Designing a modular home system to enable scalable personalization

 

OVERVIEW

Mercari’s Home experience was entering a period of rapid change. As personalization initiatives accelerated and the design system was undergoing a major transition, the Home surface risked becoming fragmented and difficult to scale. I led the design of a modular Home foundation to enable flexible, server-driven layouts while maintaining consistency, reusability, and long-term maintainability across teams.

 
 

PROBLEM & CONTEXT

At the time, Mercari’s Home experience was composed of many one-off layouts. Components existed, but their usage was inconsistent, and layout logic varied across surfaces. As personalization initiatives accelerated, this fragmentation became increasingly problematic.

In parallel, Mercari’s design system was in the middle of a major transition toward a new version. This meant that design decisions had to account for an evolving set of standards, incomplete specifications, and shifting constraints.

The core challenge was clear: how do you design a Home experience that supports personalization without hardcoding layouts, while also aligning to a design system that was still in flux.

 
 

ROLE & SCOPE

I was the lead designer responsible for the end-to-end design direction of the modular Home system.

My responsibilities included:

  • Defining the component taxonomy and modular rules

  • Establishing layout principles that could scale across content types

  • Partnering closely with engineering on feasibility and constraints

  • Influencing standards beyond a single surface to align with the evolving design system

I worked alongside another designer in the same domain. While I owned the direction and final decisions, her input was valuable in shaping and refining the system.

 
 

KEY DECISIONS & TRADE-OFFS

A central tension in this work was flexibility versus consistency.

Personalization demands flexibility. Design systems demand constraint. I made a deliberate decision to limit layout freedom in favor of predictable composition rules. This reduced visual variety in the short term, but enabled scalability, reuse, and future experimentation without redesigning from scratch.

Another major trade-off involved spacing and layout logic. Early explorations leaned toward visually expressive spacing, but engineering constraints made this difficult to implement consistently across dynamic content. I worked directly with engineers to define spacing rules that were both visually coherent and technically viable. This resulted in a system that was less expressive than the initial explorations, but far more robust.

Finally, designing against an evolving design system required making forward-looking assumptions. In several cases, I designed components based on where the system was heading rather than where it currently stood. This required close alignment with the design systems team and a willingness to accept short-term ambiguity in service of long-term coherence.

 
 

DESIGN STRATEGY & SYSTEM

Rather than designing fixed screens, I approached the Home experience as a composition problem.

I began by auditing existing Home components and patterns, identifying redundancies and inconsistencies.

From there, I defined a modular taxonomy that broke the Home experience into composable units with clear rules around hierarchy, spacing, and behavior.

Each module was designed to:

  • Stand alone as a reusable unit

  • Compose predictably with other modules

  • Support dynamic ordering and personalization logic

  • Conform to evolving design system standards

The outcome was a flexible system where layouts could be assembled without bespoke design work each time. This made it possible for teams to propose new features or content placements with a high degree of completeness, often without requiring a designer to start from zero.

 
 

DELIVERY & HANDOFF

I delivered the final system documentation and designs to the design systems team during their broader overhaul effort.

Although I left the company before full implementation, the work was structured to be durable. The rules, taxonomy, and composition logic were intended to outlive any single surface or feature.

Subsequent public releases around Mercari’s personalization efforts reflected elements of the foundation I had defined, particularly in the underlying structure and composability of the Home experience.

 
 

OUTCOME & LONG-TERM IMPACT

Because I was no longer at the company during implementation, I cannot point to direct metrics. Iimplementation occurred post-handoff, but the design principles continued to influence the Home experience roadmap, so the intended impact of this work was clear.

The modular system was designed to:

  • Enable faster feature proposals with partial completeness

  • Reduce dependency on designers for every layout iteration

  • Improve consistency across personalized Home experiences

  • Provide a stable foundation for experimentation and personalization at scale

In effect, this work shifted design from being a bottleneck to being an enabling layer.

And while this work was absorbed into a broader organizational effort and evolved beyond my direct involvement, members of the engineering and product teams later shared public write-ups that describe the technical and architectural direction of Mercari’s Home and personalization systems. These posts provide external context on how modular UI and Home experience strategies continued to evolve following my tenure.

Related product/engineering blog posts:

Further context (optional): AI personalization project page

 
 

REFLECTION

Looking back, this project reinforced the importance of designing for systems, not just surfaces.

The most meaningful work here was not the visual output, but the decisions around constraint, scalability, and long-term maintainability. Designing within an evolving design system required comfort with ambiguity and a focus on principles over polish.

If I were to do this again, I would push harder for lightweight measurement around internal efficiency and adoption. Even directional data would have helped quantify the impact of the system. That said, the work achieved its primary goal: establishing a durable foundation that could support Mercari’s personalization ambitions beyond my tenure.