Fusion — Commerce Intelligence Platform

The problem

GroupM's enterprise clients were sitting on enormous volumes of streaming retail data — but had no clear way to act on it. Existing reporting tools required analyst intervention for every query, creating bottlenecks and decisions made on stale information.

The challenge was to design a self-serve intelligence platform that let non-technical business users — brand managers, merchandisers, sales leads — access real-time insights without needing an analyst in the room.

My role

I led the full design process for Fusion from early concept through launch — responsible for user research, information architecture, wireframing, prototyping, usability testing, and the design system. I worked directly with engineering, product, and data science teams in an agile environment.

Discovery & research

I conducted interviews with business users across five enterprise clients to understand how they currently accessed and used data. The core insight: users didn't want more data — they wanted fewer decisions to make. They needed the platform to surface the right signal at the right moment.

I also audited the existing internal tooling for accessibility gaps and navigation failure points — mapping where users dropped off, what they misread, and where they lost confidence in the numbers.

Design process

I started with low-fidelity sketches to establish information hierarchy — what the user needs to see first, second, and on demand. From there I built analytical wireframes that mapped data visualization types to specific business questions.

High-fidelity prototypes were built in Figma and tested with real users via Maze. I ran multiple rounds of usability testing, tracking task completion rates and time-on-task to validate improvements between iterations.

Design system

I built and maintained a scalable component library used across the platform. Establishing shared tokens for data visualization — color scales, axis patterns, chart types — reduced design-to-development friction significantly and kept the interface visually consistent as the platform scaled.

Outcome

By combining structured usability testing with adversarial edge-case simulation — deliberately breaking flows to find what users couldn't recover from — recommendations consistently made it into production, not just into decks.