Every ad, broken down for you.

The storyboard, shot by shot. The script, word for word. The on-screen text. All of it done before you open the ad.

AI presenter contrasts vague prompt with proven fixHubSpot · 6 shots · running 24d

HOOK · 0:00–0:04.0

“You're prompting your AI wrong.”

Opening shot — on-screen text: wrong. Here's why.

Why this ad works

The ad opens with a callout that stings just enough to stop the scroll ('You're prompting your AI wrong') without being insulting, immediately creating a knowledge gap the viewer wants closed. It then walks through a mini-diagnosis (vague prompt → generic output) before resolving into a concrete, zero-risk fix: a free 1,000+ prompt download. Using a labeled AI avatar as the presenter is a clever meta-move for an AI-marketing product — the format itself demonstrates the brand's subject matter (AI content) while staying transparent about what it is.

Structure

Hook (0-4.5s): AI avatar names the viewer's mistake and promises a reveal ('here's why'), classic pattern-interrupt callout. Build (4.5-18s): the ad shows the actual failure mode (a generic 'write an Instagram caption' prompt) then pivots to the landing page itself, scrolling through real feature cards (superpowers, customizable prompts) to prove the resource exists and has substance. Payoff (18-25s): a concrete example prompt ('develop a LinkedIn content strategy...') makes the abstract benefit tangible, and the CTA to download begins. Close (25-30s): a screen-recording of the actual PDF with numbered prompts visually substantiates the '1,000+' claim before the final CTA line, 'get started with better prompting.'

What to take away

Open with a specific, common mistake your audience is making right now — a mistake-callout earns more attention than a generic benefit statement.
Show the broken behavior on screen (the vague prompt box) before showing the fix — contrast sells harder than description.
Use screen-recordings of the real deliverable (landing page, PDF) instead of static graphics — it converts a claim ('1,000+ prompts') into visible proof.
If using a synthetic or AI-generated presenter, label it plainly — transparency itself becomes a talking point and builds unusual trust for an AI-related offer.
Anchor the lead magnet with one ultra-specific example (a real prompt) rather than only describing the category — specificity makes the free asset feel immediately usable.