Scale or Personalization - Lowest Common Denominators?

Lately, I’ve been exploring a problem that comes with designing and optimizing for scale and growth. A common value in experience design is customization and personalization. At its core, that idea is closely tied to our desire for belonging and self-expression.

At some point, though, as systems and tools grow to ever larger scales, they require greater standardization. As a result we end up designing for an ever lower common denominator.

Often, that means designing for an assumed “average” user. Some human-centered design practitioners rightly reject this notion, pointing out absurdities like the “average” person with 2.3 children—the mathematical average of a user base. Yet we still tend to design for assumed average behavior, or we design outright for desired behaviors. In doing so, we exclude behaviors that live at the margins, or we require conformity to a standard we define—intentionally or not.

One area where this tension is being explored more actively right now is neurodivergence, for example.

More broadly, there may be an inherent tension between scale and personalization. Consider how personal networks with billions of users have led to a templatization of interactions — clickable reactions, character-limited replies, and so on — enabling massive scale through standardization. At the same time, we see rising desire for the handmade, the artisanal, the rare, the local — all descriptors for qualities that sit on the opposite end of that spectrum. They aren’t standardized or optimized for scale. They feel personal and personalized, which imbues them with meaning and value.

I have a hypothesis that one appeal of AI technology for many consumers right now is that it can create a sense of more personalized interaction with technology. A big contributor may be natural language processing and the ability to interact in ways that, to many people, feel more natural than navigating a graphical user interface , which is a level of abstraction that, while subtle and ubiquitous, can still feel like a corset for certain interactions. By no means do I think language-based interaction eliminates the hurdles and barriers inherent in standardization and templatization, but it may reduce how significant those barriers feel.

How we navigate this tension between scale and personalization will be an interesting challenge to watch. Systems inherently require a level of standardization to work well. Train tracks need specific widths for train cars to fit. The internet is made up of countless protocols that allow devices around the world to connect and interact with one another. And now, with the rise of agentic AI, there’s a push to design tools and services for agents, in a way standardizing for this new user type. Ironically, an agent is probably more standardized, and more “average,” than a human user would ever be. Yet the question remains what this means for human users and their unique needs, behaviors, and quirks.

Let’s explore.