The Problem

Cox Automotive's eCRM had a loyalty problem. After years of incremental updates, the platform was showing its age — and dealerships were noticing. Drive Centric, a newer competitor with a modern, polished interface, was actively positioning itself as the upgrade dealers deserved.

The threat was real. Dealers weren't just complaining about aesthetics — they were considering switching. For a platform used across thousands of dealerships nationwide, that was a business crisis, not a design request.

The project escalated quickly, reaching executive leadership including Steve Rowley. A dark mode redesign of the eCRM became the highest priority initiative on the product roadmap.

The Research

Before touching the interface, we needed to validate key layout decisions with real users. I designed and ran a placement survey distributed to 180 dealer respondents — a mix of deal-builders and other daily users — testing two competing concepts for the communication comment section: input box at the top (Concept A) vs. input box at the bottom (Concept B).

The results were consistent across every user segment tested:

• Comment placement survey 1: Concept B — 50 vs. 25

• Comment placement survey 2: Concept B — 32 vs. 19

• Comment placement survey 3: Concept B — 24 vs. 18

Dealers preferred the input box at the bottom by a roughly 2:1 margin. Their reasoning was intuitive — it mirrors the mental model established by everyday messaging apps, keeps deal details front and center, and reduces the friction of scrolling past history to add a new note.

This wasn't just a preference. It was a signal about how dealers actually think when they're in the middle of a deal.

The Design

Blind Competitive Benchmarking Ran two rounds of side-by-side dealer surveys with all branding removed so responses reflected the UI, not brand loyalty. Round one, Drive Centric won. We went back, sharpened the direction, and ran it again. Round two, ours came out on top.

Dark Mode Design System This wasn't a theme switch — it was a ground-up rebuild of color tokens, typography hierarchy, and components. Every decision was made to scale across the full platform, not just the screens in scope.

Research-Informed Layouts Survey data from 180 dealers directly shaped component decisions. The communication comment section — input box placement, thread hierarchy, reading order — was redesigned around how dealers actually think mid-deal, not how the original system was built.

AI-Assisted Critique Loop Integrated Claude AI into the iteration process to pressure-test UI decisions in real time — evaluating contrast, hierarchy, and layout logic on dark backgrounds before stakeholder reviews. It compressed the feedback cycle and kept reviews focused on strategy, not catching inconsistencies.

The Outcome

The redesign received executive sign-off and was positioned for development handoff. I was laid off before the launch, but the project was approved, research-validated, and dev-ready at the time of my departure.

What the work demonstrated was bigger than a visual refresh — it was a response to a competitive threat, backed by user research, executed at design system scale, and validated at the highest levels of the organization.

Reflection

Working under genuine business pressure with a real competitor in the picture sharpened every decision. The AI-assisted critique loop was one of the most effective process changes I made — it's now a permanent part of how I work. If I could revisit anything, I'd have pushed for usability testing on the full dark mode flows earlier in the process, before executive review rather than after.