Episode 05 features Sabrina Talma, CEO and Co-Founder of Human Made Machine, on what creative testing reveals about brand campaigns, global audiences, and the real limits of AI-generated content.
We have released the fifth episode of Season 2 of Audience Connection, a podcast exploring the science and strategy behind content that truly resonates. Host Lydia Chan sits down with Sabrina Talma, CEO and Co-Founder of Human Made Machine, a creative intelligence platform that helps global brands test and refine creative work before it goes to market. With a client roster that includes Google, the NBA, and The North Face, Talma brings a rare, data-backed perspective on why most brand campaigns miss the mark and what brands can do about it.
A central theme of the conversation is the gap between marketer instinct and audience reality. As host Lydia Chan notes, "There's been many studies done on the ability of marketers and creatives to predict how well a campaign or a piece of content performs. And guess what? It's pretty much been a flip of a coin." Talma echoes this, stating plainly: "Marketers are terrible at marking their own homework." She illustrates this with a real campaign example in which a global self-empowerment ad produced starkly different responses from Gen Z audiences in the United States and China — American respondents viewed it as a mirror, seeking personal relatability, while Chinese respondents approached it more like a window, prioritizing aspiration over identification. The finding highlights a recurring risk in global marketing: assuming that audiences, even within the same generation, interpret creative the same way.
The episode also tackles AI in creative production with a level of scrutiny that goes beyond the current industry conversation. Talma shares results from a head-to-head test between an AI-generated and a human-made ad for Puma, noting that while audiences did not identify the AI spot as machine-made, their feedback was telling: "This looks like every other sports ad." She argues that AI tends to converge on an average rather than differentiate, and that "from a longevity perspective and ideation, you still really need that human touch." On AI personas for creative testing, Talma references both her own experiments — which returned accuracy rates of around 50 percent — and a Stanford and Google DeepMind digital twins study, which found that while AI responses matched human attitudes at 85 percent for general life questions, the match rate dropped significantly when purchase decisions and risk were introduced. Her advice is straightforward: approach AI personas with caution.
The episode closes on a grounding note, with Talma pointing to David Ogilvy's 1983 principles of advertising as proof that the fundamentals of effective creative have not changed as much as the industry assumes. The Audience Connection is presented by Casual and is available now on Spotify, Apple Podcasts, and all major podcast platforms.
.svg)
