
Mission 5 of 6
Package your video so it gets clicked and then gets watched.
YouTube only shows a video to more people once a few have clicked it and stayed. You'll build the title, thumbnail and first 30 seconds so they make one promise and pay it off fast.
What you'll do
Use one formula for search and one for browse, with rules that stop a good title lying.
Render an object-plus-text thumbnail from HTML, then test three variants in YouTube's own Test and Compare.
Check the first 30 seconds, then set up chapters, end screen, description and pinned comment.
- A thumbnail that still reads at phone size
- A packaging checklist you run on every upload
- The export and description-chapters.txt from edit long-form with code
- A frame from the video and the tool's logo as PNGs in assets/
- Python with Playwright, from the editing mission, for the thumbnail render
- A YouTube channel with the Studio open
Your progress0/6
- 01Titles: one formula for search, one for browse10 min
- 02Thumbnails without a face: object, text, contrast14 min
- 03Three variants, one difference each8 min
- 04The first 30 seconds as a retention gate8 min
- 05Chapters, end screen, cards: keeping them on the channel7 min
- 06Description, pinned comment, and the packaging checklist8 min
Titles: one formula for search, one for browse
A search title is the query plus a reason to pick yours. A browse title is a specific curiosity you can pay off. Know which one each video is before you write a word.
You can make the best build-along in your lane and nobody will see it if the title doesn't earn the click. So the packaging starts with the title, and the title starts with a decision you've already made.
Go back to plan.md: each row was marked search or browse. That decides the title's shape.
Two kinds of title
A search title puts the words a person typed at the front, because that's what they scan for in results.
A browse title appears next to other videos with no query in anyone's head, so it has to create the itch itself.
Save the formulas and rules below as titles.md. They show the shape of each kind, and the rules that keep a title honest.
titles.mdSEARCH titles (the query first, then the specific promise) [Tool] vs [Tool]: which one to actually install [Tool] tutorial: [outcome] from a blank folder [Tool] review after a week of real work How to [outcome] with [tool] (no [thing people hate]) [Tool] on Windows: the setup that works BROWSE titles (specific curiosity, paid off in the video) I replaced [paid product] with a [specific thing] and it's fine I tested 5 [category] tools for a week. One is worth it. This open source [category] tool does [thing] and almost nobody uses it [Tool] is better than [tool] at the one thing that matters I let [tool] [do risky thing] on a real project Rules Under 60 characters where you can; the first 40 must carry the meaning (mobile truncates). Tool name first for search, claim first for browse. Every title must be paid off on screen. If the video can't show it, the title can't say it. Numbers are specific (5 tools, a week, 7B) never vague (tons, insane, ultimate).
Write ten, then cut to three
Write ten titles for the video. Not three, ten.
The first five are the obvious ones everyone writes, and six to ten are where the specific one appears.
Pick the three you'd click yourself and read each against the rules.
The one most people get wrong is pay-off. 'I let the AI delete my code' is a strong title only if the video shows the AI deleting your code.
If it merely nearly did, the title is 'It nearly deleted my code', which is still good and doesn't make the retention graph fall off a cliff at the minute viewers realise they were had.
Test the one you chose
Search titles have a second test: type the first four words into YouTube search and look at the autocomplete. If your phrase appears, the demand is real. If it doesn't, rephrase to the nearest one that does.
Browse titles have a different test: say it out loud to someone who doesn't care about your lane and watch whether they ask 'and?' That 'and?' is the click.