The Queensberry x DerbyFest project film shows a useful commercial lesson: a campaign film and a performance asset are related, but they do different jobs. The hero film had to sell the atmosphere of the day. The paid-social test had to answer a sharper question: which creative format gets people to click while they are still in the buying moment?
That is where AI advertising agency work becomes operational, not ornamental. Generative video production can expand a campaign world quickly, but the commercial value appears when the team turns those assets into a testable system: clear variants, consistent targeting, measured click behavior and fast decisions.
The challenge: turn local event attention into ticket-page action
DerbyFest needed to feel bigger than another day at the races. The campaign had a specific venue, a specific local catchment and a specific entertainment hook. Epsom Downs is not a generic backdrop; it is the meaning of the offer. The creative had to connect racing, music, social energy and the urgency of booking.
The opening production goal was narrative clarity. The opening business goal was sharper: make people understand the event quickly enough to leave the feed and move to the ticket page.
The creative approach: build the world, then test the buying trigger
The work split into two layers. First came the campaign world: a hero film with race-day atmosphere, Pixie Lott energy, event identity and cinematic pace. Then came the testing layer: four paid-social versions built to isolate different attention routes.
The variants included a 37-second hero film, a 15-second cutdown, an artist-led call to action and a deliberately lo-fi viral remix. The point was not to prove that one craft style is always best. The point was to learn which creative shape matched this audience, this platform and this moment.
The production stack: AI filmmaking plus disciplined performance editing
The stack combined campaign strategy, AI filmmaking, edit design, short-form versioning and paid-social measurement. In practical terms, that means the team could make the premium asset and the lower-polish test assets inside the same creative logic.
That matters for AI video commercials. Brands often over-invest in the single polished film, then under-invest in the variants that make the media spend learn. This project treated the hero film as one asset in a system, not the whole campaign.
The result: the strangest asset did the most commercial work
The strongest test was not the most polished edit. The viral remix opened with a left-field sports clip, then snapped into the DerbyFest event creative. It produced 1,067 website clicks in the 24-hour comparison, 370% more clicks than the 37-second hero film, and a 7.12% click-through rate.
That does not make polish irrelevant. It makes sequencing important. The premium film made the event feel real. The remix did the feed-native interruption job. Together, they show why AI ad creation should be judged by the role each asset plays in the funnel.
Why it matters for brands
The useful insight is simple: creative testing should not be a bolt-on after production. It should be designed into the brief. Meta's own ad guidance repeatedly points to the relationship between objective, audience, placement, creative and measurement, including the value of diversified creative and A/B testing. The DerbyFest work turns that into a compact production model for brands that need demand now, not only a beautiful recap later.
AI agents for marketing can help route briefs, produce variant checklists, monitor creative performance and surface winners faster. But the creative judgment still matters: know what each asset is supposed to do, then build enough variations to let the market answer.
Context links for the operating model
The wider platform logic supports the same conclusion. Meta's ad creative guidance stresses quality visuals, compelling messages and mixed formats. Meta's Reels ads guidance points to short-form testing and placement fit. The Jockey Club's Epsom media centre gives the venue context behind Derby Festival demand.
For brands, the originality angle is not "make more versions." It is to decide what each version is trying to prove. Hero assets build belief. Cutdowns compress the pitch. Artist-led edits test recognition. Lo-fi remixes test interruption. A stronger commercial production system makes room for all four.
What brand teams should take from it
- Write the commercial job before the shot list. A beautiful asset can still be the wrong click driver.
- Pair AI commercial production with measurement design. Generative speed only matters if the variants answer real questions.
- Use polish and platform-native interruption together. They solve different parts of the funnel.
- Treat creative learning as reusable infrastructure. Every test should improve the next campaign brief.
Vertical Haus helps brands turn story, AI production, short-form variants and performance learning into one commercial engine.