The direct answer
The 3-second hook rule holds that a short-form video has roughly three seconds to earn the viewer's continued attention before they scroll away, which means the opening frame and opening line carry disproportionate weight relative to the rest of the video's runtime.
What a working hook looks like
A spoken hook that lands in about 10-14 words and names a specific stake for the viewer outperforms a generic opener. The visual hook — what's on screen in frame one — should be legible and specific before a single word is spoken, since many viewers decide whether to keep watching before the audio even registers.
Common hook mistakes
Slow builds, brand logos before any value is delivered, and generic openers like 'in this video I'm going to show you' all burn through the 3-second window without giving the algorithm — or the viewer — a reason to stay.