One brief, two shapes: a 30-second ad in 9:16 and 16:9

Ask for “a 30-second ad” and you usually need it twice. A tall version for phones, Stories and Reels, and a wide one for a website, a YouTube pre-roll or a screen at an open day. Made as two separate jobs, they drift: different wording, different cuts, one of them checked and the other not.

We make both from a single written description of the ad, with one command. Building it that way forced us to decide a few things in advance, and those decisions are most of what makes the result reliable.

Give the ad a fixed shape

Every ad we make has the same four parts:

Hook, the first 3 seconds. One line that stops a thumb. If it fails, nothing after it is seen.
Body, about 3 to 22 seconds. Two or three beats, each one benefit. Never a list of features.
Offer, around 22 seconds. The single concrete thing being offered.
Call to action, around 26 seconds. One action, one verb.

The software enforces it. If an ad’s first beat isn’t a hook, or it has no call to action, or it runs outside 20 to 40 seconds, it won’t build. That sounds strict, and it is meant to be: a brief that can’t fit the shape is usually a brief that isn’t finished yet.

Build it for silence

Video in a social feed typically starts playing with the sound off. So the words on screen aren’t subtitles for a voiceover. They are the ad. There is no narration at all, and the music, if there is any, is decoration: losing it must cost nothing.

That changes how you write. Every line has to make sense alone, in a few words, read at a glance. And it gives us one test that matters more than any other: watch every render with the sound off before it goes anywhere.

Keep words out of the phone’s own controls

A tall video on TikTok, Reels or Stories has the platform’s own furniture on top of it: the username, the caption, the buttons. On a 1080 by 1920 frame, the bottom 280 pixels and the top 150 or so can be covered. Nothing readable goes there.

So the hook sits at the top of the frame, and the body, offer and call to action are anchored to the bottom. A two-line beat and a four-line beat then finish in the same place, and the eye doesn’t have to hunt.

Fail loudly when the words don’t fit

We cap a line of on-screen text at 26 characters in the tall version and 34 in the wide one. A longer line isn’t wrapped or shrunk to fit. It stops the build with an error.

The reason is that a quietly trimmed call to action is worse than a failed render. On the first full build, this caught a real one: a line that read “Somewhere they learn to read,” which came to 29 characters. We rewrote it. Nobody had to spot a chopped sentence in a finished video.

Choose how a wide picture fills a tall frame

Turning a wide still into a tall frame has two honest options. Fill crops it to the tall shape: punchy, and it fills the screen, but you lose the sides. Mount keeps the whole picture and places it over a blurred, darkened copy of itself, which preserves the composition but only fills about a third of the screen.

We default to fill, and use mount sparingly. On our own short vertical cuts, the mounted ones held attention for an average of only 1.67 seconds.

What this means for your ad

You give us one brief. You get both shapes back, worded the same, built to the same rules, and checked with the sound off. If you have a school, a shop or a service and want short video that works on a phone and on a screen, our social media work and our animation both start here.