The expensive way to make a video
The expensive way is to open a video tool, type something vague, look at what comes out, feel vaguely disappointed, and try again. Repeat eleven times. You have now spent an hour and a chunk of your monthly allowance to arrive at a video you would not send to your mother.
Ten minutes of planning on paper removes almost all of that. Not because planning is virtuous, but because a plan turns "make me a video" into a set of small, checkable instructions that a machine can actually follow.
In plain English
- Hook:
- The opening line or shot whose only job is to stop someone scrolling past.
- Payoff:
- The thing you promised in the hook, delivered. Without it, viewers feel tricked and do not come back.
- Shot list:
- A simple table of what is said and what is on screen, line by line, before anything is made.
- Aspect ratio:
- The shape of the frame. Tall for phones, wide for a laptop or television. Choosing late means recropping everything.
- Call to action:
- The one thing you want the viewer to do next, said out loud. Follow, visit, book, reply.
Hook, middle, payoff
Almost every short video that works has the same three-part shape.
The hook is the first three to five seconds. It makes a promise, asks a question, or shows something odd enough that stopping feels worth it. "Most sourdough is proved wrong" is a hook. "Hi everyone, welcome back to the channel" is a nap.
The middle delivers on the promise, and it holds exactly one idea. One. Two ideas in a 45-second video means neither one lands. If you have three good points, you have three videos and a content calendar.
The payoff closes the loop. It answers the question, shows the result, or gives the one instruction you want followed. It is also where the call to action lives, and it should be a single small ask. Not follow, subscribe, comment, share, visit the website and tell your friends. One thing.
Match the length to where it lives
Broad shapes, not laws, and platforms change their limits constantly, so check before you commit to a format.
Vertical social clips reward brevity: think fifteen to sixty seconds, tightly cut, with the point up front. An explainer on your own website or a product page can breathe a little longer, somewhere around one to two minutes, because the person watching already chose to be there. Anything past two minutes needs a genuine reason: a lesson, a demonstration, a story with steps.
The honest rule is that attention has to be earned per second, and shorter videos have fewer seconds to fumble.
Decide the shape of the frame before you generate anything. Tall video cropped from wide video loses the edges of every shot, and the edges are usually where you put the interesting bit.
The one-page plan
Your plan does not need software. Two columns on one page: what is said, what is seen. Write the script line in the left column and the picture that goes with it in the right.
This is the document that stops aimless generating, because now every prompt you write has a job. You are not asking for "a bakery video." You are asking for eight seconds of hands shaping dough on a floured counter, warm morning light, no faces, to sit under a specific line of narration.
Give me 10 opening hooks for a 45-second vertical video.
Topic: [your topic]. Audience: [who they are, what they already know].
Each hook must be under 12 spoken words and make a promise
the rest of the video can actually keep.
Rank them by which would stop a scroll, and say why for the top three.
Turn this script into a shot list.
For each line of narration give me: the spoken line, the length in seconds,
and a plain description of what is on screen while it is said.
Keep total run time under [45] seconds. Format it as a two-column table.
[paste your script]
Here is my video plan. Act as a blunt editor.
- What is the single idea in this video? State it in one sentence.
- If there is more than one idea, tell me which to cut and why.
- Does the ending deliver what the opening promised? Quote both.
[paste your plan]
โ Weak prompt
Prompt
Make a 30 second video about our new coffee subscription.
Output
Generic shots of coffee cups, a stock-looking cafe, floating text saying Discover Our Coffee, no clear beginning or end.
No hook, no single idea, no ending. The tool guesses, and its guess is whatever the average coffee advert looks like. You will now generate this five more times hoping it improves.
โ Good prompt
Prompt
30-second vertical video. Hook (0 to 4 seconds): a bag of beans with a roast date stamped on it, narration says most coffee you buy was roasted months ago. Middle (4 to 24 seconds): beans being bagged and posted, narration explains we roast on Tuesday and post on Wednesday. Payoff (24 to 30 seconds): a cup on a kitchen table, narration says first bag is half price, link in bio. Warm morning light throughout, no faces, no on-screen text.
Output
Three clearly separated segments that match the narration timings, with a consistent look and one instruction at the end.
Every second is accounted for and every shot has a job. Fewer regenerations, and the version you keep is the one you planned.
Checkpoint
Plan hook, middle and payoff on one page before you generate anything, because a vague prompt costs you time and credits and still gives you a video you will not use.
๐ Quiz
Question 1 of 4What is the job of the hook?