Reading the storyboard rebuild

The storyboard rebuild lives near the bottom of every Deep Bench report (after Rewrites). It's 4–6 numbered beats, each with a time window (e.g., 0–2s, 2–5s), an action description, a suggested on-screen caption, and a one-line rationale tied to retention or conversion.

It is nota list of small edits to your current footage. It's a reshoot plan — the model rebuilds the structure from scratch around the same product, then tells you what each beat is for. The shot-list edits section is for incremental fixes; this is for "the structure is wrong, start over."

Rule of thumb: trust the storyboard rebuild when your hook score is below 30 AND your native-feel score is low. That combination means the structure isn't working AND the production reads as a commercial — both signals point at a rebuild, not a patch.

Trust it less when your hook score is above 40 and your rubric scores are all in the 50–70 band. In that range, you've got a working ad that needs polish — the shot-list edits + caption rewrite section will do more for you than a teardown.

Each beat is built to answer two questions: what happens and why it earns the watch. The rationale field is the answer to "why." If you only read one field per beat, read that one.

Common patterns:

• —Beat 1 is always the hook. 0–2s, sometimes 0–3s. It will look nothing like a product shot — it's usually a face, a tension moment, a question, or a sharp on-screen claim. The rebuilt hook is almost always 3–5 words on-screen + a tight cut.
• —Beat 2 introduces the problem or curiosity gap. 2–5s. The viewer's decided to keep watching past 3 seconds; this beat earns the next 5. Stakes are introduced here — what could go wrong, what the audience is missing.
• —Beats 3–4 deliver the product reveal + proof. 5–15s. This is where most ads put the product shot — and this is where the rebuild will tell you to add B-roll, detail texture, before/after — anything sensory. If you're running affiliate, this is where the code appears on screen.
• —The closing beat is always the CTA. The rebuild will spell out both the voiceover line and the on-screen graphic. Both. Not one. The rubric scores CTA channels separately for a reason — viewers who watched with sound on need the voice, viewers muted need the graphic. A CTA without a baked-in graphic caps at 60 on the rubric no matter how clever the voiceover.

You don't have to reshoot from scratch — most users rebuild Beat 1 (the hook) and Beat N (the CTA) using existing source, and only reshoot the middle if their score still flatlines after that.

The standard workflow:

1. Re-cut Beat 1 using a different opening from your existing footage. Look for: a face close-up, a sharp gesture, a problem moment. If you don't have one in the can, this is a 30-second selfie shot.
2. Bake in the on-screen caption from Beat 1 — sticky text, high contrast, on frame 1. This alone can lift hook rate 5–10 points.
3. Use the macro-shot prescriptions (the section right above the storyboard rebuild) to fill in any B-roll the middle beats call for. Most are 5-second clips you can shoot solo.
4. Re-cut the CTA beat with the spelled-out voiceover line + baked-in graphic.
5. Re-score the new cut. If you don't see at least a 10- point hook lift, the structure issue is deeper than a re-edit — that's when you reshoot from scratch.

The rebuild is opinionated about structure but neutral about two things on purpose:

• —Music / trending sound. We don't recommend specific sounds — by the time you'd shoot the cut, the trending sound landscape would have rotated. Look at TikTok Creative Center's top sounds list the day you publish.
• —The exact wording of the voiceover. The rebuild gives you the on-screen captions and the action direction; the voiceover line is yours to write. The Rewrites section gives you three alt hooks if you need a starting point.

The storyboard rebuild is the "output" end of the v3.34 narrative layer. The pipeline that produces it:

1. Vertical archetype names the pattern top performers in your category use ("reflective-texture-reveal" for hair, "before-after-confession" for fitness, etc.).
2. Emotional positioning names what the viewer is actually buying — the deep need under the surface need. "Confidence" not "hair gloss."
3. Hook valence picks the emotional tone of the hook (confession / warning / transformation / shock / etc.).
4. Storyboard rebuild assembles all of that into a concrete 4–6 beat shoot plan.

Each piece is useful on its own, but reading them as a stack is the right way to absorb the rebuild — Beat 1 is shaped by the hook valence and emotional positioning above it; Beats 3 and 4 are shaped by the vertical archetype. The model walks its own work; you can walk the work backward to understand why each beat is what it is.