A video script second by second

An ad script for a creator is almost always written by the brand, and almost always written as text: a paragraph about the product, a paragraph about the benefits, a CTA at the end. The creator reads it aloud, the audience scrolls past. The problem isn't the words — it's that a video isn't text; it's a sequence of seconds.

01The first three seconds decide everything

On vertical video, 40 to 60 percent of those who opened it make it to the fifth second. Everything you write after that will be seen by half the audience at best. The hook is not a decoration — it's the only part worth arguing with the creator about.

A working hook shows a problem or a result, not the product. "Spent three months figuring out why leads were getting more expensive" retains. "Today I'll tell you about an analytics service" does not.

02What every second should contain

We write the script as a table: second, what's on screen, what's being said, what's on the overlay. Not because we love tables, but because it reveals gaps. If from second 12 to 20 there's a talking head and nothing happens — that's where the retention curve will drop.

A guideline for a thirty-second video: 0–3 hook, 3–8 problem with a specific example, 8–18 product in action, 18–25 result with a number, 25–30 CTA. For a sixty-second video the proportions are the same; the middle stretches.

03The words stay with the creator

We deliver the structure, not the lines. The creator speaks in their own voice, with their own filler words and their own rhythm — that's exactly why people listen to them. A brand that sends a word-for-word script gets a video the audience identifies as an ad within two seconds.

The exception is legal wording and specific numbers. Those we lock in verbatim and ask to be shown on screen, so they don't depend on pronunciation.

04How to read the retention curve

A steady decline is normal. A sharp drop at a specific second means an error: the pacing changed, a list started, a logo filled the screen. A rise in the middle is rare and means viewers are rewinding, usually to a number or a demo.

After the first five videos we usually have a list of three to four spots where this particular product's audience drops off. From then on, scripts are written with those in mind, and retention improves by 10–15 percentage points without changing creators.

05One CTA per video

One CTA per video. "Click the link and subscribe to the channel" generates fewer clicks than just "click the link": the viewer faces a choice and often does neither. Need two actions — shoot two videos.