITLES OF MY 3 BEST VIDEOS, OR "new channel"] WHAT I CAN REALISTICALLY MAKE: [talking head / screen recording / faceless with stock footage / animation / other] UPLOAD RHYTHM: [VIDEOS PER WEEK] Give me 20 video ideas, grouped into three buckets: 1. SEARCH VIDEOS (8 ideas) - answer a question someone types into YouTube or Google. For each, write the exact phrase you think they type, e.g. "how to clean a cast iron pan after rust". 2. SUGGESTED VIDEOS (8 ideas) - ride on topics that already get recommended next to bigger channels in my niche. Name the kind of video it sits beside. 3. SERIES IDEAS (4 ideas) - repeatable formats I can make 10+ episodes of without running out. For every idea give: - A working title under 60 characters - The viewer's problem or curiosity in one sentence - Why someone would click THIS video over the 50 others on the topic (the angle) - Difficulty to make: easy / medium / hard, for my production style Rules: - No generic ideas like "10 tips for beginners" unless the angle is specific. - Do not claim search volumes. Instead, tell me how to validate each idea in 2 minutes: what to type into YouTube search, and what to look for in the autocomplete and in the view counts of small channels. - Flag any idea that would need expertise, licensing or on-camera presence I said I do not have.
What you get: 20 ideas in three buckets, each with a search phrase or angle, difficulty rating and a two-minute validation method.
Tip: The validation step matters more than the list. Ideas where small channels (under 10k subscribers) have videos with high views are the ones worth making first.
Write YouTube titles that earn the click honestly
Get title options that are specific and clickable without being clickbait.
Write YouTube titles for my video. WHAT THE VIDEO ACTUALLY DELIVERS: [THE PAYOFF THE VIEWER GETS] WHO IT IS FOR: [VIEWER] MAIN SEARCH PHRASE (if any): [PHRASE] VIDEO LENGTH: [MINUTES] THUMBNAIL IDEA (if decided): [WHAT IS IN THE THUMBNAIL] Give me 12 titles across these styles, 2 each: 1. Plain search title - the phrase people type, stated clearly 2. Outcome title - the result the viewer gets 3. Curiosity gap - a question or tension the video resolves 4. Number / list - only if the video really is a list 5. Mistake / warning - what the viewer is doing wrong 6. Personal experiment - "I tried X for 30 days" style, only if true For each title: - Keep it under 60 characters so it is not cut off on mobile - Put the most important words in the first 40 characters - Mark whether it works WITH the thumbnail (title and thumbnail should say different things, not repeat each other) Then: - Pick your top 3 and explain why in one line each - Write the title as it would appear in search results next to my thumbnail idea, and check they do not repeat the same words Rules: no ALL CAPS words except one if essential, no "You won't believe", no promise the video does not keep. A title that overpromises gets clicks and then kills retention, which hurts the video more than a weaker title.
What you get: 12 titles in six styles, each under 60 characters, with a top-3 shortlist and a title-thumbnail pairing check.
Tip: Title and thumbnail should work as a pair: the thumbnail shows the emotion or the result, the title adds the context. If both say the same thing, you have wasted one of them.
Script a YouTube video that keeps people watching
Turn an outline into a script built for retention, with a strong first 30 seconds.
Write a YouTube script. TOPIC: [TOPIC] TITLE: [TITLE] THE PROMISE (what the viewer gets by the end): [PAYOFF] TARGET LENGTH: [MINUTES] MY SPEAKING STYLE: [casual / calm expert / energetic / dry humour], [PASTE A PARAGRAPH I HAVE SAID OR WRITTEN, if possible] MY NOTES / RESEARCH: [PASTE NOTES, FACTS, STEPS] Structure: 1. HOOK (0-15 seconds) - restate the promise from the title in a new way and show why it is worth staying. No "Hey guys, welcome back". No channel intro. 2. SETUP (15-45 seconds) - what the viewer will learn and why it matters now. One sentence of credibility only if it is real. 3. BODY - in sections. Every 60-90 seconds, add a "re-hook": a question, a surprise, a preview of something coming, or a pattern break. Mark each re-hook with [RE-HOOK]. 4. PAYOFF - deliver the promise fully. Do not hold back the good part for a paid product. 5. END - one clear next step: a specific related video to watch (write it as [LINKED VIDEO: topic]), not a list of social links. Format: - Write it to be SPOKEN: short sentences, contractions, no words I would not say out loud. - Put visual directions in [BRACKETS] on their own line: what should be on screen. - Put approximate timestamps at each section. - Keep the word count to roughly 150 words per minute of my target length. After the script, list: any fact from my notes you could not confirm from what I gave you (so I can check it), and the 3 moments viewers are most likely to drop off, with a fix for each.
