YouTube Description Generator
A readable, search-friendly YouTube description — opening line, key points, CTA, keywords, and optional chapters.
Fill out the form above and your results will appear here.
What is YouTube Description Generator?
Writes a full YouTube description — opening line, main description, key points, CTA, and suggested keywords — and auto-generates timestamped chapters if you paste a transcript or talking points.
How does it work?
- 1.Reads your video title, topic, audience, keywords, tone, CTA, links, and optional transcript — truncated to the first 4,000 characters if longer.
- 2.Writes for humans first and search second, explicitly avoiding keyword stuffing and repeating the same point across the opening line, description, and key points.
- 3.Only fills in chapters if you supplied a transcript or explicit timestamps; otherwise returns an empty chapter list rather than inventing timestamps.
- 4.Weaves supplied links in naturally and never invents a URL you didn't give it.
- 5.If you supply a CTA, keeps it verbatim (light phrasing tweaks only); otherwise writes one from your tone, topic, and audience.
How to use it
- 1.Enter the video title and describe what the video is about.
- 2.Fill in who the video is for, and optionally add target keywords.
- 3.Pick a tone, then optionally add your own CTA and any links to include.
- 4.Optionally paste a transcript or key talking points — this is what unlocks chapters.
- 5.Click Write my description and review the opening line, description, key points, CTA, keywords, and chapters.
Examples
Input
Title: "Why Most Freelancers Underprice Their First 10 Clients" · Audience: New freelance designers/developers · Transcript pasted
Output
Opening line hooks on the underpricing problem; chapters generated from the transcript, e.g. "[00:00] Why underpricing happens" / "[03:15] The fix"; CTA and keywords filled in from tone and topic.
Input
Same title, no transcript supplied
Output
Same opening line, description, key points, CTA, and keywords — but chapters returns an empty array instead of guessed timestamps.
Tips
- •Paste a transcript, or even just rough timestamps and talking points, specifically to unlock chapters — skipping it returns an empty chapter list by design, not a bug.
- •Leave CTA blank if you want it inferred from tone, topic, and audience; type your own exact line if you need a specific ask kept intact.
- •List links exactly as you want them to appear — the model never invents a URL, so an omitted link never shows up in the output.
- •This is the one tool running Claude Haiku instead of Gemini or GPT-4o-mini — output quality is worth the higher cost here, which is also why it has a lower daily free-tier limit (4/day vs. 10/day on most other tools).
- •Keep Keywords to actual search terms your audience would type, not generic tags — they feed suggestedKeywords in the output, not hashtags.
Use cases
- •YouTubers who want a complete, SEO-aware description without writing one from scratch per video.
- •Educational or tutorial channels that want auto-generated chapters from a transcript.
- •Channels repurposing a podcast or livestream transcript into a structured video description.
- •Anyone who needs a description with real, non-invented links woven in naturally (course links, socials).
Frequently asked questions
What exactly does this generate?
An opening line, main description, key points, CTA, suggested keywords, and — if you paste a transcript or timestamps — a chapter list.
Which AI model writes the description?
Anthropic Claude Haiku 4.5, prompted for natural, non-AI-sounding writing — no buzzwords or filler phrasing. This is the one tool where output quality justifies a pricier model.
Will it keyword-stuff the description?
No — it's explicitly instructed to prioritize readability for humans first and search optimization second.
Does it write the CTA for me, or do I need to provide one?
You can leave it blank and it's generated for you from your tone, topic, and audience — or type your own exact call to action and it'll be worked into the description instead.