Generative AI Brings Marketing Personas to Life

But what if you have little or no data at your disposal?

This challenge occurs even among the biggest brands. Until recently, even a large, international beer brewer hardly had any insight into its own target group. After all, consumers were only reached indirectly – through retail channels, events and sponsors. The marketing team was therefore forced to base the approach of advertising campaigns primarily on its own intuition, in the absence of data on the various target groups.

Machine learning and generative AI proved to be a solution to this challenge. Thanks to sponsoring and organizing events, the beer brand had access to quite a bit of social media data. Info Support developed smart models to convert this data into clusters of consumers with similar profiles. Generative AI tools were then used to bring speaking marketing personas to life; so lifelike, in fact, that the brewer’s marketing team was able to have conversations with these fictional consumers to get to know them better and test campaign ideas.

The hypothesis

The central question at hand: can we improve the effectiveness of advertising by creating personas based on data from real people, gaining a better understanding of the target audience?

The team used data sourced from social media channels. Joop Snijder, who led the project as Head of Research Center AI at Info Support: “The first exercise we did was to form clusters of consumers with similar interests. For that, we first deployed traditional machine learning models, because they are very good at clustering features. Then we deployed large language models to find similar characteristics. Because large language models use natural language processing (NLP), they can make estimates of users’ characteristics based on social media posts.”

In doing so, the team deliberately looked beyond demographic data, such as residence or age: “We especially wanted to look at characteristics that would say something about receptivity to advertising messages. So we started looking at some more complex traits. Think of language use, for example, or a DISC profile that a consumer meets: how does someone score in terms of dominance, influence, steadiness and conscientiousness?”

Emily is a 24-year-old college student from Amsterdam. This festival-goer gets totally absorbed in the moment when surrounded by music, people and good vibes. She is always up to date with festival line-ups and often goes to the club with friends. Emily also has a passion for fashion, she loves discovering fashion items that no one else has. She is a fan of Vinted and loves to score a good deal in pre-loved clothing. She also has an eye for jewelry and loves browsing through the collection of brands like Lucardi.

Emily really came alive thanks to generative tools DALLE-2 and GPT-4. The brewery’s marketing team was able to actually start the conversation with Emily. That helped determine how Emily did or did not want to be approached: she was especially looking for that unique experience, and got excited about personalized, interactive content. What she turned away from? Formal, distant advertising messages.

Higher conversion and customer satisfaction

The approach proved quite successful: the AI-generated personas made it possible to develop very targeted messages, which proved to resonate much better with the target audience, achieve higher conversion rates and higher customer satisfaction.

Joop Snijder: “With this approach, we were able to help the client significantly increase the impact of advertising. The application of machine learning and generative AI contributed to greater insight into the characteristics and preferences of the target group: who are our customers and what makes them happy? This project has paved the way to make many more strides in personalized marketing.”