What you get: A spoken-style script with timestamps, on-screen directions, marked re-hooks, and a list of drop-off risks with fixes.
Tip: The re-hooks do the heavy lifting on retention graphs. If your retention drops at the same point in several videos, it is usually a section with no re-hook for over 90 seconds.
Write the YouTube description, chapters and tags in one go
Produce a search-friendly description with chapters from your script or transcript.
Write the upload metadata for my YouTube video. TITLE: [TITLE] SCRIPT OR TRANSCRIPT (with timestamps if you have them): [PASTE] MAIN SEARCH PHRASE: [PHRASE] LINKS I WANT TO INCLUDE: [LINKS AND WHAT THEY ARE] CHANNEL CALL TO ACTION: [e.g. subscribe, newsletter, related playlist] Produce: 1. DESCRIPTION - first 2 lines (about 150 characters) must summarise the video and include the main phrase naturally, because that is all most viewers see before "more". Then 2-3 short paragraphs on what the video covers. Then links, clearly labelled. 2. CHAPTERS - start at 0:00, at least 3 chapters, each at least 10 seconds long, each title under 40 characters and descriptive (not "Part 2"). Only use timestamps that exist in my transcript. If I gave no timestamps, write the chapter titles and put [TIME] placeholders. 3. TAGS - 8-12, starting with the exact main phrase, then close variations. Tell me honestly that tags play a small role and titles/descriptions matter more. 4. PINNED COMMENT - one question that invites a real answer related to the video. 5. HASHTAGS - up to 3, only if relevant. Rules: no keyword stuffing, no lists of unrelated keywords at the bottom (YouTube treats that as spam), no misleading claims.
What you get: A description with a strong first two lines, valid chapters, a short tag list, a pinned comment and up to three hashtags.
Tip: Chapters must start at 0:00 and have at least three entries or YouTube ignores them. Make chapter titles searchable phrases - they can appear as key moments in Google.
Diagnose a video that underperformed
Work out why a video got low views or retention, and what to change next time.
Help me diagnose an underperforming YouTube video. VIDEO TITLE: [TITLE] THUMBNAIL DESCRIPTION: [WHAT IS ON IT] TOPIC AND LENGTH: [TOPIC, MINUTES] NUMBERS FROM YOUTUBE STUDIO: - Impressions: [NUMBER] - Click-through rate: [%] - Average view duration: [TIME] ([% of video]) - Where retention drops sharply: [TIMESTAMPS AND WHAT HAPPENS THERE] - Traffic sources: [e.g. browse 40%, search 30%, suggested 20%] MY CHANNEL AVERAGES FOR COMPARISON: [CTR %, AVG VIEW DURATION %] Work through it in this order and stop at the first real problem: 1. PACKAGING - if impressions are decent but CTR is below my average, the title/thumbnail is the issue. Suggest 3 alternative title + thumbnail pairs I could test. 2. PROMISE MATCH - if CTR is fine but viewers leave in the first 30 seconds, the video does not deliver what the title promised quickly enough. Rewrite my first 15 seconds. 3. PACING - if the drop is gradual, find the flat sections from the timestamps I gave. Suggest cuts. 4. TOPIC DEMAND - if impressions are very low, the topic may not have enough interest, or YouTube does not yet know who to show it to. Say which is more likely from the traffic sources. Rules: - Compare against MY averages, not general benchmarks. There is no universal "good CTR". - Do not blame "the algorithm". The algorithm follows viewer behaviour. - Finish with ONE change to make on the next upload, not ten.
