Key Takeaways
- Across more than 3,300 social video ads tested with US and UK consumers, ads made to look like creator content were 30% less likely to be top performers.
- Ads using three or more distinctive brand assets were 30% more likely to be top performers, and were the only trait that substantially lifted brand connection.
- Logos appeared in 97% of the ads and pack shots in roughly half, yet pack shots delivered the weakest lift in brand connection of any branding device.
- In a separate Canva and Harris Poll study, 78% of consumers said they would rather see ads made by people, even if AI could produce better ones.
The prevailing advice on social video has been to stop looking like an ad. Shoot it on a phone, borrow the cadence of a creator, and let the feed do the rest. New testing suggests that advice is winning the scroll and quietly costing brands the one thing the ad was paid to deliver: being remembered as theirs.
Zappi's fourth annual State of Creative Effectiveness report, released on October 5, measured real consumer response to more than 3,300 social video ads on TikTok, Instagram, Facebook and YouTube across 21 categories in the US and UK. Each ad received a Creative Power score covering resonance, brand connection, persuasion, distinctiveness and emotional intensity, and the top 20% were classed as top performers.
Looking Like a Creator Is Not the Same as Being Native
The headline finding cuts against a lot of current briefs. Ads designed to resemble creator content were 30% less likely to be a top performer, and had the weakest brand connection of any creative characteristic in the study.
The distinction that matters is who the ad sounds like. According to Advanced Television's summary of the findings, brand-made creative designed to feel native to social ranked among the strongest formats tested. The feed rewards fluency in its conventions. It does not appear to reward a brand pretending to be someone else.
That chimes with what consumers say about authenticity more broadly. In the Canva and Harris Poll study of 1,415 marketing leaders and 3,547 consumers across seven countries, 78% of consumers said they would rather see ads made by people, even if AI could produce better ones, and 70% said they can usually tell when an ad was generated by AI. Audiences are getting sharper at spotting a costume, whether the costume is a creator aesthetic or a synthetic one.
The Three Jobs No Longer Line Up
Zappi's most useful framing is that standing out, persuading and building the brand have drifted apart, and different creative choices drive each one.
Attention is the easiest to buy. Ads built around a strong visual hook were 2.4 times more likely to be top performers, with a marked gain in distinctiveness. Persuasion followed a different recipe: it was strongest when the ad carried a clear product benefit and ended on a positive resolution.
Brand connection followed neither. Ads using three or more distinctive brand assets, such as colours, slogans and recurring characters, were 30% more likely to be top performers, and those assets were the only creative characteristic studied that substantially improved brand connection. A hook gets the view. Only consistent brand codes make sure the view is credited to the right company.
"Content has scaled faster than the systems marketers use to validate it and prove its value," said Nataly Kelly, Zappi's chief marketing officer. The volume problem is real: Wistia's 2026 State of Video report, drawn from more than 900 professionals and 13 million videos, found more than a third of teams already use AI in their video workflow, while 46% are holding production budgets flat. More output on the same money leaves little room to test whether each cut actually carries the brand.
The Logo Is Doing Less Than Everyone Assumes
The branding data contains the finding most likely to change an edit. Logos appeared in 97% of the ads studied, which makes them close to universal and therefore close to invisible as a differentiator. Pack shots appeared in roughly half of all ads, yet delivered the weakest lift in brand connection of any branding device. Brand colours and slogans did more work.
In practice, the end card with a product and a logo, the default branding move in most social edits, is the least effective way to make an ad memorable as yours. The codes that work are the ones woven through the footage from the first second.
Audience age complicates the picture further. Respondents aged 25–34 were more likely than 18–24-year-olds to say they love digital advertising, and the youngest group scored relatively low across emotion, persuasion, distinctiveness and brand connection. The audience most briefs chase with creator-style work is the hardest one to move.
What Social Teams Should Change Now
- Brief for native, not for imitation. Brand-made creative built for the feed outperformed ads dressed as creator content, so write briefs around platform conventions rather than borrowed personas.
- Audit your distinctive assets. Count how many recognisable codes appear in each cut. Three or more was the threshold associated with a 30% higher chance of top performance.
- Move branding out of the end card. Pack shots gave the weakest lift. Put colours, slogans and recurring characters inside the story where viewers actually watch.
- Score the three jobs separately. A hook that wins attention will not necessarily persuade or build memory. Measure distinctiveness, persuasion and brand connection as distinct outcomes.
- Pair every hook with a benefit and a resolution. Persuasion peaked when a clear product benefit met a positive ending, so do not let the opening crowd out the payoff.
- Keep humans visible in the work. With 78% of consumers preferring ads made by people, AI is best used for speed behind the scenes rather than as the face of the brand.
None of this argues against social-first creative. It argues against confusing a format with a strategy. The ads that worked hardest in Zappi's data looked at home in the feed and still left no doubt about whose ad it was.
