CJ Williams ← All Work

Case Study

Purchasing Platform

From icon-only navigation and broken search URLs to a structured, AI-powered procurement experience.

Purchasing Platform
Client
A B2B procurement SaaS serving mid-market and enterprise buyers across North America
Role
Lead UX Designer
Timeline
2024 – Present
Team
Cross-functional team of 6 (Product, Engineering, QA, Marketing)

UX Research · Design Systems · Information Architecture · AI Integration · Usability Testing

The Challenge

The platform had a powerful AI search engine (HawkSearch) under the hood and no experience to match it. Navigation was a row of unlabeled icons. Search generated broken, dead-end URLs. The platform's most powerful feature — AI search across purchase histories, warranties, and supplier catalogs — was invisible to users and unknown to most of the team.

The mandate: make what exists work, introduce proven e-commerce patterns, then design the leap — an AI-native procurement experience.

What I Did

01 Design System

40+ button variants, no shared visual language, and every new feature compounding the inconsistency.

A token-first design system built in Figma — primitives through templates, every component documented with interaction states and responsive specs.

02 Mobile Experience

No responsive foundation — broken layouts, inaccessible touch targets, and purchasing flows that punished mobile users.

A mobile-first rebuild of the full journey — discovery, cart, checkout, support, and account — modeled on proven e-commerce patterns. Task completion rose across every flow.

03 Platform Navigation

“Buying Desk” opened a quote form. “Vendors” led to info pages. Everything else was an unlabeled icon — and the storefront and admin ran on two different design systems.

Labeled, hierarchy-first navigation with breadcrumbs, unified across storefront and admin. One coherent product instead of two disconnected ones.

04 Shopping Experience

Refrigerators filed under “Anchors & Setup.” Broken search URLs. 83% mobile cart abandonment and an 85% exit rate right after purchase.

A rebuilt taxonomy that matches how buyers search, clean faceted-search URLs, product pages that build buying confidence, and a streamlined checkout — with HawkSearch reworked into rapid, navigable results.

05 Platform Redesign

The quick wins proved the platform could work — but they were still patches on an aging foundation.

A ground-up reimagining: full design-system overhaul plus mid-fi and hi-fi concepts for a search-first, AI-native procurement experience. Demonstrated to stakeholders and strongly received; cut before shipping by a market downturn.

How It Was Validated

  • Heuristic evaluation against Nielsen's 10 usability principles
  • Tree testing of the restructured navigation taxonomy
  • Moderated usability sessions on search and registration flows
  • WCAG 2.1 AA accessibility audit before handoff

After Launch

  • Redesign concepts demonstrated to the full stakeholder group
  • HawkSearch AI capabilities documented and socialized internally
  • URL and breadcrumb patterns set as platform standards
  • Continuous Hotjar session-recording feedback loop

The Outcome

54%Improvement in session purchase completion
0→1Navigation architecture built from scratch
40+Component variants consolidated into a single system

The platform's AI was genuinely competitive — what was missing was the UX scaffolding to make it usable. Each workstream added a layer: a design system, mobile foundations, real navigation, and a validated vision for an AI-native future.