Nike: a problem-solution hook that can sharpen.

The opening, on the record.

A Nike YouTube spot leading with a problem-solution frame. Watch the first three seconds, the text is doing the work the picture should be doing.

Composite 67, SHARPEN.

Verdict ladder: SCALE ≥ 70 and no KPI < 55 · SHARPEN 55–69 · REBUILD < 55 OR two+ KPIs < 45.

Where the score comes from.

Frozen weights: 25 / 20 / 20 / 20 / 15. Bands: Strong ≥ 70 · Moderate 45–69 · Weak < 45.

// KPI breakdown, Nike YouTube US

Beat the Skip

25%

0.5s

75

Strong, dynamic motion and brand familiarity carry the watch, despite text-heavy framing.

Get Noticed

20%

2.0s

70

Strong, Nike's visual language is distinctive even before the swoosh appears.

Brand Impact

20%

4.0s

70

Strong, brand_in_first_3s = yes; the swoosh is on-screen, even competing for attention with the text.

Sell Proposition

20%

MID

65

Moderate, feature-led prop. The text says what the shoe does; showing it would be stronger.

Build Brand

15%

75%

60

Moderate, the visual brand cue arrives after the text. Text dominance flattens the memory trace.

What the pipeline tagged.

A read of the opening pattern.

A problem-solution hook delivered via on-screen text, type-heavy framing, dynamic background motion. The attention-risk tag text_dominance means the saliency layer found the text taking more visual weight than the image. The model reads this as "slide deck, not film": the viewer is being told what the ad is about, rather than shown.

Beat the Skip at 75 and Get Noticed at 70 are still Strong because Nike's brand familiarity and the dynamic background carry the watch. But Build Brand at 60 reveals the cost: the viewer's first read is the text, not the swoosh, so the brand-encoding moment is delayed.

Sell Proposition at 65 reflects feature-led copy that tells rather than demonstrates. Showing the shoe doing the thing the text describes would lift this KPI 8–10 points without changing the brief.

Three edits, ranked by lift.

END

// Evidence, the close

End frame, runner away from camera, swoosh and 'Shop Now' overlay. Strong brand-first opening AND clean close. The opening pattern hits Brand Impact at full; the lift would be a more emotionally distinctive hook second.

1. Replace the on-screen text with a face-first beat. Use existing footage from later in the cut, an athlete mid-stride, as frame 1. Let the type land at 1.5–2 seconds. Get Noticed and Beat the Skip both rise; the text_dominance flag clears.
2. Show, don't tell, the feature. Replace the text claim with a 1-second product moment that demonstrates the same benefit. Sell Proposition shifts from feature-led to demonstration-led.
3. Add a closing CTA. CTA timing is "None", a simple shop/learn-more prompt converts the existing watch into action without breaking the cut.

Where this sits in the pool.

Against the SaliencyLab benchmark for YouTube · Fashion / Sport · US, composite 67 sits in the upper-mid band, above the median for the platform-category combination but capped by the text-dominance pattern.

Pool-wide, ads flagged with text_dominance lose an average of 8–12 composite points to identical creative without text-heavy openings (held-out OOS, YouTube view counts ρ +0.32, n=403, 5-fold CV as of 2026-05-05). The pattern is one of the strongest negative signals in the model.

The general principle.

Text overlays in the first three seconds are an attention tax. The viewer's eye reads text before it processes image, and on a 5-second skip window, that's a luxury you can't afford. The fix is almost always to push the text to second 1.5–2 and let the image earn the opening alone.

The second teaching is about feature-led copy. Telling the viewer that the shoe is breathable scores lower than showing the shoe in motion. Build Brand and Sell Proposition both reward demonstration over description, because demonstration creates two encoding moments (visual + claim) while description creates one (claim alone).

Deeper: First two seconds, a frame-by-frame teardown and Hook rate vs thumb-stop.