What you get: A step-by-step diagnosis identifying the first real problem, with rewritten packaging or opening and one change for the next upload.
Tip: Always diagnose in this order. Fixing pacing on a video nobody clicked is wasted effort.
Turn one long video into Shorts that point back to it
Pick the best moments from a long video and script them as standalone Shorts.
I want to turn my long YouTube video into Shorts. TRANSCRIPT WITH TIMESTAMPS: [PASTE] LONG VIDEO TITLE: [TITLE] AUDIENCE: [WHO WATCHES] Find 5 moments that can stand alone as Shorts (under 60 seconds each). For each: 1. TIMESTAMP RANGE to cut from the original 2. WHY IT WORKS ALONE - does it have a complete idea, a surprise, a clear before/after, or a strong opinion? 3. FIRST 2 SECONDS - the on-screen text or spoken line that stops the scroll. If the original moment starts slowly, write a new opening line to record or place as text. 4. ON-SCREEN TEXT - short captions for key words 5. ENDING - how it ends without "link in bio". Suggest a natural line that makes the long video the obvious next step, and tell me to use YouTube's related video link feature for the Short. 6. TITLE for the Short Rules: - A Short must make sense to someone who has never seen my channel. - Do not cut mid-sentence or mid-thought. - Rank the 5 by how likely they are to work, and explain the ranking.
What you get: Five ranked Short candidates with timestamps, scroll-stopping openings, captions, endings and titles.
Tip: The best Shorts from a long video are usually opinions or surprising results, not the tutorial steps. Steps rarely stand alone.
Plan a month of uploads around one content pillar
Build a realistic 4-week YouTube plan where each video feeds the next.
Plan my next month of YouTube uploads. CHANNEL: [TOPIC AND AUDIENCE] UPLOADS PER WEEK I CAN SUSTAIN: [NUMBER] TIME I HAVE PER WEEK: [HOURS] ONE BIG TOPIC TO BUILD AROUND THIS MONTH: [PILLAR TOPIC] WHAT I ALREADY HAVE (footage, research, old videos): [LIST] Produce: 1. THE PILLAR VIDEO - one comprehensive video on the big topic, scheduled in week 1 or 2 2. SUPPORTING VIDEOS - narrower videos that each answer one question around the pillar and link back to it 3. SHORTS - 2-3 per long video, cut from the long ones, not extra production 4. A CALENDAR TABLE: Week | Day | Video | Type (long/short) | Main search phrase | Links to | Production steps | Hours needed 5. A PRODUCTION CHECK - total hours needed vs the hours I have. If the plan does not fit, cut videos, do not ask me to work more. 6. A PLAYLIST - name and order so viewers binge the set Rules: be realistic about time. A plan I abandon in week 2 is worse than a smaller plan I finish.
What you get: A four-week calendar built around one pillar video, with supporting videos, Shorts, a playlist and an hours check.
Tip: The hours check is the part people skip and regret. Consistency beats volume for small channels.
Where AI actually helps here
- Generating and validating video ideas against what small channels already get views for
- Writing many title options quickly so you can pick and test
- Turning notes into a spoken-style script with re-hooks every minute or so
- Producing descriptions, chapters and pinned comments from a transcript
Where it falls down
- Knowing real search volumes – it will guess numbers confidently if you let it
- Judging your thumbnail without seeing it
- Knowing what YouTube’s current policies say – check the official pages
The mistake almost everyone makes: Writing the script before the title
If you cannot write a title and thumbnail that people would click, the video will not be found however good the script is. Package first: write the title, sketch the thumbnail, then script to deliver exactly that promise in the first 15 seconds.
Free tool: Social Character Counter
Runs in your browser. No sign-up, nothing uploaded.
Questions people ask
Can AI write my YouTube scripts?
It can draft them well from your notes and research. The best results come from giving it your own speaking style and facts, then editing the draft out loud. Scripts that sound like you keep viewers longer.
What is a good click-through rate on YouTube?
There is no universal number – it depends on your niche, traffic source and how many impressions a video gets. Compare each video with your own channel average in YouTube Studio.
Do tags still matter?
A little, mainly for misspellings. Titles, thumbnails, descriptions and viewer behaviour matter far more.