This post will explore some central ideas on bridging the divide between personas and intelligent systems in the context of online advertising. The current state-of-the-art persona systems are interactive, but they are not intelligent. Hence, the goal is to move from interactive persona systems to intelligent persona systems — “intelligent” includes interactive by definition but also several other characteristics, such as the ones we explore in the following. Each characteristic is first defined and then we reflect on its relevance for intelligent persona systems.
- Adaptability – Intelligent systems must be able to adjust and adapt to changing environments and user requirements. Intelligent persona system examples: Automatic updating of personas when customer behavior or attributes change (concept drift). Personalization of ad content for different personas (either generating from scratch or modifying human-created content).
- Learning – Intelligent systems must be able to learn from their environment and past experiences in order to improve their performance. Intelligent persona system examples: Automatically recommend personas to target based on observed performance on goal metrics defined by the advertiser.
- Reasoning – Intelligent systems must be able to make decisions and solve complex problems by using logical reasoning. Intelligent persona system examples: Almost none do – skip. (…or, maybe explain *why* a certain action would have a certain effect; explainable algorithms to increase stakeholder trust and actionability.)
- Perception – Intelligent systems must be able to sense and interpret information from their environment. Intelligent persona system examples: Connect to web and social media APIs to extract real-time information.
- Interaction – Intelligent systems must be able to interact with humans in a meaningful way. Intelligent persona system examples: Recommend specific marketing actions and personas for whom to address those actions – offer an intuitive UI to accept or refuse actions.
- Self-Awareness – Intelligent systems must be aware of their own state, be able to determine the best course of action, and be able to adjust their behavior accordingly. Intelligent persona system examples: Record the performance of different personas and marketers and change recommendations accordingly (reinforcement learning).
From an evolutionary point of view, the persona system evolution can be viewed as follows: automatic persona generation systems => interactive persona systems => intelligent persona systems. The latter part has not yet been achieved in either research or practice. Interactive persona systems are systems that provide corporate users (e.g., advertisers) the ability to generate, filter, and learn more about the personas that represent their users. The currently most advanced systems are interactive but not intelligent. “How can we get to intelligence?” is the key question.
Ideas here are based on academic research: Kaate, Ilkka, Salminen, Joni, Jung, Soon-gyo, Olkkonen, Rami, and Jansen, Bernard J. (2023). How Can Intelligent Persona Features Support Online Advertising Work? In the Proceedings of The Sixteenth International Conference on Advances in Computer-Human Interactions. IARIA, p. 65-67. Venice, Italy. ISBN: 978-1-68558-078-0.
Curious to do your Bachelor’s, Master’s, or Doctoral thesis on personas? Contact me for collaboration opportunities and research topics: Dr. Joni Salminen, email: jonisalm@uwasa.fi.