The Prehook¶

A short visual hook at the very start of the video — designed to stop the scroll in the first 1-2 seconds. The prehook is structurally separate from the main body, which is why it gets its own concept page.

What a prehook is¶

The first 1-3 seconds of a video, before the avatar starts the main script. Usually one of:

A tight selfie of the avatar mid-sentence, looking surprised / shocked / leaned-in
An on-screen text overlay teasing the content ("If you have these 3 symptoms...")
A quick visual signal (the product appearing briefly, a face flash, a graphic)
A pattern-interrupt frame (something visually unexpected)

The prehook is not the main hook of the script. The main hook is the avatar's opening line. The prehook is what stops the scroll long enough for the main hook to land.

Why prehooks matter¶

TikTok / Reels / Facebook all decide whether to surface your video based on early engagement signals — specifically, what percentage of viewers don't swipe away in the first 2 seconds.

A great script with a generic visual opener (just the avatar looking at the camera) often loses the first-second swipe-test. A great script with a prehook (something visually arresting before the avatar speaks) holds the viewer long enough for the hook line to do its work.

When to use a prehook¶

Add a prehook when:

The first-3-second retention rate has been weak (your videos drop off too fast)
The opening of the main script is strong but slow to ramp up
You're targeting a hyper-competitive niche where most videos look similar
You're trying a new format and want to maximize chances of cutting through

Skip a prehook when:

The main script's opening line is itself a strong pattern interrupt
The platform format already has a built-in attention-grabbing structure (e.g., a recipe demo's first ingredient drop)
You're doing a video copy of a source video that doesn't have one
The brief budget is tight and you don't need the extra clip

How a prehook fits in the workflow¶

flowchart LR
    P[Prehook clip<br/>1-3 seconds] --> M[Main body<br/>scene 01]
    M --> M2[Scene 02]
    M2 --> M3[Scene 03]
    M3 --> CTA[CTA]

The prehook is a separate clip with:

Its own image gen (the still frame)
Its own Veo3 gen (the animated clip)
Its own Approve gate

It's typically wired before Scene 01 in the workflow. In a .nbflow, it lives as its own scene node — not part of the main body's wiring.

Adding a prehook to an existing workflow¶

Adding a prehook is a Lvl 4 structural change — it adds nodes to the graph. See Chapter 5 — Lvl 4 Structural Change.

The diff:

Original:  Scene 01 → Scene 02 → ... → Scene 08
New:       Prehook → Scene 01 → Scene 02 → ... → Scene 08

Adds: image gen + Veo3 + Approve for the prehook clip. The rest stays untouched.

Designing a prehook¶

The Visual Planner produces the prehook spec during storyboarding (when a prehook is requested in the brief). Typical specs:

Tight selfie of the avatar surprised / leaning in: Camera close to the face, eyes wide, mouth slightly open mid-word, body leaned forward. Reads as "wait, this is interesting."
Text overlay teaser: On-screen text appears for 1-2 seconds before the avatar speaks. Examples: "Most women don't know this" or "If you have brain fog AND...". Bold serif, animated word-by-word.
Product / object signal: A quick frame showing the product hero, then cut to the speaking. Works for product-aware audiences who recognize the product immediately.
Pattern-interrupt frame: Something visually unexpected — a face from an unusual angle, a hand entering frame, a sudden color flash. Used carefully because too much weirdness loses people.

The choice depends on the audience and the rest of the workflow's style. Don't put an aggressive pattern interrupt on a calm wellness workflow — tonal mismatch.

Prehook vs. hook (the script's first line)¶

A common confusion:

Prehook	Main hook
Visual element at the very start	The first spoken sentence
1-3 seconds	First 4-6 seconds
Designed to stop the scroll	Designed to keep the viewer watching past the scroll-stop
Lives as its own scene (own image + Veo clips)	Lives in the main body's Scene 01
Optional (some workflows skip it)	Required (every script has an opening line)

Both can exist in the same workflow. They work together: prehook = scroll-stop, main hook = retention.

Prehook in post-production¶

The prehook is generated as a standalone clip — in post-production, the editor:

Lays the prehook clip on the timeline first
Cuts directly to the first speaking scene (no fade, no transition — abrupt cut)
The prehook clip's audio is usually silent (or muted) — the main script's voiceover starts at Scene 01

If the prehook has text overlay, the text is added in post (the Image Prompter doesn't include text in the generated still — the post-production editor adds it as a caption layer).

Common prehook mistakes¶

Making the prehook longer than 3 seconds: Defeats the purpose. The prehook is a teaser, not a setup. If it's longer than 3 seconds, it's just a slow opening.
Using the same prehook across multiple variants: The prehook is where you want to test most aggressively — it's the highest-leverage second of the video. Vary the prehook between variants more than the main script.
Forgetting to test without it: Sometimes the main hook is strong enough that the prehook adds nothing or even slows engagement. A/B test: same script, with and without prehook. The data tells you which works.
Treating prehook generation as part of B-roll: It's not. B-roll is a cutaway within the main video; prehook is the opening. They're structurally different and generated differently.

When you're ready¶

→ Next: PiP Format — picture-in-picture composition, the mixed-media alternative to full-frame talking head.