CASE STUDY

Arcadia: AI-powered insight platform. $25M+ in new business.

01

My Role

Founder and design lead for Arcadia, an AI-powered insights platform I conceived and built from the ground up while Director of Experience Design at Havas.

How did you make AI insights accessible to clients?

Making AI outputs feel human. Raw sentiment scores meant nothing to clients. I needed to show them stories. I designed a visual language that mapped emotions to journey moments, turning “sentiment: -0.73” into “frustration at checkout.” The trick was balancing detail with clarity: complex enough to be credible, simple enough for exec decisions.

How did you convince stakeholders to invest in AI in 2018?

I didn’t pitch AI. I pitched solving a $500K problem. Clients paid for research that nobody used because analysis took weeks. I built a working demo with real client data and showed leadership: 3 weeks → 3 hours. That shifted the conversation from “scary new tech” to “competitive advantage.” The $25M business case came later.

02

Opportunity

Clients had fragmented research data across multiple sources but no way to turn it into actionable customer insights.

How did you validate AI accuracy with real client data?

Parallel testing. Our research teams analyzed datasets manually while Arcadia processed the same data. After 20+ projects, a pattern emerged: Arcadia caught insights researchers missed 30% of the time, not because it was smarter, but because it didn’t get tired analyzing thousands of responses. Human researchers still won on nuance, but AI won on scale.

03

Result

$25M+ in new business
Became a standalone product offering for Havas.

  • Built sales enablement materials and trained teams across Havas on how to use AI for customer insights—this was 2015-2018, when most agencies had never touched AI
  • Conceived the end-to-end product vision and designed the complete AI workflow from research input to insight visualization
  • Created a methodology that replaced subjective UX research with data-driven, AI-powered journey mapping

If you rebuilt Arcadia today, what would change?

I‘d build the “so what?” layer from day one. Arcadia showed teams what customers felt but left them asking “now what?” Today I’d embed prescriptive guidance: suggested design fixes, AB test hypotheses, even draft wireframes based on pain points. The best insights are useless if teams don’t know how to act on them.