Harnessing the Power of AI in Marketing is a Double-Edged Sword! While generative AI tools can drastically increase the speed of marketing processes, the absence of robust safeguards may lead professionals astray.
When it comes to demystifying how generative AI operates, it’s all too easy to default to language that personifies the technology. Terms such as "think" or "design" give the impression that tools like ChatGPT are more than just algorithms responding to user prompts. As a result, many everyday users are increasingly turning to these AI platforms as substitutes for mental health professionals, spiritual guides, or even medical advice.
For marketers, this inclination to interpret AI-generated content as possessing deeper meaning brings both opportunities and pitfalls. Take AI audience personas, for example. These personas leverage large language models (LLMs) like ChatGPT and others to enable brand managers to sift through vast amounts of consumer data, making actionable insights more accessible.
Traditionally, audience personas have been indispensable in marketing strategies. These analytical tools help differentiate consumer demographics and understand intricate linkages between various data points. In fact, prior to 2022, marketing teams invested substantial amounts of time and resources—often weeks—developing these personas to guide their media planning and strategies. However, the landscape has evolved, and now many turn to AI tools for assistance. According to a survey conducted by SAS, an impressive 85% of marketers are now utilizing generative AI in various capacities.
Take the example of Lavazza, an Italian coffee brand that started employing AI personas in its marketing efforts during the second quarter of this year. This strategy allows their team to not only streamline development time but also to engage with the data behind these personas in innovative ways. For instance, they have created a persona named "Adele" for the French market, while targeting British and American audiences with another persona called "Lucy."
Simone Ballarini, who leads business intelligence and consumer insights at Lavazza, shared, "With this interface, we have all the advantages of a human conversation blended with the objectivity of quantitative information." In Lavazza’s approach, these AI personas are crafted using insights from extensive sources, including 5,000 consumer interviews conducted by Kantar, consumer surveys from Nielsen, and additional data supplied by the National Coffee Association. The wealth of information at their disposal is overwhelming, far beyond what any individual could comprehend at a reasonable pace. Moreover, the company can continually update these personas to reflect the latest research, ensuring their relevance.
These personas serve multiple purposes: generating initial feedback for creative concepts and aiding in media strategy. Thor Olof Philogène, CEO and co-founder of Stravito, adds that the intent behind their solution is to expedite these processes rather than completely eliminate the need for traditional testing investments. "Personas enhance early-stage exploration, allowing teams to commit resources only to the most promising ideas. Think of it as an early warning system, not a final validation tool," he explained. Ballarini noted that Lavazza had already utilized these personas for initial evaluations of branding and creative assets within their Dolcevita product line.
Similarly, at Stagwell agency Code & Theory, the implementation of AI personas has expanded to include every client on their roster, which features well-known brands such as Champion, Albertsons, and Marriott Bonvoy. This surge in the adoption of AI-generated personas is not an isolated phenomenon. The ease with which agencies can now create audience personas using LLM technology has led to a substantial increase in their use.
For example, Jellyfish, a digital marketing services company, has integrated an audience persona agent into its Pencil platform. John Dawson, the head of strategy at Jellyfish, elaborates, "You can evolve a persona from a simple slide into a vital component for developing concepts and go-to-market strategies."
However, despite the clear advantages, industry leaders have expressed caution about the potential pitfalls of over-reliance on these AI-generated personalities. Marketers in 2025 are already feeling uneasy about making creative or media decisions without solid backing from various studies and surveys, which often sidelines their own instincts and experiences. Additionally, many decisions are becoming automated, especially with the rise of sophisticated media buying tools such as those offered in Meta's Advantage+ suite.
The most pressing concern is that marketers—like any other users—may start to treat AI personas as infallible authorities. Assigning names, distinct visual features, or even AI-created voices (as Code & Theory has done to impress clients) further enhances their anthropomorphic appeal. Notably, even Piper, who urges caution against viewing personas as human, describes the technology as being similar to a living entity.
As the use of personas becomes more entrenched within the marketing world, dependence on aggregated feedback poses a legitimate risk. "They aren't real human beings, even if they possess faces. They're instruments, not truths," Piper reminds us.
So, how can agencies and their clients ensure that team members do not regard feedback from AI personas as gospel? One strategy suggested by Jellyfish's Dawson is training. Agencies and in-house teams must ensure that everyone involved comprehends the data sources feeding into the AI personas and understands their inherent limitations, as well as the role they play in utilizing these AI-generated profiles. For example, Lavazza’s audience profiles include footnotes that reference the data sources underpinning the generated insights, similar to the footnote format used by Perplexity.
Additionally, employing specific prompting techniques can help maintain the appropriate use of personas. Piper mentions that Code & Theory consistently instructs its persona agents to critically evaluate inputs to counteract some LLMs' tendencies to be overly accommodating. "LLMs are crafted to satisfy," commented George Forge, senior vice president of client technology and product development at Quad.
Marketers can also establish safeguards to prevent uninformed decisions based solely on AI persona recommendations. For instance, Lavazza’s media agency, Wavemaker, is granted access only to the insights generated by these personas, not the personas themselves, thereby insulating part of the decision-making process from excessive AI influence.
Piper further advises that human oversight should be embedded at different stages of the AI persona research process. Insights that seem compelling should undergo scrutiny through surveys and feedback panels, and creative concepts that receive a positive response from personas must still be validated by human focus groups.
The overarching message is that marketers leveraging these personas still rely on their judgment. As Piper aptly concludes, "Without human involvement, you’ll eventually end up with generic and uninspired outcomes."
Join the Conversation!
What are your thoughts on the rise of AI personas in marketing? Do you think they enhance or undermine the creative process? Let us know in the comments!
