The practice of creating personas has long been a valuable tool for gaining insights into consumer behavior and preferences. The analysis of a large data stream—coupled with interviews and survey—is an excellent way to establish workable audience profiles.
With the latest advancements in generative artificial intelligence (AI), marketers can now use tools that instantly leverage massive amounts of data. This led us to believe that generative AI could be exceptionally well-suited for audience insights. Read on to learn all about our “experiment” to compare generative AI and human-generated personas—the results may surprise you!
How we used to develop personas
Before delving into the AI-driven persona development process, let’s establish a clear understanding of what personas are. Personas are data-informed personifications representing distinct customer segments/audiences based on research, data analysis, and user insights. They go beyond demographic information and aim to capture the motivations, needs, behaviors, and pain points of specific groups within the target audience.
Traditionally, creating personas involves gathering data through syndicated data tools, custom surveys, qualitative interviews, and other market research. While these methods provide valuable insights, they can be limited by relatively small sample sizes, time constraints, and an associated cost. The degree of rigor is much less for B2B audiences as they are so much more challenging to get both large and in-depth data to support.
In theory, generative AI-powered personas should be as accurate or more accurate than those created by people. With the advent of generative AI, B2B businesses now have access to vast amounts of data from various sources, including social media, website analytics, customer feedback, and more. Given this big data, we are hopeful to uncover new opportunities for persona development.
Our experiment to compare human-crafted and AI-generated personas
Earlier this year, DAC was tasked to develop a specific persona, “decision maker looking for IT solutions”, for both the public and private sectors. Given that syndicated tools are limited in B2B capabilities—and typically require custom market research to build robust data sets for B2B personas—we thought it was a unique opportunity to see if generative AI could produce better personas to unlock valuable, actionable B2B insights.
Can generative AI come to the rescue? Can it be a knight in shining armour? Can it be a lifeboat? We set out to answer these questions by providing ChatGPT Plus with the same type of information we would input into syndicated tools:
Then we asked ChatGPT Plus a series of questions to give a very precise framework to the answers we were seeking:
Our client is an IT consultancy firm that is looking to reach out to you to sell their services. Here is a list of questions that will help us understand you better:
- Why do you go to work every day?
- What motivates you at work?
- What are you looking for in a new IT partner?
- How do you learn about new IT partners online and offline?
- Describe your personality at work.
- Describe your decision-making criteria for a new IT partner.
- In 5 years, what role and where do you want to work?
- Who is your man crush?
- What’s your favourite TV show?
- What’s your favourite film?
- What’s your favourite dessert?
- What’s your favourite news source?
- Do you listen to podcasts? If yes, what are your top 5 favourite podcasts?
Results and learning
ChatGPT Plus was able to build a detailed persona, “Alex”, that included demographic information such as the age of his children, overall media consumption, and even his preferred podcast platform.
But when we asked ChatGPT to make distinctions between the public and private sector versions of Alex, the tool seemed to reach the limits of its current capabilities, resulting in repetitive and circular data.
ChatGPT Plus is exceptionally good at summarising, but this strength becomes its weakness in the context of personas. The tool generalises, mitigating the key differences of each persona and creating stereotypes, which is exactly what personas try to avoid. As a result, we will continue to approach B2B personas with our best-in-class syndicated toolkit and custom market research. This package creates distinctive personas that lay the groundwork for comprehensive, data-informed marketing strategies.
AI is just one piece of the puzzle
Generative AI opens up new possibilities for businesses to gain deeper insights into their target audience, but only if we consider it as one source of information among others, such as syndicated tools, custom market research, social listening, and more. The risk of making it your sole source is that it would generalise and mitigate differences into stereotypes, preventing effective marketing strategies from being developed and implemented.
As generative AI continues to advance, persona development will undoubtedly become an indispensable tool for businesses looking to thrive in the age of data-driven decision making. But human interpretation and intuition remain absolutely critical in understanding the nuances of customer behavior.