Paid Social

Zappi Report: Celebrity Ads Grab Attention but Fail to Drive Sales

Zappi Report: Celebrity Ads Grab Attention but Fail to Drive Sales
Zappi Report: Celebrity Ads Grab Attention but Fail to Drive Salesean254 / pixabay

Zappi's latest report finds celebrity and influencer partnerships often fail to move sales, while ads anchored in distinctive brand assets perform best, its CMO told an ADWEEK House panel.

By Sophie Lindqvist

3 min read

At a glance

  • Zappi's annual report "The State of Creative Effectiveness" found celebrity and influencer partnerships often do not impact sales.
  • Zappi CMO Nataly Kelly said top-performing ads draw a throughline to distinctive brand assets kept front and center.
  • Steve Phillips said scroll ads have become formulaic because guidelines outrank creative in brand processes.
  • Zappi built an AI-generated synthetic audience to test campaigns with thousands of assets at scale.
  • Key campaign pieces should still be tested with humans, Phillips said, while derivative versions can use synthetic consumers.

Simply partnering with a celebrity or producing influencer content often has no measurable impact on sales, according to Zappi's latest annual report, "The State of Creative Effectiveness," shared during an ADWEEK House Advertising HQ fireside chat.

The finding cuts against a core assumption of social-first marketing budgets: that borrowed attention converts. Zappi's report analyzed ads across social media to determine what actually engaged viewers, and the data points to a gap between attention and connection.

"What the highest performing ads do is draw a throughline to distinctive brand assets and make sure that they are front and center," said Nataly Kelly, CMO of Zappi, who presented the findings alongside Steve Phillips, the company's chief innovation officer, founder and chair, in conversation with ADWEEK editor-in-chief Ryan Joe.

What counts as a brand asset?

Phillips pushed back on the common definition. "Every client thinks they know what their distinctive brand assets are, but it's not just the logo, it's not just the colors," he said. "It might be the type of story arc in your advertising or the type of humor related to you, and that storytelling can be a much more distinctive brand asset."

For creative teams, the implication is direct: assets that carry recognition — a recurring narrative structure, a signature tone of humor — may outperform visual identity markers in feed environments where users decide in fractions of a second whether to keep scrolling.

Why does 'struggling with scroll' matter?

Phillips described what he calls "struggling with scroll" — the competition among brands to stand out with meaningful ads on devices. His assessment of where the industry stands was blunt: marketing teams are strong at making TV ads, but social media creative still needs work.

He traced the problem to process. "A lot of the scroll ads are naturally formulaic because the only way brands can manage the process is by putting guidelines in, and then the guidelines become more important than the creative," Phillips said.

The cost of that formulaic output is not negative reaction but indifference. An indifferent response, Phillips argued, means a campaign had no impact beyond the brand having "spent some money."

His prescription: "Marketers need to think creatively about making a story and making it authentic for the consumer, the creator, and the brand."

Can AI testing replace human panels?

Zappi has developed a synthetic audience — an AI-generated, data-driven simulation of real customers — that brands can use to test large campaigns at scale. The use case targets volume: a campaign with thousands of assets cannot realistically be evaluated through human research alone.

"Testing all of them with humans is too time-consuming, too expensive," Phillips said. "However, you do want feedback on every piece of creative. You want to know whether it's good or bad. You want to know whether you can improve it, but also critically, you want to understand: Does it fit with the rest of the campaign?"

He drew a clear boundary for practitioners: key pieces of a campaign should still be tested with humans, while derivative versions and lower-risk elements can go to synthetic consumers.

The panel's broader signal for 2025 planning is that attention metrics alone won't predict commercial outcomes — and as synthetic testing matures, the constraint on testing every asset before launch is shifting from budget to workflow design.

Source: Adweek Social Media

Sophie Lindqvist

Show full bio

Correspondent covering people, launches and day-to-day practice in Social platform news and policy at Reach Brief.

More from Sophie Lindqvist →

Related articles

« Previous articleNext article »