Death of the persona

Death of the persona: Embracing AI-driven personalization

It’s time to harness real data and make a real impact.

In the evolving landscape of product design, the rise of AI-driven personalization challenges the traditional use of personas. While personas have long served as a valuable tool for understanding user needs and behaviours, user dynamic and ever-changing nature demands a more adaptive and responsive approach. Although personas may still have a place in some situations, integrating personas with accurate user data can create a more dynamic user experience at the core of design systems. Product design has the potential to evolve from static personas to personalized, contextually relevant interactions.

The limitations of personas

Personas are fictional characters representing user demographics, behaviours, and goals, traditionally used in the design thinking process to empathize with users and guide the development of user-centred products. However, as Christin Roman in his article “The problem with personas” notes, personas often suffer from several flaws:

• Superficial Data: Personas sometimes rely on demographics, needing deeper insights into user motivations and behaviours.
• Overemphasis on Details: Designers may focus on unimportant aspects of personas, leading to misaligned design decisions.
• Lack of Real Data: Personas often include assumptions or fictional quotes and may lack grounding in user data.
• Absence of Scenarios: Without scenarios, personas lack context, reducing their effectiveness in guiding design choices. (Also noted by Jared. M Spool)

We need a more nuanced and effective approach to understanding and addressing user needs.

The rise of AI-driven personalization

As mentioned by Carol Flanders, AI-driven personalization leverages real-time data and machine learning algorithms to create adaptive, responsive, and contextually relevant user experiences. Bogdan Anghel has listed three core topics of AI in design that could also be applied to the idea of AI within design systems:

1. Adaptive design patterns

AI can create adaptive design patterns that automatically adjust UI elements. This can enhance accessibility and engagement by tailoring the design to fit the user's needs and requirements. For instance, an AI system might modify navigation, information architecture, or component styles dynamically to suit the user's needs and requirements.

2. Anticipating user needs

By analyzing user data, AI can predict trends and preferences, enabling the creation of intuitive UI elements that align with user behaviours. For example, AI can adjust the layout of an interface to better suit the needs of the users based on their changing environment or activities. This can make it easier for users to find what they need and when they need it.

3. Dynamic environment-based adjustments

The future will be multi-modal, so it is wise to think about these scenarios now — Romina Kavcic

AI can adapt UI elements to different environments and device types, optimizing the user experience based on situational context. A user's needs in bed looking for an optimal hiking route will look significantly different than when they are actually trekking through nature. The AI can tailor components based on the user’s browsing history and current context, displaying relevant information and adjusting the layout for optimal viewing on various devices.

Balancing benefits and risks

As highlighted by Creative 27, several pioneering companies have harnessed AI to offer personalized user experiences. Netflix’s recommendation engine analyzes user interactions to suggest shows and movies, while Spotify employs AI to curate personalized playlists. These examples illustrate the tangible benefits of AI in crafting bespoke user experiences and enhancing engagement.

However, it also raises several important consumer and social welfare considerations. Omid Rafieian and Hema Yoganarasimhan’s research highlights critical implications such as:

• Search cost: AI can significantly reduce search costs by delivering better product recommendations and helping users find what they need more quickly and efficiently.
• Privacy: This personalization relies heavily on large-scale consumer-level data collected through online tracking, raising significant privacy concerns that must be addressed through transparency and robust data protection measures.
• Fairness: If different consumers receive different treatments based on their data profiles, prompting ethical questions about equity.
• Polarization: Personalized content delivery can create echo chambers that reinforce biases and deepen societal divides, highlighting the need for a balanced approach that exposes users to diverse perspectives.

One significant challenge for designers is the presence of preexisting biases within AI models. Designers need to ensure that algorithms are built on diverse and accurately labelled datasets that represent all user segments. Furthermore, continuous learning and human feedback from real users (and customers) are essential for refining AI systems and preserving their effectiveness.

As designers, we also need to consider these implications as part of our design process to enhance user experiences, respect privacy, ensure fairness, and promote a more informed and balanced society.

Embracing the future of design systems

As we progress, integrating AI into design systems will become increasingly critical for companies aiming to stay relevant at scale. Romina Kavcic, founder of The Design System Guide, has written an excellent article highlighting how AI is redefining design systems, which I won’t re-hash here.

AI-driven personalization is not merely a trend but a foundational shift in how businesses engage with their users.

By embracing AI, designers can unlock new levels of engagement and satisfaction, creating interfaces that genuinely understand and cater to individual users. Using Atomic design principles, see the opportunity for intervention below.

We can use the power of consistency and flexibility that tokens/components provide to support these personalization opportunities within the broader patterns and features.

10 questions – we can start asking at a system level

This is by no means an exhaustive list, but just a few I would want to experiment with:

1. What are all the goals and user needs connected to this feature or existing pattern? User interviews, ethnographic studies, or diary studies can be used to identify the connections.
2. Does my user have existing preferences outside of this feature that can inform their needs or behaviour? For example, existing mental models.
3. Will this feature be used across multiple devices and within different situations? For example, while running and sitting at a desk.
4. Will my user's workflow differ based on a set measurement or trigger? For example, at a specific time of the day or previous input.
5. Will my user’s role change while using this feature? For example, from author to reviewer.
6. Will this feature have to be displayed in other areas of my application or device that is/isn’t controlled by me? For example, notifications or placement of the application.
7. Will my user need to switch input methods while using this feature? For example, from text input to taking a picture.
8. Can my user control these changes? Is this customization instead of personalization? For example, give me healthier options instead of following my unhealthy behavioural data.
9. How do I measure if the personalization I have introduced is effective or useful? For example, shorter time in the app or fewer interactions in specific areas.
10. What user data will be required to personalize this component, and how will I ensure its accuracy, relevance, transparency and avoiding bias? For instance, an opt-in approach with clear explanations can help people understand how their data is being used.

What other questions should we ask when designing for personalization?

By leveraging adaptive design patterns, anticipating user needs, and making dynamic environment-based adjustments, AI offers the potential to create truly personalized, contextually relevant user experiences. As businesses embrace this shift, they can unlock unprecedented levels of user engagement and satisfaction, paving the way for a more responsive and user-centric future in product design. Design can contribute to this space if we start asking the right questions at the right time. I am excited, are you?

For further reading on this topic, see the following:

• When large language models meet personalization
• User Perceptions of Algorithmic Decisions in the Personalized AI System
• How Much Personalization Is Enough in UX Design?
• Personalization in product design: Tailoring UX for individual users

Thank you for reading! As this is a continuous learning journey for me, please feel free to reach out with your feedback or any questions you may have 🌸

Death of the persona was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.